Can you please do a video about the eagles run game and their O line and Jalen from the past 8-9 weeks. I think it would be a good video to show blocking schemes and different styles of runners.
Just wanted to say, you can be arrogant on here Brett. You singled out people (including videos of them in your videos), criticizing Mayfield for not making it work with OBJ. Now, it's time for you to eat some sh*t, on why you were wrong. You were wrong. Mayfield was the main reason reason it didn't work with OBJ. I love your channel, but your arrogance to criticize others, is why sometimes you look like a complete jacka$$. Hope you improve in the future.
it is unbelievable how good this channel is, Brett Kollmann putting out FREE content that most big name talk shows will never come close to in terms of quality
Because TV sports shows have to have as much content about every subject as possible between commercials. And they’re generally personality driven puff pieces designed to support the sport in all facets.
As an individual who does data analysis for a living and had to do analytics for my last job, and to some extent at my current job, the biggest issue I see is most people don't understand how to view analytics. I'm not just talking about the math but when people say "Analytics says you have to do X" well analytics doesn't tell you to do anything. To keep it simple, analytics gives you probabilities of what is more or less likely to occur in a given situation, it doesn't mean you should always do what the probabilities say you should probably do. We don't ever look at a situation and think "Well, there is a 70% chance of this happening so I have to go for it." We look to see why it is 70%, what were the reasons for the 30% then assess the situation to see are we in a situation where we're closer to the 70 or the 30? Those of us who do this don't always go with what the probabilities say we should, we evaluate and will strategically go against the odds if we think the situation calls for it. The process for why something does or does not occur is vital and it's something every person who does data analysis focuses on, we don't simply focus on results, we focus on the process that leads to those results. Too many people think it's math nerds producing numbers, no, it's math nerds who look at the data, evaluate the data, evaluate the processes that led to the data, and use that to try and inform us in our decision making when looking at the future while being very aware of outside variables that can effect those numbers. That's why when something doesn't work people in data analysis aren't surprised, we know even if the data says probabilities are high for converting on 4th down and 2 we know it's not going to work every single time, in fact the data says it won't work 100% of the time so we expect failures and we expect quite a few of them. It doesn't mean analytics didn't work, in fact it shows analytics does work because if teams are going for it on 4th and 2 and converting 100% we would go back and try to figure out how we got it so wrong since the data says 100% shouldn't be occurring. Analytics are good, they've been used forever in some capacity, it's just now we have more data, more ways to gather and produce data to help evaluate what we're seeing.
What an awesome comment! I play poker for a part of my income and all 70 30's are indeed not equal lol. All coin flips are not either its really strange and hard to explain exactly why but I've played since i was 16 and am closer to 40. I can say without a doubt you are correct and only using the math as a reason to do something isn't all there is to a situation.
You’re correct, but it’s not everyone in analytics that thinks that way. Some people for PFF very much just say you have to do what the higher % play is in a vacuum, when the game is v much not played in one
@@ho31488 example i have just a open ended straight draw. Guy goes all in for 500, i have 500 my ev is obviously negative... only 8 cards help me with 2 cards to come. 32% about. However girl to my right makes horrible call with over pair not putting him on bottom set. Now my ev is positive and even though I'm going to still lose 68% of the time my EV is now positive!
@@SidelineSteve absolutely. Some analytics supporters, of which I'm obviously one, often make the same mistakes as detractors as repeating numbers but not delving into context. Brett did a very good job here going over the variables involved.
I share the same sentiment. Been mostly dealing with the increase of information for the NBA for a decade now and the one thing I had to contend with are folks who odd enough want to help regurgitate a specific narrative and using numbers to correlate situations to support it. Then try apply the narrative through a broad brush stroke when they ignore the other important variables such as possible outliers or patterns that better support another perspective. Those folks I try stop having a discourse with because their aim IS to stop a discourse by jumping to conclusions and dying on a hill with how concrete the data is. After all we got more numbers from situations we aren't used to seeing so its best to give a refinement to the process for interpreting these situations.
As someone who does analytics and statistical modeling for a career, I thought it was worth mentioning the decision right before halftime that you talked about. First things first: this is a great video and you did an excellent job walking through how analytics work and I think this video should be recommended to anyone who "doesn't understand analytics football". Regarding the 4th down play, you can treat this specific example as an expected value problem. Put simply, all you are doing is comparing the EV of option 1 to the EV of option 2. Let's call option 1 kicking a field goal, and to get the EV you simply multiply the odds of converting with the value received. For the sake of simplicity, let's assume the odds of hitting a chip shot there are 100%. So the EV of option 1 (value)*(probability) = 3*1 = 3. For simplicity again, let's assume a TD is worth 7 even though it's not due to missed PATs, but it's easier to evaluate and the logic is the same. Put simply, you want the EV of going for it to be > 3 to make it worth going for it. To solve for that, simply plug it in to (value)*(probability)>3. 7*probability of converting > 3. Solving gets us Probability of scoring > 3/7 or about 0.43. This means if the Chargers have a chance of scoring that's higher than 43% then going for it is a higher expected value than kicking a FG. This is the statistic that was likely thrown out and if the Chargers are converting at that level or higher then going for it was the correct call. You have a point that you can also consider other factors like if taking a guarantee negates the slightly higher EV of going for it, but from a purely numerical standpoint that is what the numbers say. And this brings us to a major point of analytics is that the results are always subject to interpretation or that there could be more to consider beyond just a simple EV comparison. A simple consideration is what if the odds of winning increase substantially if they had taken those 3 points but they don't increase by much more if they score a TD vs getting a FG. This is the challenge of analytics and as you pointed out it takes a lot to know when to strictly follow the numbers and also to identify the difference between bad analysis vs bad execution. Great video, really loved your breakdown. For someone who claims to "suck at math", I think you killed it.
@Vagonius Thicket-Suede I think that's a very valid point and exactly one of those other factors I was talking about. If a FG increases your chance to win from say 52% to 55% but a TD goes from 52% to 65% then the analytics as well as outcome if you convert both say to go for it. And that's really all that goes into analytics in sports - expected value, risk assessment, probability, and regression analysis. There's still more to any game than these factors, but analysis in any sport is like knowing how to play blackjack optimally when going to a casino. It's not a guarantee it will work, but it is the best chance you can give yourself when it comes to decision making.
I was gonna basically make the same comment. While the guarentee of 3 points is nice, you would have to need some pretty extreme circumstances to give up 0.5-1 expected point in order to make going for it wrong. At 6 or 7 points down, I think that going for a FG and making it a two score game is more valuable, but even then, it's the Chiefs. Teams need to be much more aggressive against the Chiefs
6:26 It's actually about somewhere between 4% and 5%. The odds of the chargers not succeeding is about 40 percent if the odds of them succeeding is about 60 percent. The odds of a failure(40%)* the odds of a cheifs TD from the 5(about 10%) = about 4%
Also unless I'm thinking incorrectly here, the idea of the 4th and goal at the end of the half is still "correct" from a probability perspective. Unless the Chargers red zone success rate is
@@johnterrell2825 it's correct, assuming 3 points are "worth" 3/7s of 7 points. It depends on the score differential and the difference in talent between the teams
In the words of Jon Finkel, a professional Poker and Magic: The Gathering player, when talking about the similarities between Poker and Magic: The Gathering: "I think the biggest thing is the deep seeded emotional understanding that the right play is the right play regardless of outcomes. The ability to make a decision 5 straight times, lose 5 times because of it, and still make it the 6th time if it's the right play." If you know and trust the math, then you've got to be willing to keep making that play, even if the last time you did it, it didn't work. Of course, Football's a fair bit different from Poker and Magic, since the odds are ironclad in the latter two games and are more squishy in Football, but the point remains.
I will live and die with Harbaugh. I love him. He will end up in Canton one day. Believe that. Just because they didn’t work well recently doesn’t mean it doesn’t help the team win. Love your stuff Brett
@DeSean Jackson #Jaccpot #0ne0fone #1hunna Congrats! Managed to watch the video and came away with exactly nothing but the most basic observation possible. It's amazing how stupid people can be.
Honestly I’m so happy to hear that chargers player is doing ok. I’m a chiefs fan and that broke my heart on the chargers first drive when they went for it and he was unfortunately hurt
One of the best episodes ever! Amazing content and the video quality with all the visuals has gotten insane! Love to see the channel grow so much - great job Brett!
@@BrettKollmann Just wanted to say, you can be arrogant on here Brett. You singled out people (including videos of them in your videos), criticizing Mayfield for not making it work with OBJ. Now, it's time for you to eat some sh*t, on why you were wrong. You were wrong. Mayfield was the main reason reason it didn't work with OBJ. I love your channel, but your arrogance to criticize others, is why sometimes you look like a complete jacka$$. Hope you improve in the future.
First off, fantastic video. The explanation and analysis was great, as always. Second, I wanted to mention a statistics concept called expected value, which could possibly better explain the early game calls (when time/score scenarios matter less, and you're just trying to maximize points). Basically, expected value is the average of all outcomes, weighted by how likely each outcome is. Taking the end of half play, say your probably of making the field goal is 100%, and the probably of getting the TD is 60%, roughly the 4th down conversion rate. Then the expected value of the FG is 3, and the expected value of the TD is 0.6*6+0.4*0=3.6, since there's a 60% chance of getting the TD and scoring 6, and a 40% chance of scoring 0 (accounting for PAT or 2PT is more complicated but would only increase the expected value, since we all outcomes are >=0). These results would indicate that on average, you'll score slightly more by trying for the TD than kicking the FG. Add on that you're playing a high-powered offence who you're expecting to score a lot, and you might be more inclined to take the high-risk high-reward option than if you were playing a more anemic offence. As for whether or not this aspect of analytics is good/bad, I don't mind it. If anything, it generally makes coaches more willing to go for the exciting, high risk calls, than the safe (sometimes boring) old-school ones. It certainly opens then up to more criticism when thing go wrong, but also makes for more "oh my gosh I can't believe they did it" moments too. Anything that makes the game more exciting (while still being reasonably sane) is good with me.
Love the video as always Brett, especially the points about execution vs. decision-making! Couple questions on the Chargers points. 1. When you say LAC has a 61.9% 4D conversion rate, does that factor in that they're more likely to convert 1 yard than 5? Feel like the math has to differ based on the distance, right? 2. And if not, why is the decision to go for it at the end of the half the wrong one? A 61.9% chance to get 7 points is an expected 4.33 points, which is higher than the 3 points you'd get for the FG (assuming it's made).
Your correct is an average of all there 4th down attempts, it's gonna change on distance. And yes your correct again the math at the end of the half is right too 60% at 6 points is better than 90% at 3. I would have to do a further deeper dive into the math of getting all 7 points though. However i agree that basically 100% 3 points to end a half is better than 40% of getting 0... so the math doesn't always give yourself the true best way to win. Football and math over lap but don't perfectly sink. This is why for so long football was conservative take the points rather than go for it. Breaking that mold is good for the game but never taking free money isn't the best call just cause math says it could be. I used the momentum argument why a goal line stand hurts your team even worse then whatever the math says because it doesn't factor that in, it can't!
@@Melcavic42 Additionally, lots of 4th and goal calculations bring into account where the opposing team will get the ball if the offense fails and this, how likely they will be to score on that new possession (i.e. turning the ball over on 4th and goal near the opposing goal line is much less punishing because the opposing team then had to drive the length of the field). The decision to go for it before the half was the only one I really disagreed with because if you fail, you don't get the opportunity to take advantage of the opposing offense backed up deep and needing to drive the whole field (although I thought the Chargers should have gone for 2 to potentially go up 9 in the 4th). Hope what I'm broadly gesturing at makes sense.
1. That's right he was oversimplifying, but the models can account for the down, distance , field position, etc. 4th and goal at the 5 is around 35% conversion rate.
Unfortunately I do not have data splits by certain 4th down distances so it's tough for me to stay what their differences are in 4th and 1 vs 4th and 5. The teams themselves have that data for their own models tho.
Man I feel like I need Scott Steiner to translate these percentages for me. Great vid with plenty of information to digest but I think the final point made is what it's all about. Analytics have been around in sports under god knows how many names for as long as sports have existed so it's odd to think teams can win without them or that they're new. Having MORE info than ever before doesn't mean teams NEVER had information to base decisions off of. Conversely, teams/fans who only live by analytics will almost always miss out because they fail to value things that spreadsheet data can't quantify IE: who's on the field/what the conditions are like etc. So I guess the best answer on "Are analytics ruining football?" is Reverend Lovejoy's "No with an if, yes with a but!" No if you use them properly, yes if you're solely reliant on them for any decision you make. The idea teams can win without some combination of probability driven data and things like an understanding of talent, chemistry, momentum and circumstances feels like folks picking a side when there's really no reason to.
@@freddiesimmons1394 Funnily enough there was an ESPN article recently about momentum and whether it's real or not but I won't be disingenuous and act I've like read it yet.
@@fackes43 I just read it. The most important quote: "Indeed there is ample data suggesting momentum isn't predictive of some new direction for the game, but rather a generic description of events that have already occurred. ...Instead, we fund their actual win percentage is just a tick less than the expected outcome, which... might signify that the team that was losing before the big play was likely the inferior team all along"
This was an amazing look into the way analytics play into play calling. I feel that oftentimes statistics based decisions are made by a coach just staring at a spreadsheet of statistics and just screaming go go go into his headset. This video was amazing because it showed how coaches use those decisions to inform their gut.
Yeah I think it also depends if the coach has a specific plan to execute against something to exploit with the opponents' defense. Some coaches are conservative about it but they usually think of something pretty crafty to beat a defense to nearly favor that high leverage situation for them. In Staley's case since he's racked up so many 4th and 1 conversion attempts in different key situations (red zone, 20-30 yrd mark) its easier to pick holes more on his play calling side and the offense not executing rather than the decision itself to actually play that down instead of punting.
Analytics are ruining football media coverage - too much talk about percentages instead of the drama of the actual on-field deeds and human emotions that make sport meaningful. No-one feels inspired by the story of who theoretically would win 51% of the time. (And in fact, each moment happens only once and is entirely under the players' control. They're actions, not randomly occurring data.)
this is one of the best vidoes this channel has produced. Fantastic job. Clear, organized, kept it simple for the viewers while also including all the relevant info to make the points. Great job man
I love you brought up execution. Many fans forget the players play, coaches coach. A bad play is not always on the coach, it can be execution in a lot of cases.
This reminds me of that one scene from Blue Mountain State where they bring in a new offensive coordinator and he brings up analytics by saying the enemy team gets its pressures from the left side and the head replies: No shit, their best player is the defensive end on that side. Analytics needs context to be truly useful. Really enjoyed this video Brett!
Analytics isn't just math. It's a study of human performance and behavior. It's often used to develop a better and deeper understanding of what is going on with room for deviation (which is known as mathematical odds. One of the big things I see with going for it on 4th down is that coaches usually implement a different play call than they normally have been doing. For instance, a team could be averaging 5 yards per play during the game and primarily using 11 personnel. But when they get into 4th and 1 from say their own 38 yard line, it's an Ace Jumbo package where the opponent knows they are going to run the ball about 95% of the time. The average yards per play from that formation and that play then dips to say 1.5 yards per play. Then they don't convert the 4th down and everybody blames the decision to go for it when the real problem was the actual playcall. Or like on the Chargers 4th and 2, Herbert is looking to basically make a 1-read, quick throw and likely hurt the odds of converting compared to a play with slightly longer routes that allowed Herbert to make more than 1-read and if that's not open, take the ball and run for the first down himself. It's basically a crux of it not being quite as easy as analytics people (like myself) tend to think it is, but it's nowhere near as difficult as the anti-analytics crowd makes it out to be either. Some of the things that Staley is going thru are things that Kevin Kelley wen tthru and then he started to learn the pitfalls to them and he adapted to help his teams use analytics even more to their advantage.
Yup well said. I always love the decision to go for it, but with the chargers the playcall always leaves me dumbfounded... (just look at thier last two weeks of this season)
Good job explaining some analytics! Success rates, situational football, possessions per game. A lot of what I use to explain up why it’s beneficial going for the fourth and shorts (or short enoughs haha). Having an aggression index to sort of tone up or tone down when you’ll go for it given game flow. It takes a good coach to understand how to bend it as in my defense is dominating them today, this WR is a matchup nightmare or maybe giving your own defense more of a break if you can maintain possession, or make theirs stay on the field. In my mind I like being aggressive with a: dominant or good Oline (can win on short yardage, but also protect well) a great to elite QB (give more chances for a great QB to shred), good OC/playcaller (give playcaller a chance to shred), good matchup (maybe injury thinned a position group or just having your guy beating theirs all day).
The production in this video is incredible. The content was very interesting & visually it was very enjoyable as well. In particular I enjoyed the box that popped up when you were demonstrating the difference between Cook & Bolton on the 4th 4th down attempt. Extremely well done.
@@BrettKollmann Just wanted to say, you can be arrogant on here Brett. You singled out people (including videos of them in your videos), criticizing Mayfield for not making it work with OBJ. Now, it's time for you to eat some sh*t, on why you were wrong. You were wrong. Mayfield was the main reason reason it didn't work with OBJ. I love your channel, but your arrogance to criticize others, is why sometimes you look like a complete jacka$$. Hope you improve in the future.
Thank you Brett. I had to explain to my 70-year-old father who thankfully values my opinion all of this stuff in earnest, and he began to understand it, but not everyone has someone like me to explain the math and strategy behind it to them. I am sending this video to him to drive the point home, even though he already agrees with me, just so he can visually see the math, and how it was the play calls and/or execution that screwed these teams up.
another worthwhile note on the BAL/GB decision: by the end of that game the ravens had 10 safties/cornerbacks on the sideline... Rodgers would've won in OT without much trouble.
One of the best parts of Bretts channel is that he asks a question in the thumbnail then answers it within the first 2 mins of the video unlike most of these channels
Another point about the Chargers going for it to end the half. If they kick and make it it gives them a 7 point lead at half and also allows them to kick later in the game.
I don’t think they are. The problem is a 17 game season is too small of a sample size. If every team followed analytics it would become obvious pretty quick that it’s the right thing to do.
Once again Brett giving us real in-depth analysis, not just some a talking head with a loud opinion. The quality of your content and the way you deliver it will always keep me coming back to watch more
Please look into the percentages rate of winning a game when down by 7 points with about 4 minutes left in the game and your team is on defense. In sum, with onside kicks being a complete fail, once your down by 9 plus points with 2-4 minutes left, the game is mathematically over. That’s why you see so many bs calls when a team is down by 7 points in the middle of the 4th quarter. They know if a large portion of games were over by the middle of the 4th quarter, fans would want to reverse the onside kick safety rules or straight up remove it.
Not that your video editing and animations were bad before, not at all, but with this editor... Everything just looks so smooth and professional, I love it
This is one of your best videos ever. Love hearing the explaination behind these decisions. Analytics is a tool at a coach's disposal in the same way that film study is. I think a lot of people believe that coaches live and die by it, but good coaches simply see it as insight into a situation to help guide their decision making.
Feel like there's a few a coaches who live and die by it to their own detriment. Some guys make a dumb decision and then say but analytics say it was the right call. Analytics don't take into count momentum, and game flow etc. Modern NFL feels like madden where the annoying guy you're playing goes for it every time regardless of the situation any time they get past their own 35.
That's obviously true for any sport. Analytics are supposed to help guide teams' and coaches' decisions to put them in better situations to succeed over the course of a season. Just because analytics say it's more beneficial to throw the ball on average than run it doesn't mean that you should never run the ball ever because that ignores situational context and the reality of the game. But if it pushes a team to be a little more willing to pass over the course of a season, it probably makes them more successful in the aggregate.
Execution can be improved on and in football is better if you can fool the defense at the same time. It just means further refining how to play that part of situational football. Belichick coaches his guys specifically for how to play in rainy/snow weather so at the very least he can say that his squad is prepared to execute better than his opponents in that kind of game. There are also specific special teams stuff he coaches so if there is an off situation comes up for that, his squad will be better prepared to make a play for it than his opponent at least.
as a stats nerd and someone with plans to become a sports analytics guy, I enjoyed this video so much. Please incorporate it into other videos. I could listen to it all day. Great job as per usual Brett
I think it’s important to understand that just like anything else, analytics are not always 100% successful. That’s not the point. Even if you could know every variable and intangible and they all said to go for it on 4th and 2, you can still fail. Using analytics will likely lose you some games that you should have won and win some games you should have lost. The point is that it’s designed to work over a large sample size, so theoretically it’s going to win you more games than it loses you in the long run. It’s also important to keep in mind that just because a play didn’t work, doesn’t mean it’s was a bad decision. That happens, that’s football, and that’s life. I’m not saying going for it on 4th down every time were good decisions, but it’s not as simple as good result = good play and bad result = bad play.
Brett outlined this. Some of those 4th downs were good attempts but HORRIBLE play calling and execution. Staley may exactly be doing a lot of these to gather data in terms of refining the situation so he knows he is going fail on many of these. In the context against the Chiefs however they are good calculated risks overall because you don't want to be trailing against them especially in the 2nd half.
Thank you so much for all this great content on football. It's a sport I love, but an o so complicated one to fully grasp, and you make it really easy! Always entertaining and informative times whenever your videos come out! Merry Christmas to you mister Kollmann!
Another important point in the Ravens game: Harbaugh should have gone for 2 the second-to-last TD. The math is hard to describe in a YT comment but I try: 50% to convert the 2ptc -> trailing by 6 = TD+XP wins the game in regulation 50% 50% to fail the 2ptc -> trailing by 8 -> 50% to convert the second 2ptc -> OT -> 50% to win in OT (probably less in this game) = 12.5% to win in OT 50% to fail the 2ptc -> trailing by 8 -> 50% to convert the second 2ptc -> OT -> 50% to lose in OT (probably less in this game) = 12.5% to lose in OT 50% to fail the 2ptc -> trailing by 8 -> 50% to fail the second 2ptc = 25% to lose in regulation In summary: 62.5% to win 37.5% to lose
Score, Time remaining, yards to go, FG length, Team's ability needs to be crunched into the simulator. I do risk analysis for a living, you are definitely describing the process correct. The goal is to win, not maximize points. Going with FG, to put you up 10 vs going for it late in game is easy to see FG is better decision. The decision tool needs to figure how much time on clock makes going for it on 4th down the better decision.
Very nice video as always, Brett. However, I don't totally agree with your argument for the 2-pt conversion try in the Ravens-Packers game. You said it yourself: the chance of Rodgers getting to FG range are pretty high (~.48 I think you mentioned). So if the Ravens were playing not to lose first, their chance of not losing when kicking the XP based on DSR(-> getting to OT) would be P(tucker hitting XP, so probably .99) x (1 - .48). So pretty much .50. The chance of not losing when going for 2 would hence be P(getting 2pt) x (1 - .48), which would be way lower than .5. So with that being said, why not try to get to OT, maybe win the cointoss and get a nice drive for your offense. I know, the chances of Rodgers beating you in OT would also be substantial, but I just can't see how chosing the quick death is more desirable. With that being said, thank you so much for the great content you put out for us. Happy holidays to you and your family!
Analytics in football is like counting cards in blackjack. If you follow basic strategy and listen to the math, you're guaranteed to win over the long term. It may not work out every single time, but it will absolutely work out over the long term
@@emmettberryhill9449 Yes, but the basic strategy is the same. Follow the statistics and follow the math. Those both exist for a reason. The Chargers objectively made the right decision every time, except for maybe the one right before halftime
Hey numb-skull.... you can play like 500 hands of blackjack in a night... You play SEVENTEEN football games in a year. Variance is and SHOULD be a factor.
@@d3eztrickz Okay? And I addressed that when I said it's not going to work out every time. There are card counters Who have lost $20,000 or more a night for seven straight nights. The point is that over the long term, the strategy is going to work
Even without the analytics (which I think is good for the most part as I think it encourages more 'aggressive' decision making) if you're the Ravens you can look at it like this: You tie the game with a PAT, but now in order to win the game you must: 1. Stop Rodgers, with 40 some seconds left and who only needs FG (as a Niner fan, I've seen this a couple times...) 2. Go into Overtime and hope that you win the coin toss. 3a. Stop Rodgers from getting a TD if you lose the toss and score a TD yourself 3b. Score a TD yourself if you win the toss or hold Rodgers to no points if you only manage a FG *Keep in mind, we're assuming that your OT TD drive is going to have to be somewhere between 70-80 yards. Alternatively you can go for 2, which then you must: 1. Convert a 2pt conversion from the 2 yard line. 2. Stop Rodgers from getting into FG or hope the FG misses Yeah which one sounds easier? Take away the mathematics of DSR, and considering the context of who you are playing against... it still makes sense to go for 2. I think you summed it up best Brett: "You either die quickly or die slowly"
its one of those situation where you go for 2 and the problem wasn't the decision. it was a horrible playcall and was totally expecting some sort of spread empty designed qb keeper or draw play
Going for 2 against the Packers makes total sense for the reasons you mentioned. That same decision against the Steelers was not the right call imo since the Ravens are much better than the Steelers, and even with Humphrey out, the Ravens probably would've won in OT whether they got the ball first or not. That Steelers offense is hot garbage 😂
@@freddiesimmons1394 I'm not saying you disregard the analytics, I'm saying that even if you were to, it would still make sense to go for 2. The eye-test can be deceiving and the statistics can be misleading. Usually it requires a combination of both to evaluate how good a team is.
Yeah. I’m So glad I stumbled on Brett one day. Will never look back for finally finding a smart and high quality football analysis channel. Not enough of them around and none even near Brett. Keep it up dude
Regarding the 4th down at the end of the half: taking the "higher-odds of points" isn't always the best option. Field goals are pretty automatic at that range, so you have an "expected points" of 3. The chargers would then look at their probability of getting a touchdown. They multiply that probability by 7 points to calculate their "expected points" from going for it. If (theoretically) their probability of scoring a TD is 50% from that position, then their expected points is 3.5 (7x0.5). Thus they should go for it rather than kick the FG. As James Marshall mentions, you also just have to consider the context: are there advantages going into halftime with a guaranteed 3 points, or possibly having the momentum of a TD?
Great video as always Brett. One thing you got me thinking about is the notion of DSR and how it translates to points and field position. Obviously this was only "napkin math" as you described, but I was curious about how big plays or chunk plays factor in. Teams rarely march down the field picking up first downs 10-15 yards at a time all the way and many drives are successful based on if a team can rip off a long play and eliminate some of the opportunity for failure over the course of a longer drive with more plays. Anyway, I'm no math whizz, but something that caught my attention. Thank you!
Analytics are a great rule of thumb. The problem can be whether or not your model makes assumptions correctly. Even if those assumptions are generally correct they might not be in a specific situation (i.e. a team has had major injuries in a game, the weather is a factor, etc).
i come here because you are a straight guru. not for 1 particular type of video or breakdown. but because you make me(a couch coach) feel like i kno what im talking about. thanks Brett for all of your content, video, breakdowns and everything else.
Happy holidays to you too Brett. Thanks for all the videos you've put out over the years. This was definitely a present of sorts getting a video of my Chargers right before the holidays so thank you for everything you do Brett 🙏
I just don't know where they get the numbers from. I watch just about every game to some extent on Sunday and I've almost never seen going for it on 4th down every time work. Or going for 2 down by one with over 30 seconds left on the clock
The drive success rate statistic Brett is using in the video comes from Football Outsiders (or at least, that's where I found it). For other statistics, people in the games just chart every play. So, for example, unless you watch every Chargers game and do the same for their 4th down conversions, it's hard to see just how often they succeed on these 4th downs. And again, it is hard to do which is why the NFL or analytics websites pay people to chart all the plays
Analytics work in baseball because it's an 162 game season, over that large a sample size any advantage gained pays off in the long run. In the playoffs though, which more resemble the brevity of a NFL season, it has mixed success (Ex Blake Snell 2020 World Series). The NFL regular season is almost 10x shorter meaning any gain from analytics is miniscule. There's definitely a part for them in the game, but nothing replaces experience and the eye test when it comes to a do or die play. I'd also like to point out analytics has ruined baseball whereas in the NFL it leads to an exciting 4th down play, so I wouldn't worry about it too much as a fan.
Awesome point and why i think football stayed very conservative for decades. I totally agree that even in black Jack you have to know when to just stay in the game and hope rather then try for a higher point total in such a short period of time. Great comment
@@calincampbell5123 Well, where do I start lol. This will be long. It's led to the decline in the starting pitcher and reliance on bullpens. Starting pitchers are stars whereas bullpen arms are mostly nobodies. It would be like pulling Brady or Rodgers in the 3rd quarter because stats tell you their backup does better then since the defense hasn't seen them yet. There's an emphasis on home runs, which are exciting plays but only happen a few times a game on average, and in the meantime has led to more strikeouts and less base hits, meaning less action. On the flipside pitchers are told to strikeout people, which has led to again less action, and burns out pitchers. Defensive shifts have always been a part of the game but are everywhere now, meaning less base hits, and batters trying for home runs to hit over it, which problems I said above. On Base Percentage is the new "Golden Stat" instead of average meaning walks are just as good as hits, leading to longer counts, longer games, and less action because walks are boring. On the financial side, teams realized it's way cheaper to pay young players entry level salaries instead of 30+ year old free agents, leading to tanking for the draft. Those are the main ones, there's many others.
@@calincampbell5123 The game is boring. Home run, Walk, Strike out. They have reduced the importance of defense, base running/stealing, etc. Theo Epstein worked on a Poll with MLB where they found out that triples, stolen bases, and defense were actually in the top 5 “most exciting aspect of baseball” to casual fans.
The oversaturation of football analysis is ridiculous, but I still LOVE watching these videos because they are always levelheaded and bring something new to the conversation. Keep up the great work Brett. You're awesome. Merry Christmas. Love live Mac Jones ;-)
Chiefs vs Chargers was my fav game of the year. They played like how i do in Madden lol. playing 4 downs makes defending so much harder and the game is more exciting tbh
Coming from someone who works in an analytical field, I thought this was pretty well done. One thing I would have liked to see you do is use expected values, for instance a 25% chance at a FG and a 25% chance at a TD are not the same thing. This would have been helpful when deciding if the math played out on the Chargers 4th & goal before half. Let's assume from that distance the FG make rate is 100%, so by kicking a FG the expected value is 3. Now, let's assume the odds of the Chargers scoring a TD was 47% (roughly the odds of converting from there), that would give the Chargers an expected value of 2.82 + another 0.93 for the XP (93% chance of making the XP). That gives the Chargers an expected value of 3.75 by going for the TD. So, yes the math did say to go for it there as opposed to kicking the FG.
Although I do understand what you're saying particularly with the packers game it seemed to that the ravens were heating up offensively and defensively (hell even got a 3 and out the drive before) i think the ravens had a decent enough shot at keeping rodgers out of fg range and extending the game in ot. Plus given rodgers and the packers luck in ot its highly likely that ravens atleast score 3. Call me old fashioned but its better to fight for another drive then not to fight at all
Yes I've also heard at home you tie the game, away you go for 2 because already the computer's have figured out your basically already up 3-0 at home! Even the bet line's figured that out years before the tech did! Anyway in this instance i agree you tie the game at home!
If you want a good subject for a follow up on this analytics topic, peep the Bills vs Pats game. McDermott def paid attention to details as he executed on ideas he would need to get the one up on Darth Hoodie. Went very aggressive on early downs so he didn't see many 3rd and longs that would've been difficult to convert. And when he didn't, he KNEW from then that they were going for it in 4th (and consequently short) down situations. To the point where he did NOT PUNT against Belichick at all the whole game. He also didn't try 'force' the run and designed the offense and play calling to the limits of Josh Allen's full utilization. He was the pass game AND the potent run game. I think this game is where analytics and game situation met with the execution from the right personnel to get the winning results! Especially against an opponent who was playing pretty conservative despite his opposing coach (McDermott) being very aggressive the entire game.
Danny Kleinman, one of the first analyst of Backgammon, compared Football with Backgammon. He even wrote a Book with Daniel Ortega about a crucial match standing, four away vs. two away. It's like a 4th and go! decision in a late football/loopsided match. A play call is here a doubling decision, whereas his execution a decision is about moving the checkers. At 4a vs. 2a, the trailing player can take a double, doubling back in the next move, thus doubling out the gammons of the leading player. The doubling and take points are shifting, depending also on the GWC%. So are late game decisions in Football. And don't forget: Frank Frigo, the CEO of edjsports, won once apon a time the Backgammon World Championchip in Monte Carlo. Their engine is like 'extreme gammon' for Backgammon. Would you backup Frank or the computer?
As a Ravens fan with an MS in Statistics, I can say that John Harbaugh sometimes uses analytics as a crutch when they should just be a tool. I agree with the 2-pointer against the Steelers, because we were on the road, the Steelers had come back from a deficit, and game flow was not in our favor. I don't agree with the 2-point call against the Packers, because we were the ones who had come back, we were at home, and our defense was actually playing well. Analytics are great, but football isn't played on a spreadsheet. The best users of analytics will balance it with a strong coach's intuition.
I couldn't agree more. I have no issue with being aggressive but you should never chase points in the 3 quarters of a game. Especially when you have the greatest kicker of all time. Tucker is 98.7% successful on kicks inside 40 yards including extra points.
I personally agree with both 2 pt attempts by harbaugh purely because of the awful overtime rules in the NFL. I just think it’s disrespectful to the effort your players put in for 4 full quarters to tell them “hey we’re gonna flip a coin to find out if we have a chance to win”. I think that if you have confidence in your offense to score in OT you should have confidence in them to convert a 2pt
@@samuelault4723 I don't mind the overtime rules. The defense gets paid too, and touchdowns aren't just guaranteed. The Ravens were in a spot to force overtime because they forced 2 punts late in the fourth quarter.
The sad thing is with Harbaugh's decision is that we'd have to stop Rodgers either way to win. It essentially becomes do we trust ourselves to punch it in from 2 yards out on one play or not...and while Harbs may have made a good moral decision in trusting his guys, I can't say I imagine we score in most of those situations
People gotta give Staley a break bc LAC was having massive kicker problems for nearly the entire season At the same time, statistics are not pre- determined outcomes. You still gotta call plays and give your players the opportunity to make game changing plays.
This is awesome. Like you said, these calculations are rough and meant to demonstrate the type of thinking analytics staff do to help make these decisions. One amendment that I'd make is your take on LA's TD attempt at the end of the half instead of a FG. You wouldn't make the decision based on "which option has the greatest odds of getting points" but rather which option has greatest expected point value. For example, if FGs from the 3 yd line are converted at a 97% rate, you'd have an expected point value of ~2.91 (3*0.97), and if TD attempts from the 3 yd line are converted at a 45% rate, you'd have an expected point value of ~3.15 (7*0.45) (but slightly less when you incorporate the probability of an extra point conversion). So I would say that the decision to go for the TD is probably mathematically correct. Otherwise, this was such a great crash course into analytical thinking and really well done in a viewer-friendly way!
@@BrettKollmann no, thank you! Someone with your level of football knowledge explaining these concepts and why they are helpful decision-making tools is so helpful and appreciated ✌🏾
I’m sixteen and coaching my varsity football team while playing on it literally only because of Brett. My coach just realized I knew so much it would just be easier if I coached lol… and that can only be attributed to Brett
There's also that 2-point conversion late in the Raven's game against the Steelers that they failed on that I believe was also the right call, and almost succeeded.
The problem with an analytical approach is that it takes the fact that people aren't robots out of the game...people get frustrated or inspired or tired or defeated... when they did not get that touchdown the defense for the chargers and the team as a whole was demoralized...emotions effect your play...
That's week 3. This is December football and their second meeting of the season. Teams are more prepared for them. Not to mention the margin of error is near zero even if the playcall gets someone open and half the time it works and when it doesn't it goes terribly wrong
@@EZR725 The math is in the video. 7% of the time the worst case scenario happens. It's the right call as long as the players on the field wanted to go for it, and it seems like they did
The thing with the DSR in low-time situations is that DSR is primarily accounting for success when given time, having to go 40 yards in three to five plays is very different (and less likely) than having to go 40 yards in eight to fourteen plays, so the two point try makes even more sense, if the chargers are scoring on 20% of drives going in to OT vs the odds of the ravens winning which is 1-(chance of getting 2 points times chance of packers going fast and making a long field goal). Oversimplified a bit but the point stands. If you have a good two point play, make rogers beat you in 40 seconds rather than handing him the ball in OT. Like the video said, the problem is in the execution, your backup quarterback was predictable and didn’t check his other receiver who was open
Another game theory reason for going for 2 right away to go down 7 instead of going down 8 and trying to go for 2 at the end is time management. You’re going to have to go for 2 at one point or another so it’s much better to know if you got it or not right away so you can plan on whether you need 1 or 2 more scores to win.
dude, seriously? 7:48 in the 3rd quarter and you're like "OMG THE NUMBERS ARE TELLING THIS IS MY LAST CHANCE OR WHATEVER" wtf man. this is massive overthinking.
The thing I've always pondered is how determinant meta trends in football dictate analytics. For example: If the analytics say its overwhelmingly beneficial to go for it on 4th and 4 on the opponents 40 and the teams on offense start to play into that, then what happens to the meta? Was the likelihood of success so high to the relatively low chance of teams going for it in that situation, and that an increase in that situation would lead to a significant variation in the probability? And what about defenses using analytics of their own to better game plan the play the offense will run given that situation, hence potentially leading to the analytics indicating its no longer advantageous? So basically, I'm asking if analytics are as susceptible to change as schemes like the Wildcat, RPO, or bunch formations?
Exactly right. The decision to go for it or not is a process completely divorced from the play call. That is why you have to measure the correctness of the decision not on the result, but on how it came to be made.
There's also the momentum effect of converting on these 4th downs. The Bolts this year have been able to successfully maintain momentum of huge plays by extending drives on those 4th downs. The only game where we played conservative was the Vikings game and it looked like a team with no identity
I think the main problem is that statistical models are really most useful when describing high volumes of identical situations. The easiest example is of course the coin toss. There might be slight imperfections in the coin but mostly the factors are the same so one can predict a 50/50 result over many tries. The 61.9% 4th down conv rate does not mean the Chargers are that successful on 4th and goal with 5yds to go. Not to mention thousands of other variables that really make statistics really irrelevant in a NFL season where sample sizes are minimal and conditions are constantly shifting.
The model Brett was presenting is greatly simplified. The actual model the Chargers use is much more complicated. Of course even the best possible model will not be perfectly accurate given the low sample sizes and low number of possessions per game. However I think you're greatly underestimating our understanding. The models are still a more accurate representation of reality than the previous decades of gut intuitions represented. The proof is in the pudding. Teams are more aggressive now than ever because it works. Humans have a general tendency to prefer the conservative option over the new option if one can find even a couple edge cases that favor the conservative option. This need for overwhelming proof in order to adopt new ways of doing is evolutionarily useful because a wrong move can mean death and the end of your genetic line. It's worthwhile to shed these biases when the stakes aren't literally life and death like in football where there's always another down, another drive and another game. After all, we can only learn by trying. Every 4th down play is another data point.
Only thing that’s missing would be win percentage and expected points added. I feel like those r two more crucial percentages and numbers that coaches look at when making these 4th down decisions
Always enjoy your work. Also like your blogging set up. Do you have a list anywhere of the camera & lighting equipment you use? Or would you mind listing that below your videos? Thanks!
For the try at the end of the half, the thing to look at is the probability of scoring a touchdown vs. getting a field goal. If they have roughly a better than 40% chance of scoring a touchdown, they should go for it.
Hey, animations and graphics look FUCKING AMAZING. Just wanted to make that clear. I’m a bit behind on your videos so idk if they’re new in this video, but holy crap, it’s awesome
Nice video but several key points to talk about, not trying to hate but I think it's worth correcting the math (and I encourage anyone to correct/challenge me further because I didn't spend ages on this). Let's start with the math on likelihood of the Chiefs scoring a touchdown, this sort of underpins the video so is important. Firstly the assumption of 15 yards per series conversion is really important so it's worth actually calculating. Through the season the Chiefs are at 16.5 which does change the math quite a bit even though the numbers are close. Also if you are going to use Red Zone TD success rate (which I agree you should) you should calculate how many series to get to the red zone and then apply the rate, which will again change the odds. Let's take an example- starting at the 25 you initially calculated 5 seres conversions for a 24.3% success rate. Firstly you added a series by mistake because 0.79^5 is 31%, not 24.3% which would be 6 series. Right off the bat is a big difference, because close to 1 in 3 drives being a touchdown vs. close to 1 in 4 is a pretty big difference as a heuristic to tell a coach when making these decisions. Anyway just taking the actual yards per series conversion, 16.5, brings these odds up to 34.2%. That being said this isn't really relevant anyway because we're better off using the SCR up to the red zone then using the RCR. Doing that and using the 16.5 yards per first down brings the odds of getting to the red zone from the 25 to 45.5%, and the odds of scoring a touchdown at 27.0%. This is pretty close to the video but I can't give too much credit because if you are using the RCR in the video you should use it once they get in the 20, which would give the chance of scoring a touchdown from the 25 of 25.0% at 15 yards per first down. Overall there it doesn't make a huge huge difference from the 25 but it will make a difference from the 5. From the 5 if we use the actual yards/first down and the RCR from the 20 the Chiefs have a 20.3% chance of scoring a touchdown, which is 1 in 5 drives from that start from the five (wild). This is almost twice as more likely than stated in the video which obviously matters a lot. The other half of this decision which really matters is the likelihood that the Chargers score, which as others have mentioned is lower than their series 4th down conversion average due to the fact that it is 5 yards and in the red zone, both factors that make it more difficult. Ben Baldwin's model gave the play an estimated success rate of 35%, and an article from the 33rd team (www.the33rdteam.com/4th-down-decisions-2/) had it at 33.68%. I'll use the first one because I trust Baldwin's model, which would give the 'worst case scenario' a probability of roughly 13%, not huge but still almost double what is stated in the video. In terms of the decision itself I agree that Staley was right to go for it, Baldwin's win probability based model says to go for it, and my quick back of the napkin expected points from going for it is higher than kicking the field goal (2.99 vs 2.94), although it is pretty close. This does take into account Hopkin's extra point % this year too, just FYI. The go for it from the 1 is interesting as well. You are pretty definitive that the Staley made a mistake but decide to ignore the math and go back to classic football logic. The math says that he was again right to go for it - Baldwin's model has the play succeeding at 60% and on average adding 2% to the Charger's win probability. Looking at expected points going for it is clearly advantageous, at 4.49 expected points to 2.97. It's definitely not 'the mathematically proven thing to do' to just take the FG, and pretending it is is weird. Your analysis of the other decisions I don't have any real problems with, apart from a carry over of bad math from the beginning. I also agree with the overall sentiment about Staley and Harbaugh. Nice video overall!
Hey Brett, big fan. The value of going for it on fourth before the half is Probability of Touchdown * 7 - Probability of Successful Field Goal * 3 = Point Value of Going for it. So in a simple model of 61.9% fourth down conversion rate it would be 7*.619 - 3 =1.333 which is why they went for it.
Only at 7:27 but the problem seems to be how does your defense stack against the offense. They may have in general an 80% drive success rate, but what if their DSR jumps to 90% vs your defense. At least for the first drive where you don’t know how well you stack up in the matchup, I would lean taking points
Patreon link to vote for the next topic for everyone who is a $1 Patron or above: www.patreon.com/posts/60254665
It’s rig, nfl is the new WWE
Can you please do a video about the eagles run game and their O line and Jalen from the past 8-9 weeks. I think it would be a good video to show blocking schemes and different styles of runners.
Just wanted to say, you can be arrogant on here Brett. You singled out people (including videos of them in your videos), criticizing Mayfield for not making it work with OBJ. Now, it's time for you to eat some sh*t, on why you were wrong. You were wrong. Mayfield was the main reason reason it didn't work with OBJ. I love your channel, but your arrogance to criticize others, is why sometimes you look like a complete jacka$$. Hope you improve in the future.
On the 2nd , a fg there woulda been long range try and put KC in better field position 7 yards back.
@@m56wds3😊
it is unbelievable how good this channel is, Brett Kollmann putting out FREE content that most big name talk shows will never come close to in terms of quality
It really is insane how much better he is than ALL the TV 📺 pundits... go team Brett! 🔥
Because TV sports shows have to have as much content about every subject as possible between commercials. And they’re generally personality driven puff pieces designed to support the sport in all facets.
Thank you! I’m just glad you guys still like this show after all these years lol
@@patrickfuller6132 kkkkkkkkkl
Agree mate.
As an individual who does data analysis for a living and had to do analytics for my last job, and to some extent at my current job, the biggest issue I see is most people don't understand how to view analytics. I'm not just talking about the math but when people say "Analytics says you have to do X" well analytics doesn't tell you to do anything. To keep it simple, analytics gives you probabilities of what is more or less likely to occur in a given situation, it doesn't mean you should always do what the probabilities say you should probably do.
We don't ever look at a situation and think "Well, there is a 70% chance of this happening so I have to go for it." We look to see why it is 70%, what were the reasons for the 30% then assess the situation to see are we in a situation where we're closer to the 70 or the 30? Those of us who do this don't always go with what the probabilities say we should, we evaluate and will strategically go against the odds if we think the situation calls for it. The process for why something does or does not occur is vital and it's something every person who does data analysis focuses on, we don't simply focus on results, we focus on the process that leads to those results. Too many people think it's math nerds producing numbers, no, it's math nerds who look at the data, evaluate the data, evaluate the processes that led to the data, and use that to try and inform us in our decision making when looking at the future while being very aware of outside variables that can effect those numbers.
That's why when something doesn't work people in data analysis aren't surprised, we know even if the data says probabilities are high for converting on 4th down and 2 we know it's not going to work every single time, in fact the data says it won't work 100% of the time so we expect failures and we expect quite a few of them. It doesn't mean analytics didn't work, in fact it shows analytics does work because if teams are going for it on 4th and 2 and converting 100% we would go back and try to figure out how we got it so wrong since the data says 100% shouldn't be occurring.
Analytics are good, they've been used forever in some capacity, it's just now we have more data, more ways to gather and produce data to help evaluate what we're seeing.
What an awesome comment! I play poker for a part of my income and all 70 30's are indeed not equal lol. All coin flips are not either its really strange and hard to explain exactly why but I've played since i was 16 and am closer to 40. I can say without a doubt you are correct and only using the math as a reason to do something isn't all there is to a situation.
You’re correct, but it’s not everyone in analytics that thinks that way. Some people for PFF very much just say you have to do what the higher % play is in a vacuum, when the game is v much not played in one
@@ho31488 example i have just a open ended straight draw. Guy goes all in for 500, i have 500 my ev is obviously negative... only 8 cards help me with 2 cards to come. 32% about. However girl to my right makes horrible call with over pair not putting him on bottom set. Now my ev is positive and even though I'm going to still lose 68% of the time my EV is now positive!
@@SidelineSteve absolutely. Some analytics supporters, of which I'm obviously one, often make the same mistakes as detractors as repeating numbers but not delving into context.
Brett did a very good job here going over the variables involved.
I share the same sentiment. Been mostly dealing with the increase of information for the NBA for a decade now and the one thing I had to contend with are folks who odd enough want to help regurgitate a specific narrative and using numbers to correlate situations to support it. Then try apply the narrative through a broad brush stroke when they ignore the other important variables such as possible outliers or patterns that better support another perspective. Those folks I try stop having a discourse with because their aim IS to stop a discourse by jumping to conclusions and dying on a hill with how concrete the data is. After all we got more numbers from situations we aren't used to seeing so its best to give a refinement to the process for interpreting these situations.
Wow Brett I can’t believe you’d suggest analytics are ruining football! What a disgusting statement!
The nerds will pay for this
@@BrettKollmann 🤣🤣🤣🤣
Bengal here being the human computer he is
That was a disgusting act by Brett Kollmann -Joe Buck
I don't need to hear anything from you Bengal until i get that Riverside Royal drop! 😉
As someone who does analytics and statistical modeling for a career, I thought it was worth mentioning the decision right before halftime that you talked about. First things first: this is a great video and you did an excellent job walking through how analytics work and I think this video should be recommended to anyone who "doesn't understand analytics football".
Regarding the 4th down play, you can treat this specific example as an expected value problem. Put simply, all you are doing is comparing the EV of option 1 to the EV of option 2. Let's call option 1 kicking a field goal, and to get the EV you simply multiply the odds of converting with the value received. For the sake of simplicity, let's assume the odds of hitting a chip shot there are 100%. So the EV of option 1 (value)*(probability) = 3*1 = 3. For simplicity again, let's assume a TD is worth 7 even though it's not due to missed PATs, but it's easier to evaluate and the logic is the same. Put simply, you want the EV of going for it to be > 3 to make it worth going for it. To solve for that, simply plug it in to (value)*(probability)>3. 7*probability of converting > 3. Solving gets us Probability of scoring > 3/7 or about 0.43. This means if the Chargers have a chance of scoring that's higher than 43% then going for it is a higher expected value than kicking a FG. This is the statistic that was likely thrown out and if the Chargers are converting at that level or higher then going for it was the correct call.
You have a point that you can also consider other factors like if taking a guarantee negates the slightly higher EV of going for it, but from a purely numerical standpoint that is what the numbers say. And this brings us to a major point of analytics is that the results are always subject to interpretation or that there could be more to consider beyond just a simple EV comparison. A simple consideration is what if the odds of winning increase substantially if they had taken those 3 points but they don't increase by much more if they score a TD vs getting a FG. This is the challenge of analytics and as you pointed out it takes a lot to know when to strictly follow the numbers and also to identify the difference between bad analysis vs bad execution.
Great video, really loved your breakdown. For someone who claims to "suck at math", I think you killed it.
@Vagonius Thicket-Suede I think that's a very valid point and exactly one of those other factors I was talking about. If a FG increases your chance to win from say 52% to 55% but a TD goes from 52% to 65% then the analytics as well as outcome if you convert both say to go for it. And that's really all that goes into analytics in sports - expected value, risk assessment, probability, and regression analysis. There's still more to any game than these factors, but analysis in any sport is like knowing how to play blackjack optimally when going to a casino. It's not a guarantee it will work, but it is the best chance you can give yourself when it comes to decision making.
I was gonna basically make the same comment. While the guarentee of 3 points is nice, you would have to need some pretty extreme circumstances to give up 0.5-1 expected point in order to make going for it wrong. At 6 or 7 points down, I think that going for a FG and making it a two score game is more valuable, but even then, it's the Chiefs. Teams need to be much more aggressive against the Chiefs
6:26
It's actually about somewhere between 4% and 5%.
The odds of the chargers not succeeding is about 40 percent if the odds of them succeeding is about 60 percent. The odds of a failure(40%)* the odds of a cheifs TD from the 5(about 10%) = about 4%
Ah shit you’re totally right that’s my bad! That being said, that makes it even more insane that it happened tho!
@@BrettKollmann yeah
@@BrettKollmann that's why you did the intro!
Also unless I'm thinking incorrectly here, the idea of the 4th and goal at the end of the half is still "correct" from a probability perspective. Unless the Chargers red zone success rate is
@@johnterrell2825 it's correct, assuming 3 points are "worth" 3/7s of 7 points. It depends on the score differential and the difference in talent between the teams
Also having a backup SECONDARY out there isn't exactly working in your favor if you're the Ravens
In the words of Jon Finkel, a professional Poker and Magic: The Gathering player, when talking about the similarities between Poker and Magic: The Gathering: "I think the biggest thing is the deep seeded emotional understanding that the right play is the right play regardless of outcomes. The ability to make a decision 5 straight times, lose 5 times because of it, and still make it the 6th time if it's the right play." If you know and trust the math, then you've got to be willing to keep making that play, even if the last time you did it, it didn't work.
Of course, Football's a fair bit different from Poker and Magic, since the odds are ironclad in the latter two games and are more squishy in Football, but the point remains.
Can't say I was expecting to see a Magic: the Gathering and GOAT MTG player referenced in a football video, but I'm here for it.
I will live and die with Harbaugh. I love him. He will end up in Canton one day. Believe that. Just because they didn’t work well recently doesn’t mean it doesn’t help the team win.
Love your stuff Brett
I know I’m a biased Ravens fan but I rank him top 3 coach in the league behind Belichick and Reid
Which one?
@DeSean Jackson #Jaccpot #0ne0fone #1hunna Congrats! Managed to watch the video and came away with exactly nothing but the most basic observation possible. It's amazing how stupid people can be.
@@poeyplayz John Harbaugh not Jim
@@ProphetOfTruth_ thx
Honestly I’m so happy to hear that chargers player is doing ok. I’m a chiefs fan and that broke my heart on the chargers first drive when they went for it and he was unfortunately hurt
One of the best episodes ever! Amazing content and the video quality with all the visuals has gotten insane! Love to see the channel grow so much - great job Brett!
Thank you! Best decision I ever made was getting an editor that’s better than me lol
@@BrettKollmann Just wanted to say, you can be arrogant on here Brett. You singled out people (including videos of them in your videos), criticizing Mayfield for not making it work with OBJ. Now, it's time for you to eat some sh*t, on why you were wrong. You were wrong. Mayfield was the main reason reason it didn't work with OBJ. I love your channel, but your arrogance to criticize others, is why sometimes you look like a complete jacka$$. Hope you improve in the future.
First off, fantastic video. The explanation and analysis was great, as always.
Second, I wanted to mention a statistics concept called expected value, which could possibly better explain the early game calls (when time/score scenarios matter less, and you're just trying to maximize points). Basically, expected value is the average of all outcomes, weighted by how likely each outcome is. Taking the end of half play, say your probably of making the field goal is 100%, and the probably of getting the TD is 60%, roughly the 4th down conversion rate. Then the expected value of the FG is 3, and the expected value of the TD is 0.6*6+0.4*0=3.6, since there's a 60% chance of getting the TD and scoring 6, and a 40% chance of scoring 0 (accounting for PAT or 2PT is more complicated but would only increase the expected value, since we all outcomes are >=0). These results would indicate that on average, you'll score slightly more by trying for the TD than kicking the FG. Add on that you're playing a high-powered offence who you're expecting to score a lot, and you might be more inclined to take the high-risk high-reward option than if you were playing a more anemic offence.
As for whether or not this aspect of analytics is good/bad, I don't mind it. If anything, it generally makes coaches more willing to go for the exciting, high risk calls, than the safe (sometimes boring) old-school ones. It certainly opens then up to more criticism when thing go wrong, but also makes for more "oh my gosh I can't believe they did it" moments too. Anything that makes the game more exciting (while still being reasonably sane) is good with me.
This, thank you!
I was thinking of exactly that concept with the play with 3 seconds to go to end the half
Love the video as always Brett, especially the points about execution vs. decision-making!
Couple questions on the Chargers points.
1. When you say LAC has a 61.9% 4D conversion rate, does that factor in that they're more likely to convert 1 yard than 5? Feel like the math has to differ based on the distance, right?
2. And if not, why is the decision to go for it at the end of the half the wrong one? A 61.9% chance to get 7 points is an expected 4.33 points, which is higher than the 3 points you'd get for the FG (assuming it's made).
Your correct is an average of all there 4th down attempts, it's gonna change on distance. And yes your correct again the math at the end of the half is right too 60% at 6 points is better than 90% at 3. I would have to do a further deeper dive into the math of getting all 7 points though. However i agree that basically 100% 3 points to end a half is better than 40% of getting 0... so the math doesn't always give yourself the true best way to win. Football and math over lap but don't perfectly sink. This is why for so long football was conservative take the points rather than go for it. Breaking that mold is good for the game but never taking free money isn't the best call just cause math says it could be. I used the momentum argument why a goal line stand hurts your team even worse then whatever the math says because it doesn't factor that in, it can't!
@@Melcavic42 Additionally, lots of 4th and goal calculations bring into account where the opposing team will get the ball if the offense fails and this, how likely they will be to score on that new possession (i.e. turning the ball over on 4th and goal near the opposing goal line is much less punishing because the opposing team then had to drive the length of the field).
The decision to go for it before the half was the only one I really disagreed with because if you fail, you don't get the opportunity to take advantage of the opposing offense backed up deep and needing to drive the whole field (although I thought the Chargers should have gone for 2 to potentially go up 9 in the 4th). Hope what I'm broadly gesturing at makes sense.
1. That's right he was oversimplifying, but the models can account for the down, distance , field position, etc.
4th and goal at the 5 is around 35% conversion rate.
Unfortunately I do not have data splits by certain 4th down distances so it's tough for me to stay what their differences are in 4th and 1 vs 4th and 5. The teams themselves have that data for their own models tho.
@@BrettKollmann Davis Mills vid?
Man I feel like I need Scott Steiner to translate these percentages for me.
Great vid with plenty of information to digest but I think the final point made is what it's all about. Analytics have been around in sports under god knows how many names for as long as sports have existed so it's odd to think teams can win without them or that they're new. Having MORE info than ever before doesn't mean teams NEVER had information to base decisions off of. Conversely, teams/fans who only live by analytics will almost always miss out because they fail to value things that spreadsheet data can't quantify IE: who's on the field/what the conditions are like etc. So I guess the best answer on "Are analytics ruining football?" is Reverend Lovejoy's "No with an if, yes with a but!" No if you use them properly, yes if you're solely reliant on them for any decision you make. The idea teams can win without some combination of probability driven data and things like an understanding of talent, chemistry, momentum and circumstances feels like folks picking a side when there's really no reason to.
But analytics can account for weather and the roster. And I'm pretty sure math has disproved the concept of momentum
@@freddiesimmons1394 Funnily enough there was an ESPN article recently about momentum and whether it's real or not but I won't be disingenuous and act I've like read it yet.
@@fackes43 then I'll look it up
@@fackes43 I just read it. The most important quote: "Indeed there is ample data suggesting momentum isn't predictive of some new direction for the game, but rather a generic description of events that have already occurred. ...Instead, we fund their actual win percentage is just a tick less than the expected outcome, which... might signify that the team that was losing before the big play was likely the inferior team all along"
@@fackes43 I believe in momentum in only one sporting event I've ever seen: the falcons chokejob
This was an amazing look into the way analytics play into play calling. I feel that oftentimes statistics based decisions are made by a coach just staring at a spreadsheet of statistics and just screaming go go go into his headset. This video was amazing because it showed how coaches use those decisions to inform their gut.
Yeah I think it also depends if the coach has a specific plan to execute against something to exploit with the opponents' defense. Some coaches are conservative about it but they usually think of something pretty crafty to beat a defense to nearly favor that high leverage situation for them. In Staley's case since he's racked up so many 4th and 1 conversion attempts in different key situations (red zone, 20-30 yrd mark) its easier to pick holes more on his play calling side and the offense not executing rather than the decision itself to actually play that down instead of punting.
Refs are “ruining” football more then analytics are. But still neither of them are ruining football. Footballs still great
Fax
True. They're not ruining it into the ground, but are making it easier to blame them for Ls
Very well put. Refs are pretty bad though. But I’ve always been of the mindset if you leave it up to them you live and die with whatever happens 🤷🏽♂️
Analytics are ruining football media coverage - too much talk about percentages instead of the drama of the actual on-field deeds and human emotions that make sport meaningful.
No-one feels inspired by the story of who theoretically would win 51% of the time. (And in fact, each moment happens only once and is entirely under the players' control. They're actions, not randomly occurring data.)
...Commentators who think of sportsmen as tumbling dice miss the whole point of sport.
this is one of the best vidoes this channel has produced. Fantastic job. Clear, organized, kept it simple for the viewers while also including all the relevant info to make the points. Great job man
Thank you! I appreciate that!
I love you brought up execution. Many fans forget the players play, coaches coach. A bad play is not always on the coach, it can be execution in a lot of cases.
This reminds me of that one scene from Blue Mountain State where they bring in a new offensive coordinator and he brings up analytics by saying the enemy team gets its pressures from the left side and the head replies: No shit, their best player is the defensive end on that side. Analytics needs context to be truly useful. Really enjoyed this video Brett!
this really helped me understand the thought process behind analytical football, thanks Brett
Analytics isn't just math. It's a study of human performance and behavior. It's often used to develop a better and deeper understanding of what is going on with room for deviation (which is known as mathematical odds.
One of the big things I see with going for it on 4th down is that coaches usually implement a different play call than they normally have been doing. For instance, a team could be averaging 5 yards per play during the game and primarily using 11 personnel. But when they get into 4th and 1 from say their own 38 yard line, it's an Ace Jumbo package where the opponent knows they are going to run the ball about 95% of the time. The average yards per play from that formation and that play then dips to say 1.5 yards per play. Then they don't convert the 4th down and everybody blames the decision to go for it when the real problem was the actual playcall.
Or like on the Chargers 4th and 2, Herbert is looking to basically make a 1-read, quick throw and likely hurt the odds of converting compared to a play with slightly longer routes that allowed Herbert to make more than 1-read and if that's not open, take the ball and run for the first down himself.
It's basically a crux of it not being quite as easy as analytics people (like myself) tend to think it is, but it's nowhere near as difficult as the anti-analytics crowd makes it out to be either. Some of the things that Staley is going thru are things that Kevin Kelley wen tthru and then he started to learn the pitfalls to them and he adapted to help his teams use analytics even more to their advantage.
Yup well said. I always love the decision to go for it, but with the chargers the playcall always leaves me dumbfounded... (just look at thier last two weeks of this season)
Good job explaining some analytics! Success rates, situational football, possessions per game. A lot of what I use to explain up why it’s beneficial going for the fourth and shorts (or short enoughs haha). Having an aggression index to sort of tone up or tone down when you’ll go for it given game flow. It takes a good coach to understand how to bend it as in my defense is dominating them today, this WR is a matchup nightmare or maybe giving your own defense more of a break if you can maintain possession, or make theirs stay on the field.
In my mind I like being aggressive with a: dominant or good Oline (can win on short yardage, but also protect well) a great to elite QB (give more chances for a great QB to shred), good OC/playcaller (give playcaller a chance to shred), good matchup (maybe injury thinned a position group or just having your guy beating theirs all day).
The production in this video is incredible. The content was very interesting & visually it was very enjoyable as well. In particular I enjoyed the box that popped up when you were demonstrating the difference between Cook & Bolton on the 4th 4th down attempt. Extremely well done.
thank you! Just hoping to improve a little bit every episode
@@BrettKollmann Just wanted to say, you can be arrogant on here Brett. You singled out people (including videos of them in your videos), criticizing Mayfield for not making it work with OBJ. Now, it's time for you to eat some sh*t, on why you were wrong. You were wrong. Mayfield was the main reason reason it didn't work with OBJ. I love your channel, but your arrogance to criticize others, is why sometimes you look like a complete jacka$$. Hope you improve in the future.
Thank you Brett. I had to explain to my 70-year-old father who thankfully values my opinion all of this stuff in earnest, and he began to understand it, but not everyone has someone like me to explain the math and strategy behind it to them. I am sending this video to him to drive the point home, even though he already agrees with me, just so he can visually see the math, and how it was the play calls and/or execution that screwed these teams up.
another worthwhile note on the BAL/GB decision: by the end of that game the ravens had 10 safties/cornerbacks on the sideline... Rodgers would've won in OT without much trouble.
One of the best parts of Bretts channel is that he asks a question in the thumbnail then answers it within the first 2 mins of the video unlike most of these channels
The day analytics call a terrible PI that ruins the outcome of a game is the day I’ll say they have ruined football
Another point about the Chargers going for it to end the half. If they kick and make it it gives them a 7 point lead at half and also allows them to kick later in the game.
Love the rewind vhs effects, and all the other nuances in the editing. Great work as always!
Idk but that enjoy your family and stay out of trouble hit me the right way. Happy holidays!
I don’t think they are. The problem is a 17 game season is too small of a sample size. If every team followed analytics it would become obvious pretty quick that it’s the right thing to do.
Once again Brett giving us real in-depth analysis, not just some a talking head with a loud opinion. The quality of your content and the way you deliver it will always keep me coming back to watch more
Your lighting is amazing man, bravo. I took 3 film classes and lighting was the hardest thing for everyone learn and I still never got it right.
Please look into the percentages rate of winning a game when down by 7 points with about 4 minutes left in the game and your team is on defense. In sum, with onside kicks being a complete fail, once your down by 9 plus points with 2-4 minutes left, the game is mathematically over. That’s why you see so many bs calls when a team is down by 7 points in the middle of the 4th quarter. They know if a large portion of games were over by the middle of the 4th quarter, fans would want to reverse the onside kick safety rules or straight up remove it.
Not that your video editing and animations were bad before, not at all, but with this editor... Everything just looks so smooth and professional, I love it
One of your absolute best episodes ever Brett. Also, keep up the tremendous visual upgrades!
This is one of your best videos ever. Love hearing the explaination behind these decisions. Analytics is a tool at a coach's disposal in the same way that film study is. I think a lot of people believe that coaches live and die by it, but good coaches simply see it as insight into a situation to help guide their decision making.
Feel like there's a few a coaches who live and die by it to their own detriment. Some guys make a dumb decision and then say but analytics say it was the right call. Analytics don't take into count momentum, and game flow etc. Modern NFL feels like madden where the annoying guy you're playing goes for it every time regardless of the situation any time they get past their own 35.
You can crunch all of the numbers you want, it still comes down to the execution of said play
That's obviously true for any sport. Analytics are supposed to help guide teams' and coaches' decisions to put them in better situations to succeed over the course of a season.
Just because analytics say it's more beneficial to throw the ball on average than run it doesn't mean that you should never run the ball ever because that ignores situational context and the reality of the game. But if it pushes a team to be a little more willing to pass over the course of a season, it probably makes them more successful in the aggregate.
Execution can be improved on and in football is better if you can fool the defense at the same time. It just means further refining how to play that part of situational football. Belichick coaches his guys specifically for how to play in rainy/snow weather so at the very least he can say that his squad is prepared to execute better than his opponents in that kind of game. There are also specific special teams stuff he coaches so if there is an off situation comes up for that, his squad will be better prepared to make a play for it than his opponent at least.
5 seconds in and BOOM SIDE ANGLE. Levelling up the production sir.
as a stats nerd and someone with plans to become a sports analytics guy, I enjoyed this video so much. Please incorporate it into other videos. I could listen to it all day. Great job as per usual Brett
I think it’s important to understand that just like anything else, analytics are not always 100% successful. That’s not the point. Even if you could know every variable and intangible and they all said to go for it on 4th and 2, you can still fail. Using analytics will likely lose you some games that you should have won and win some games you should have lost. The point is that it’s designed to work over a large sample size, so theoretically it’s going to win you more games than it loses you in the long run.
It’s also important to keep in mind that just because a play didn’t work, doesn’t mean it’s was a bad decision. That happens, that’s football, and that’s life. I’m not saying going for it on 4th down every time were good decisions, but it’s not as simple as good result = good play and bad result = bad play.
Brett outlined this. Some of those 4th downs were good attempts but HORRIBLE play calling and execution. Staley may exactly be doing a lot of these to gather data in terms of refining the situation so he knows he is going fail on many of these. In the context against the Chiefs however they are good calculated risks overall because you don't want to be trailing against them especially in the 2nd half.
Thank you so much for all this great content on football. It's a sport I love, but an o so complicated one to fully grasp, and you make it really easy! Always entertaining and informative times whenever your videos come out! Merry Christmas to you mister Kollmann!
go brett! love what you’ve been doing with the channel, happy to watch you grow!
Another important point in the Ravens game: Harbaugh should have gone for 2 the second-to-last TD.
The math is hard to describe in a YT comment but I try:
50% to convert the 2ptc -> trailing by 6 = TD+XP wins the game in regulation 50%
50% to fail the 2ptc -> trailing by 8 -> 50% to convert the second 2ptc -> OT -> 50% to win in OT (probably less in this game) = 12.5% to win in OT
50% to fail the 2ptc -> trailing by 8 -> 50% to convert the second 2ptc -> OT -> 50% to lose in OT (probably less in this game) = 12.5% to lose in OT
50% to fail the 2ptc -> trailing by 8 -> 50% to fail the second 2ptc = 25% to lose in regulation
In summary:
62.5% to win
37.5% to lose
XP: 50/50 to tie
Coin flip: 50/50 to get possession
OT:50/50 chance to win
I dig this, keep it up. I respect that you're still putting out new types of content.
Score, Time remaining, yards to go, FG length, Team's ability needs to be crunched into the simulator. I do risk analysis for a living, you are definitely describing the process correct. The goal is to win, not maximize points. Going with FG, to put you up 10 vs going for it late in game is easy to see FG is better decision. The decision tool needs to figure how much time on clock makes going for it on 4th down the better decision.
Very nice video as always, Brett. However, I don't totally agree with your argument for the 2-pt conversion try in the Ravens-Packers game. You said it yourself: the chance of Rodgers getting to FG range are pretty high (~.48 I think you mentioned). So if the Ravens were playing not to lose first, their chance of not losing when kicking the XP based on DSR(-> getting to OT) would be P(tucker hitting XP, so probably .99) x (1 - .48). So pretty much .50. The chance of not losing when going for 2 would hence be P(getting 2pt) x (1 - .48), which would be way lower than .5.
So with that being said, why not try to get to OT, maybe win the cointoss and get a nice drive for your offense. I know, the chances of Rodgers beating you in OT would also be substantial, but I just can't see how chosing the quick death is more desirable.
With that being said, thank you so much for the great content you put out for us. Happy holidays to you and your family!
I can't comprehend the math you just did but I would have kicked the field goal and bet on my defense
Analytics in football is like counting cards in blackjack. If you follow basic strategy and listen to the math, you're guaranteed to win over the long term. It may not work out every single time, but it will absolutely work out over the long term
Yeah that’s cool and all, but football and blackjack are completely different
@@emmettberryhill9449 Yes, but the basic strategy is the same. Follow the statistics and follow the math. Those both exist for a reason. The Chargers objectively made the right decision every time, except for maybe the one right before halftime
Hey numb-skull.... you can play like 500 hands of blackjack in a night... You play SEVENTEEN football games in a year. Variance is and SHOULD be a factor.
@@d3eztrickz Okay? And I addressed that when I said it's not going to work out every time. There are card counters Who have lost $20,000 or more a night for seven straight nights. The point is that over the long term, the strategy is going to work
@@d3eztrickz And those same people still earn a solid profit of over $60,000 a year, or more. And I'm probably being conservative
Nice touch with the sfx of the film reel clacking and the snow on the screen!
Even without the analytics (which I think is good for the most part as I think it encourages more 'aggressive' decision making) if you're the Ravens you can look at it like this:
You tie the game with a PAT, but now in order to win the game you must:
1. Stop Rodgers, with 40 some seconds left and who only needs FG (as a Niner fan, I've seen this a couple times...)
2. Go into Overtime and hope that you win the coin toss.
3a. Stop Rodgers from getting a TD if you lose the toss and score a TD yourself
3b. Score a TD yourself if you win the toss or hold Rodgers to no points if you only manage a FG
*Keep in mind, we're assuming that your OT TD drive is going to have to be somewhere between 70-80 yards.
Alternatively you can go for 2, which then you must:
1. Convert a 2pt conversion from the 2 yard line.
2. Stop Rodgers from getting into FG or hope the FG misses
Yeah which one sounds easier? Take away the mathematics of DSR, and considering the context of who you are playing against... it still makes sense to go for 2. I think you summed it up best Brett: "You either die quickly or die slowly"
its one of those situation where you go for 2 and the problem wasn't the decision. it was a horrible playcall and was totally expecting some sort of spread empty designed qb keeper or draw play
Going for 2 against the Packers makes total sense for the reasons you mentioned. That same decision against the Steelers was not the right call imo since the Ravens are much better than the Steelers, and even with Humphrey out, the Ravens probably would've won in OT whether they got the ball first or not. That Steelers offense is hot garbage 😂
"Take away the mathematics" and going with "which one sounds easier" is throwing out the instructions and trying assemble the furniture by eye
@@freddiesimmons1394 I'm not saying you disregard the analytics, I'm saying that even if you were to, it would still make sense to go for 2. The eye-test can be deceiving and the statistics can be misleading. Usually it requires a combination of both to evaluate how good a team is.
class vid, merry xmas brett
Yeah. I’m So glad I stumbled on Brett one day. Will never look back for finally finding a smart and high quality football analysis channel. Not enough of them around and none even near Brett. Keep it up dude
holy shit i’m so glad i found this channel. big network football shows don’t even compare
Regarding the 4th down at the end of the half: taking the "higher-odds of points" isn't always the best option. Field goals are pretty automatic at that range, so you have an "expected points" of 3. The chargers would then look at their probability of getting a touchdown. They multiply that probability by 7 points to calculate their "expected points" from going for it. If (theoretically) their probability of scoring a TD is 50% from that position, then their expected points is 3.5 (7x0.5). Thus they should go for it rather than kick the FG.
As James Marshall mentions, you also just have to consider the context: are there advantages going into halftime with a guaranteed 3 points, or possibly having the momentum of a TD?
Halftime kinda ruins momentum though. Just take your points I think allot of these coaches are out coaching themselves with these things.
Great video as always Brett.
One thing you got me thinking about is the notion of DSR and how it translates to points and field position. Obviously this was only "napkin math" as you described, but I was curious about how big plays or chunk plays factor in. Teams rarely march down the field picking up first downs 10-15 yards at a time all the way and many drives are successful based on if a team can rip off a long play and eliminate some of the opportunity for failure over the course of a longer drive with more plays.
Anyway, I'm no math whizz, but something that caught my attention. Thank you!
Analytics are a great rule of thumb. The problem can be whether or not your model makes assumptions correctly. Even if those assumptions are generally correct they might not be in a specific situation (i.e. a team has had major injuries in a game, the weather is a factor, etc).
Merry Christmas to you and your loved ones Kollmann.
Now let's watch our gift community.
i come here because you are a straight guru. not for 1 particular type of video or breakdown. but because you make me(a couch coach) feel like i kno what im talking about.
thanks Brett for all of your content, video, breakdowns and everything else.
Happy holidays to you too Brett. Thanks for all the videos you've put out over the years. This was definitely a present of sorts getting a video of my Chargers right before the holidays so thank you for everything you do Brett 🙏
“a f*cking moron” that was hilarious asfff 😭
I just don't know where they get the numbers from. I watch just about every game to some extent on Sunday and I've almost never seen going for it on 4th down every time work. Or going for 2 down by one with over 30 seconds left on the clock
The drive success rate statistic Brett is using in the video comes from Football Outsiders (or at least, that's where I found it). For other statistics, people in the games just chart every play. So, for example, unless you watch every Chargers game and do the same for their 4th down conversions, it's hard to see just how often they succeed on these 4th downs. And again, it is hard to do which is why the NFL or analytics websites pay people to chart all the plays
Analytics work in baseball because it's an 162 game season, over that large a sample size any advantage gained pays off in the long run. In the playoffs though, which more resemble the brevity of a NFL season, it has mixed success (Ex Blake Snell 2020 World Series). The NFL regular season is almost 10x shorter meaning any gain from analytics is miniscule. There's definitely a part for them in the game, but nothing replaces experience and the eye test when it comes to a do or die play.
I'd also like to point out analytics has ruined baseball whereas in the NFL it leads to an exciting 4th down play, so I wouldn't worry about it too much as a fan.
I see this a lot, why has analytics been bad for baseball?
Awesome point and why i think football stayed very conservative for decades. I totally agree that even in black Jack you have to know when to just stay in the game and hope rather then try for a higher point total in such a short period of time. Great comment
@@calincampbell5123 Well, where do I start lol. This will be long. It's led to the decline in the starting pitcher and reliance on bullpens. Starting pitchers are stars whereas bullpen arms are mostly nobodies. It would be like pulling Brady or Rodgers in the 3rd quarter because stats tell you their backup does better then since the defense hasn't seen them yet. There's an emphasis on home runs, which are exciting plays but only happen a few times a game on average, and in the meantime has led to more strikeouts and less base hits, meaning less action. On the flipside pitchers are told to strikeout people, which has led to again less action, and burns out pitchers. Defensive shifts have always been a part of the game but are everywhere now, meaning less base hits, and batters trying for home runs to hit over it, which problems I said above. On Base Percentage is the new "Golden Stat" instead of average meaning walks are just as good as hits, leading to longer counts, longer games, and less action because walks are boring. On the financial side, teams realized it's way cheaper to pay young players entry level salaries instead of 30+ year old free agents, leading to tanking for the draft. Those are the main ones, there's many others.
making the right play is correct even if the variance is higher due to smaller sample size.
@@calincampbell5123 The game is boring. Home run, Walk, Strike out. They have reduced the importance of defense, base running/stealing, etc.
Theo Epstein worked on a Poll with MLB where they found out that triples, stolen bases, and defense were actually in the top 5 “most exciting aspect of baseball” to casual fans.
I wish more people separated play call vs execution of play when talking about the malcolm butler pick
The oversaturation of football analysis is ridiculous, but I still LOVE watching these videos because they are always levelheaded and bring something new to the conversation. Keep up the great work Brett. You're awesome. Merry Christmas. Love live Mac Jones ;-)
Merry Christmas!
Incredibly interesting video Brett, thanks for doing it! Merry Christmas
Chiefs vs Chargers was my fav game of the year. They played like how i do in Madden lol. playing 4 downs makes defending so much harder and the game is more exciting tbh
Coming from someone who works in an analytical field, I thought this was pretty well done. One thing I would have liked to see you do is use expected values, for instance a 25% chance at a FG and a 25% chance at a TD are not the same thing. This would have been helpful when deciding if the math played out on the Chargers 4th & goal before half. Let's assume from that distance the FG make rate is 100%, so by kicking a FG the expected value is 3. Now, let's assume the odds of the Chargers scoring a TD was 47% (roughly the odds of converting from there), that would give the Chargers an expected value of 2.82 + another 0.93 for the XP (93% chance of making the XP). That gives the Chargers an expected value of 3.75 by going for the TD. So, yes the math did say to go for it there as opposed to kicking the FG.
This is hands down the best video I’ve seen in months. Thank you, Brett, for the amazing content.
Although I do understand what you're saying particularly with the packers game it seemed to that the ravens were heating up offensively and defensively (hell even got a 3 and out the drive before) i think the ravens had a decent enough shot at keeping rodgers out of fg range and extending the game in ot. Plus given rodgers and the packers luck in ot its highly likely that ravens atleast score 3. Call me old fashioned but its better to fight for another drive then not to fight at all
Yes I've also heard at home you tie the game, away you go for 2 because already the computer's have figured out your basically already up 3-0 at home! Even the bet line's figured that out years before the tech did! Anyway in this instance i agree you tie the game at home!
Packer fan here agrees 👍
If you want a good subject for a follow up on this analytics topic, peep the Bills vs Pats game. McDermott def paid attention to details as he executed on ideas he would need to get the one up on Darth Hoodie. Went very aggressive on early downs so he didn't see many 3rd and longs that would've been difficult to convert. And when he didn't, he KNEW from then that they were going for it in 4th (and consequently short) down situations. To the point where he did NOT PUNT against Belichick at all the whole game. He also didn't try 'force' the run and designed the offense and play calling to the limits of Josh Allen's full utilization. He was the pass game AND the potent run game. I think this game is where analytics and game situation met with the execution from the right personnel to get the winning results! Especially against an opponent who was playing pretty conservative despite his opposing coach (McDermott) being very aggressive the entire game.
“I’m am what the kids today call a fucking moron “
I wasn’t expecting that 😂😂😂😂😂😂😂😂😂😂😂😂😂😂😂😂
I love your videos brett I got your jet chip wasp hoodie for Christmas and has been rocking it since, great quality merch!!!
Danny Kleinman, one of the first analyst of Backgammon, compared Football with Backgammon. He even wrote a Book with Daniel Ortega about a crucial match standing, four away vs. two away. It's like a 4th and go! decision in a late football/loopsided match. A play call is here a doubling decision, whereas his execution a decision is about moving the checkers. At 4a vs. 2a, the trailing player can take a double, doubling back in the next move, thus doubling out the gammons of the leading player. The doubling and take points are shifting, depending also on the GWC%. So are late game decisions in Football. And don't forget: Frank Frigo, the CEO of edjsports, won once apon a time the Backgammon World Championchip in Monte Carlo. Their engine is like 'extreme gammon' for Backgammon. Would you backup Frank or the computer?
As a Ravens fan with an MS in Statistics, I can say that John Harbaugh sometimes uses analytics as a crutch when they should just be a tool.
I agree with the 2-pointer against the Steelers, because we were on the road, the Steelers had come back from a deficit, and game flow was not in our favor.
I don't agree with the 2-point call against the Packers, because we were the ones who had come back, we were at home, and our defense was actually playing well.
Analytics are great, but football isn't played on a spreadsheet. The best users of analytics will balance it with a strong coach's intuition.
I couldn't agree more. I have no issue with being aggressive but you should never chase points in the 3 quarters of a game. Especially when you have the greatest kicker of all time. Tucker is 98.7% successful on kicks inside 40 yards including extra points.
I personally agree with both 2 pt attempts by harbaugh purely because of the awful overtime rules in the NFL. I just think it’s disrespectful to the effort your players put in for 4 full quarters to tell them “hey we’re gonna flip a coin to find out if we have a chance to win”. I think that if you have confidence in your offense to score in OT you should have confidence in them to convert a 2pt
@@samuelault4723 I don't mind the overtime rules. The defense gets paid too, and touchdowns aren't just guaranteed. The Ravens were in a spot to force overtime because they forced 2 punts late in the fourth quarter.
yeah cause Aaron Rodgers has never made something happen with a minute or less to go in a game
@@Benisuber1 if he had, the extra point wouldn't have helped us anyway
The sad thing is with Harbaugh's decision is that we'd have to stop Rodgers either way to win. It essentially becomes do we trust ourselves to punch it in from 2 yards out on one play or not...and while Harbs may have made a good moral decision in trusting his guys, I can't say I imagine we score in most of those situations
Very interesting video. Its hard to measure things like momentum and execution. You can make informed decisions but players still gotta make plays
Wow Brett, for me as a Casual that was incredibly fascinating. Thank you for your hard work all year and Merry Christmas!
People gotta give Staley a break bc LAC was having massive kicker problems for nearly the entire season
At the same time, statistics are not pre- determined outcomes. You still gotta call plays and give your players the opportunity to make game changing plays.
This is awesome. Like you said, these calculations are rough and meant to demonstrate the type of thinking analytics staff do to help make these decisions.
One amendment that I'd make is your take on LA's TD attempt at the end of the half instead of a FG. You wouldn't make the decision based on "which option has the greatest odds of getting points" but rather which option has greatest expected point value. For example, if FGs from the 3 yd line are converted at a 97% rate, you'd have an expected point value of ~2.91 (3*0.97), and if TD attempts from the 3 yd line are converted at a 45% rate, you'd have an expected point value of ~3.15 (7*0.45) (but slightly less when you incorporate the probability of an extra point conversion). So I would say that the decision to go for the TD is probably mathematically correct.
Otherwise, this was such a great crash course into analytical thinking and really well done in a viewer-friendly way!
Thank you! I’m still trying to learn a lot about the subject myself so that little EPA note is helpful, thank you!
@@BrettKollmann no, thank you! Someone with your level of football knowledge explaining these concepts and why they are helpful decision-making tools is so helpful and appreciated ✌🏾
I’m sixteen and coaching my varsity football team while playing on it literally only because of Brett. My coach just realized I knew so much it would just be easier if I coached lol… and that can only be attributed to Brett
There's also that 2-point conversion late in the Raven's game against the Steelers that they failed on that I believe was also the right call, and almost succeeded.
The problem with an analytical approach is that it takes the fact that people aren't robots out of the game...people get frustrated or inspired or tired or defeated... when they did not get that touchdown the defense for the chargers and the team as a whole was demoralized...emotions effect your play...
Obviously not lol. People ripping on Staley clearly forgot that the main reason the Chargers won in week 3 was because he was aggressive on 4th down
That's week 3. This is December football and their second meeting of the season. Teams are more prepared for them. Not to mention the margin of error is near zero even if the playcall gets someone open and half the time it works and when it doesn't it goes terribly wrong
And a shit defense. But don't forget that like you did.
@@EZR725 The math is in the video. 7% of the time the worst case scenario happens. It's the right call as long as the players on the field wanted to go for it, and it seems like they did
The thing with the DSR in low-time situations is that DSR is primarily accounting for success when given time, having to go 40 yards in three to five plays is very different (and less likely) than having to go 40 yards in eight to fourteen plays, so the two point try makes even more sense, if the chargers are scoring on 20% of drives going in to OT vs the odds of the ravens winning which is 1-(chance of getting 2 points times chance of packers going fast and making a long field goal). Oversimplified a bit but the point stands. If you have a good two point play, make rogers beat you in 40 seconds rather than handing him the ball in OT. Like the video said, the problem is in the execution, your backup quarterback was predictable and didn’t check his other receiver who was open
Another game theory reason for going for 2 right away to go down 7 instead of going down 8 and trying to go for 2 at the end is time management. You’re going to have to go for 2 at one point or another so it’s much better to know if you got it or not right away so you can plan on whether you need 1 or 2 more scores to win.
dude, seriously? 7:48 in the 3rd quarter and you're like "OMG THE NUMBERS ARE TELLING THIS IS MY LAST CHANCE OR WHATEVER" wtf man. this is massive overthinking.
The thing I've always pondered is how determinant meta trends in football dictate analytics. For example: If the analytics say its overwhelmingly beneficial to go for it on 4th and 4 on the opponents 40 and the teams on offense start to play into that, then what happens to the meta? Was the likelihood of success so high to the relatively low chance of teams going for it in that situation, and that an increase in that situation would lead to a significant variation in the probability? And what about defenses using analytics of their own to better game plan the play the offense will run given that situation, hence potentially leading to the analytics indicating its no longer advantageous? So basically, I'm asking if analytics are as susceptible to change as schemes like the Wildcat, RPO, or bunch formations?
"Who's winning the game?"
"Nobody. Miami is just losing more slowly."
-from the 70's
Exactly right. The decision to go for it or not is a process completely divorced from the play call. That is why you have to measure the correctness of the decision not on the result, but on how it came to be made.
There's also the momentum effect of converting on these 4th downs. The Bolts this year have been able to successfully maintain momentum of huge plays by extending drives on those 4th downs. The only game where we played conservative was the Vikings game and it looked like a team with no identity
I'm pretty strongly under the idea of momentum is largely unsubstantiated
I think the main problem is that statistical models are really most useful when describing high volumes of identical situations. The easiest example is of course the coin toss. There might be slight imperfections in the coin but mostly the factors are the same so one can predict a 50/50 result over many tries. The 61.9% 4th down conv rate does not mean the Chargers are that successful on 4th and goal with 5yds to go. Not to mention thousands of other variables that really make statistics really irrelevant in a NFL season where sample sizes are minimal and conditions are constantly shifting.
The model Brett was presenting is greatly simplified. The actual model the Chargers use is much more complicated. Of course even the best possible model will not be perfectly accurate given the low sample sizes and low number of possessions per game. However I think you're greatly underestimating our understanding. The models are still a more accurate representation of reality than the previous decades of gut intuitions represented. The proof is in the pudding. Teams are more aggressive now than ever because it works.
Humans have a general tendency to prefer the conservative option over the new option if one can find even a couple edge cases that favor the conservative option. This need for overwhelming proof in order to adopt new ways of doing is evolutionarily useful because a wrong move can mean death and the end of your genetic line. It's worthwhile to shed these biases when the stakes aren't literally life and death like in football where there's always another down, another drive and another game. After all, we can only learn by trying. Every 4th down play is another data point.
Only thing that’s missing would be win percentage and expected points added. I feel like those r two more crucial percentages and numbers that coaches look at when making these 4th down decisions
This is becoming my fav channel
Always enjoy your work. Also like your blogging set up. Do you have a list anywhere of the camera & lighting equipment you use? Or would you mind listing that below your videos? Thanks!
I’ll start putting that in the description below, good idea!
For the try at the end of the half, the thing to look at is the probability of scoring a touchdown vs. getting a field goal. If they have roughly a better than 40% chance of scoring a touchdown, they should go for it.
Hey, animations and graphics look FUCKING AMAZING. Just wanted to make that clear. I’m a bit behind on your videos so idk if they’re new in this video, but holy crap, it’s awesome
Nice video but several key points to talk about, not trying to hate but I think it's worth correcting the math (and I encourage anyone to correct/challenge me further because I didn't spend ages on this).
Let's start with the math on likelihood of the Chiefs scoring a touchdown, this sort of underpins the video so is important. Firstly the assumption of 15 yards per series conversion is really important so it's worth actually calculating. Through the season the Chiefs are at 16.5 which does change the math quite a bit even though the numbers are close. Also if you are going to use Red Zone TD success rate (which I agree you should) you should calculate how many series to get to the red zone and then apply the rate, which will again change the odds.
Let's take an example- starting at the 25 you initially calculated 5 seres conversions for a 24.3% success rate. Firstly you added a series by mistake because 0.79^5 is 31%, not 24.3% which would be 6 series. Right off the bat is a big difference, because close to 1 in 3 drives being a touchdown vs. close to 1 in 4 is a pretty big difference as a heuristic to tell a coach when making these decisions. Anyway just taking the actual yards per series conversion, 16.5, brings these odds up to 34.2%.
That being said this isn't really relevant anyway because we're better off using the SCR up to the red zone then using the RCR. Doing that and using the 16.5 yards per first down brings the odds of getting to the red zone from the 25 to 45.5%, and the odds of scoring a touchdown at 27.0%. This is pretty close to the video but I can't give too much credit because if you are using the RCR in the video you should use it once they get in the 20, which would give the chance of scoring a touchdown from the 25 of 25.0% at 15 yards per first down. Overall there it doesn't make a huge huge difference from the 25 but it will make a difference from the 5.
From the 5 if we use the actual yards/first down and the RCR from the 20 the Chiefs have a 20.3% chance of scoring a touchdown, which is 1 in 5 drives from that start from the five (wild). This is almost twice as more likely than stated in the video which obviously matters a lot. The other half of this decision which really matters is the likelihood that the Chargers score, which as others have mentioned is lower than their series 4th down conversion average due to the fact that it is 5 yards and in the red zone, both factors that make it more difficult. Ben Baldwin's model gave the play an estimated success rate of 35%, and an article from the 33rd team (www.the33rdteam.com/4th-down-decisions-2/) had it at 33.68%. I'll use the first one because I trust Baldwin's model, which would give the 'worst case scenario' a probability of roughly 13%, not huge but still almost double what is stated in the video.
In terms of the decision itself I agree that Staley was right to go for it, Baldwin's win probability based model says to go for it, and my quick back of the napkin expected points from going for it is higher than kicking the field goal (2.99 vs 2.94), although it is pretty close. This does take into account Hopkin's extra point % this year too, just FYI.
The go for it from the 1 is interesting as well. You are pretty definitive that the Staley made a mistake but decide to ignore the math and go back to classic football logic. The math says that he was again right to go for it - Baldwin's model has the play succeeding at 60% and on average adding 2% to the Charger's win probability. Looking at expected points going for it is clearly advantageous, at 4.49 expected points to 2.97. It's definitely not 'the mathematically proven thing to do' to just take the FG, and pretending it is is weird.
Your analysis of the other decisions I don't have any real problems with, apart from a carry over of bad math from the beginning. I also agree with the overall sentiment about Staley and Harbaugh. Nice video overall!
Hey Brett, big fan. The value of going for it on fourth before the half is Probability of Touchdown * 7 - Probability of Successful Field Goal * 3 = Point Value of Going for it. So in a simple model of 61.9% fourth down conversion rate it would be 7*.619 - 3 =1.333 which is why they went for it.
As a charger fan your take on staley is perfect , we won a lot of games because of those calls , sometimes the execution is just off , thats football
Only at 7:27 but the problem seems to be how does your defense stack against the offense. They may have in general an 80% drive success rate, but what if their DSR jumps to 90% vs your defense. At least for the first drive where you don’t know how well you stack up in the matchup, I would lean taking points