I Simulated The Premier League Run-In 4,000 Times.
Вставка
- Опубліковано 14 лют 2024
- In this video, James Allcott simulates the Premier League season 4,000 times to find out who wins the title, which teams can qualify for the Champions League and which team could be relegated.
UnderthePitch: / underthepitch
How it works:
The simulations work by predicting each game for the rest of the season and updating the table as it goes.
The predictions are made by a (simple) AI model. The model looks at 10 factors:
Points of the home team
Average home points-per-game of the home team
Average goals scored at home for the home team
Average goals conceded at home for the home team
Form of the past 5 games for the home team
Points of the away team
Average away points-per-game of the away team
Average goals scored away for the away team
Average goals conceded away for the away team
Form of the past 5 games of the away team
The model doesn’t predict an exact outcome. It gives probabilities of each team scoring a certain amount of goals. For example, for the home team the prediction might look like:
0 goals: 30%
1 goal: 40%
2 goals: 20%
3 goals: 6%
4+ goals: 4%
This process is repeated for each game left in the season, and the final positions are saved. Then this is repeated for ‘x’ number of simulations, to allow a spread of results to be generated.
JOIN THE DISCORD SERVER
👉 / discord
JOIN US ON THE JAFFN PATREON 🎧
👉 / jamesandflav
FOLLOW JAMES ON TWITTER:
👉 / jamesallcott
FOLLOW JAMES ON INSTAGRAM:
👉📸 @jlallcott jlallcott?...
James Allcott content focuses on Premier League, Champions League, EFL Championship talking about Manchester United, Arsenal, Liverpool, Man City, Tottenham, Barcelona, England, Chelsea, Jurgen Klopp, Erik Ten Hag, Mikel Arteta, transfer news, tactical analysis and much more. James has made content with Mark Goldbridge from the United Stand, Rory Jennings, Ben Foster, Spencer FC, The Club, Thogden, JaackMaate, ESPN and many more.
#PremierLeague #JLA #footballit - Спорт
![[柴犬ASMR]曼玉Manyu&小白Bai 毛发护理Spa asmr](http://i.ytimg.com/vi/0TsXQ7z2Dh4/mqdefault.jpg)
Many thanks to @UnderthePitch on X for his amazing help on this video! 🙏🏼
Here’s a message from him on how this works:
The simulations work by predicting each game for the rest of the season and updating the table as it goes.
The predictions are made by a (simple) AI model (I don’t like using AI as it is such a buzzword nowadays). The model looks at 10 factors:
Points of the home team
Average home points-per-game of the home team
Average goals scored at home for the home team
Average goals conceded at home for the home team
Form of the past 5 games for the home team
Points of the away team
Average away points-per-game of the away team
Average goals scored away for the away team
Average goals conceded away for the away team
Form of the past 5 games of the away team
I have built my own dataset of about 26,000+ games, spanning 10+ leagues and 10+ seasons in order to train this model to look at these above 10 factors for each game and look at the final outcome. This was the toughest bit of the project, as there are no datasets out there that have these exact factors taken right before a game was played. For example, if I’m training on a Brighton vs Wolves game from 2018, I need to know all of the values of all of these factors right before that game was played (like what was the form of these teams going into the game, what was Brightons average home points-per-game that season right before that game was played, etc).
From this, the model has no pre-existing knowledge of individual teams. It just looks for patterns and trends between the aforementioned factors and the final outcome of a game for all games it is trained on. For a simple example, it will probably learn that if the home team score a lot at home, and the away team concede a lot away from home, then the home team will be predicted to score a lot. It will learn this, along with other patterns that you or me might not identify at first glance.
Some other models out there include things like xG, etc. which has its benefits, however, I wanted to keep my factors simple in order to be able to use this on smaller leagues where xG data is unavailable.
The model doesn’t predict an exact outcome. It gives probabilities of each team scoring a certain amount of goals. For example, for the home team the prediction might look like:
0 goals: 30%
1 goal: 40%
2 goals: 20%
3 goals: 6%
4+ goals: 4%
Then, the model will use a weighted randomiser to ‘predict’ the number of goals scored by the home team. Another way of imagining this is thinking of putting everything on a scale, where ‘0 goals’ represents 0-30, ‘1 goal’ represents 30-70, ‘2 goals’ as 70-90…. Etc. Then the model will pick a random number between 0-100, and wherever it lands, it selects that number of goals. This is done to allow the ‘most likely’ outcome to still occur most of the time, however, there are also small chances of upsets occurring (like Hull finishing 2nd in 0.05% of simulations).
So this process is repeated for each game left in the season, and the final positions are saved. Then this is repeated for ‘x’ number of simulations, to allow a spread of results to be generated.
It’d be interesting to see how this model would compare to a model using expected goals and expected goals against. Would be interesting. Much appreciated though!
A bit confused you prioritized low level leagues over high quality statistics in order to simulate a high level league
So Liverpool got the highest average because of their total home games to the end of season?
I’ll take 4% at top 4 given United’s goal difference 😅
Need that “could they” bit every week 😂
Take a drink every time the "could they" segment appears
@ I found it completely unamusing and distracting. Thumbs down from me.
We're all living in a simulation... Specifically, simulation 3971 where little old Brentford get walloped back to the championship... Could they?
Could they?
Liverpool fan. My brain predicts Man City as they have Pep, Haaland, KDB and broke 1000 ffp rules.
Its fairly simple
@@Dwayndibb150*
At the end of the season can you do another video showing if any of the 4,000 simulations were right? And when none of them are will you revisit your methodology?
Yes, if none of them are correct, let’s get together and laugh at AI and simulations and predictive analysis.
The prediction for each team has an accuracy of 1% better than flipping a coin. Check out the lad who did it.
The likelihood of him getting a correct league table from 4000 simulations is 0.
Best creator out there james keep it up❤
Amazing stuff James, more like this please!
Loving the content Jim!
Villa has an awful injury crisis at the moment mainly in the defensive unit. They are down to their 4th tier defenders (Chambers and Lenglet) with Konsa, Torres, Mings, Carlos all sidelined. Plus the team has had 3 major knee injuries with Mings, Karama, and Buendia. I will be surprised if Villa finish in 6th I think top 4 is out of the question at this point.
They were in 2nd because of the lack of any injuries in the first half of the season anyway. It was only a matter of time till that luck caught up with them.
Stop making excuses , Tottenham played two fullbacks in centre defence, No Maddison, Son, Bissouma and Papa Sarr and still ground out results, getting 3-3 away to the ethihad, beating Newcastle, in form Everton and Bournemouth with a patched up team,
Another comment already talked about Tottenham, United have missed there entire starting midfield and defence (and second choice defence as well) outside of Bruno for half the season. Everyone has had injury issues, they're just getting the same problem everyone has already had
Villa's injury crisis has gotten worse but we were 2nd despite Mings, Buendia, Ramsey and Moreno barely kicking a ball and just about every player missing games, Watkins is our only ever present. The idea we haven't had major injury problems is ill informed rubbish.
Tonight after a well deserved win at Fulham we are back into 4th and 12 points above 7th placed Newcastle.
With 13 games to go I'd be amazed if we don't finish in the top 6.
liverpool have 7 first team players injured
I enjoyed this video last year so I’m excited to see how this one compares in a few months
The "would they?" alone, gets you a like on this video sir, so good 😂
Great video Jimbo
Glad you enjoyed it
Holy crap, this video was actually amazing. I thought it was gonna be boring but I’m thoroughly impressed.
I love the silly cutaways, especially when youre not looking at the camera haha
Wanna run that 4000 simulations again?! lol
Looks like you forgot to add the 'Spursy' variable. This sim is as close to top 4 as they're getting.
finally you ran a sim that had variables
Good video 😊
Do you do predictions for matches ?
Out of curiosity, did this factor in the chance some teams might drop points due to charges? Or that Everton might get some of their points back?
No , they just subtracted 10 pts from whatever theor simulated points total is
What software are you doing these simulations on?
Even if it’s a short, at the end of season be interesting to see the closest prediction and the worst one
This is very interesting to see 👀
I reckon this would be more interesting if it was done in a few weeks time. Everyone has very similar run ins because there's 14 games left
Man Ill take 40 percent (champions league) for Villa with all the freaking devastating injuries we've been having.
Everyone has injuries mate. Literally every clubs fans are complaining about it constantly.
Did you see how many injuries Spurs had from start of November until just recently? That’s why we dropped off from top of the league.
Spurs had soooo many injuries in November too, almost all teams do
lol. Bros talking like other teams don’t have injuries 💀. Look at spurs and United. And even my club rn Liverpool
Can’t blame injuries everyone has them…
50% of this video is just James saying "could they" 🤣
saying this now, if Newcastle didn't have the absolutely ridiculous record breaking injuries this season, the champions league group would have gone significantly differently, and in the league I think they would have been around where villa are currently, sadly as we know there hasn't really been a solid fit 11 at the club since the start of November last year, it's been a few of the same players in certain positions but 95% of the entire squad has missed games at some point through injury,whether that's just a couple of games or full on 6 month jobs, it's actually ridiculous that the media are pushing this nonsense that eddie howe is under any sort of pressure when you look at the state of the playing staff he has had to work with this season, and don't get started on the sandro tonali issue, all in all I think the club are doing phenomenal all things considered, just a shame that tripper screwed the pooch and basically gave the EFL cup away.
Villa are in an injury crisis, spurs had no cbs for weeks, united haven't had a full squad all season. Newcastle just aren't that good, not to say howe should be under pressure but most teams have faced the same issues at some point
@@louies7868 here's a question for you, out of the clubs you have tried to claim have had anywhere near the levels of injuries of Newcastle, how many of them had enough fit players to still be able to make a substitution in a match? I stated quite clearly and on purpose that Newcastle's injury situation this season was record breaking because only one other team in premier League history has had a worse injury situation than Newcastle and it was all the way back in the late 90's.
Interesting video, however I doubt the human factor has been truly calculated in. City have overcome big points differences late in the season on numerous occasions. Liverpool and Arsenal have been psychologically damaged by that. The other issue is form, can all 3 clubs maintain their levels up to and including week 38? I doubt it. Difficulty levels of remaining fixtures is a factor but 3 points is 3 points and the better team CAN beat everyone. I agree with the swing value of city’s Anfield fixture.
Klopp leaving on a pl title won by a single point would be beautiful and very much deserved
For who
for football
@@sandg507
@@sandg507 lfc fans and him
@@SopranoLFC lfc fans are terrible. Arsenal deserve it way more.
@@lars3712 why do they?
Exceptional
The fulham european fact reminded me that Ipswich have also not lost a European match at home, and i think they've played more than Fulham
Does the model consider potential injuries?
4000 simulations - Lets see if it predicted the one that actually happens
The two biggest games of the season remaining are Liverpool VS City and City vs Arsenal City have to win both.
Based on average points, Liverpool has a 1.96 and 1.45 point advantage over Arsenal and City respectively. Meaning there's only a 0.51 average point difference between City and Arsenal. BUT, in true Arsenal disrespect fashion, you group Liverpool and City together at the top, "and then Arsenal get third." In reality, according to the average points, Pool is the clear winner, and City and Arsenal battle for 2nd.
Remember, for the price of nothing, you can sponsor an Arsenal fan of your very own, and bring some joy to the hearts of fans.
#EndArsenalHate #RespectTheCannon #Gooner4Life
This is brilliant. Fascinating. Genuinely.
However. Something that’s perplexing about these findings. Specifically relating to Arsenal & City.
Looking at these figures, City apparently have a better chance of finishing 1st than Arsenal, despite being a position below them currently. Meanhwile, at the same time, Arsenal have a lower chance of finishing top 4 (2nd-4th) than City.
How can these both be true at the same time? They contradict each other. One suggests City are more likely to finish above Arsenal… while the other, at the same time, states Arsenal are more likely to finish above City (just lower down the table).
City just dropped 2 points.Thank You City.❤
Was this filmed before United’s run? Feels hard to believe considering United’s form, Villa’s form, and the recent head to head matchup.
I don't think a simulation is going to take perceived or actual form into account.
They've still been awful
So you think Utd will get 6 more points than spurs or villa before the end of the season??
@lms_steve6693 yes. They play each other and both are falling off. Pressure will he applied and United will shithouse their way to 72 points and a 4th place finish.
Scousers, Gooners, Man City with the Brummies and Man U to fight it out for fourth with Everton and Forest surviving the drop are my predictions.
Another crucial game would Man City vs Arsenal game on 31 March.
How many times in your predictions did Man City win/draw/loose against Chelsea? It will give a good idea of how good the model is and if you want to rerun it?
Dedication in full
Can you do this again please? Alot has changed in 11 days 😅
something is off with this model. In fact, believe it or not, Arsenal should be favourites to win the title. They have the current highest expected points and best xG and xGA differential. City are close and Liverpool are some points behind. So considering the remaining opponent difficulty it should be Arsenal then without about 40% chance, City at 35% and pool at 25%
The could they bit it great 😂
When you say simulated the remaining games of the season,simulated them using what?
Would be interested to know if this weekend's results changed things at all?
Arsenal keep getting stronger while Man C get weaker...
I see that this prediction concerning the CL places was before Spurs lost at home to wolves?
Man City already dropped points
Of course you want us relegated 😂
Why does he always flirt with us with those "could they" moments
Do this every gameweek..
Arsenal fan here - man, losing the title to a goal differential of 1 would be rough (with Aston Villa erasing the current 13 GD deficit); I've been thinking about the probabilities lately, and in my head it's something like 48% Man City, 30% Liverpool, 20% Arsenal, so maybe giving Man City too much credit. But it seems foolish to expect anything other than another title for Man City. Ah well, COYG!
Loved your content for years James but a little less “could they” might make it a bit more enjoyable and unique
A bit like End Game it only takes one
City should be 2nd in your probability table. 81.34>80.83
2%
So you're saying there's a chance😋
"The gap is getting wider" because the three promoted teams are most likely to go down and yet LAST season all three stayed up so...
The three relegated teams: Leicester, Leeds and Southampton are also in a good position to come back up...
In all seriousness though..............could they?
Shows how hard Burnley have had it compared to Luton etc
Damn... city dropped points at chelsea
No, City dropped points to Chelsea at home
Would be nice to see how the Liverpool v City result affects Arsenal's chances especially in case of a draw
Doesn't matter. How much ever all these youtubers may want you to believe that arsenal is in the title race they are not. It's always been a 2 horse race. Except last year where it was a 1 horse race
@@popsiclemonster858 why? Level on goal difference with City
Hitting form at the right time
Players that will make us stronger about to return
I mean they are level due to a statistical anomaly in one game where they scored 6. Arsenal have struggled with scoring this season up to this point.@@MrUltrAdaman
@@popsiclemonster858it doesn't matter that you think it doesn't matter. A draw between Liv and city will affect Arsenal's chance, despite what muppets like you say. It would raise their chance, just not to above 50% so by end of season simpletons like you will think they were right
@@MrUltrAdaman 200m quid spent and your knocked out of both cup competitions in the 3rd round and 7 points worse off than last season.
But all you have left to hold onto is you're level on goal difference with City and the clown Zinny the Fool, Thomas Part timer, Jesus Offside, Flimsy Fabio, Tomitinfoil Jurrien Deadwood, Made of Glass ESR and Cedric the Clown all returning from injury loooooooooool
@MrUltrAdaman is a dry lunch
Man City and Liverpool draw, and Arsenal win away to Man City, and the Gunners win the title. If the MCvsL match is not a draw, the winner wins the title. All three teams are in great form and only fatigue/injuries can stop them winning the other games.
Is the simulation FM football manager? 😂
Disappointed Manchester City aren’t dominant enough to win 100% on simulations
Seems to give a lot of weight to recent form this
And after Chelsea drae at the Etihad lol
I simulated your channel once.
James Lawrence Allcott wich 3 teams do you think will go down from Premier League?
Does this sim take into account the Everton and Forest points deductions and appeals? I feel like relegation isn't decided until those get sorted.
'easier' games based on what?
JLA take a hit before recording?👀
Everton have been f*ked in the arse this season I hope they survive
Personally, I think they could
I’m a simple man. I see new JLA posts, I press it & hit like
The could they bit has scarred me
And won’t be watching again
*could they?*
Well, Arsenal definitely not winning it now. It was fun, lads.
Run in analysis at Soccerstats shows Forest has the second easiest schedule.. 1.35 points for opponents. 1.53 pts for Luton so what gives?
Run in analysis at Soccerstats????
Ffs lol
This made just before spurs lost to wolves haha
6:55 we would not win the leugle it will be liverpool man city arsenal aston then man u totteham
No but the real question is, could they?
Your simulation… does it take to account european games? Just head to head games cannot be a good basis for this sort of simulations(?)
Villa won’t get top 7 now. Not with these injuries to all of their defenders and most importantly Kamara. Gonna have to play Callum Chambers 😭
Haha. Nah,villa will be 4th. They are back at 4th. They are a fighter.
Pau was a precaution and Konsa is back in training
"you've learnt something. hit the like button thankyou."
AI ???. How about on Fm?
The key question is... Could they?
Proof that AI is far from being intelligent
so if liverpool don’t win the league, have they statistically bottled it?
No one will give them grief like Arsenal got though
Because there’s a difference between being at the top and dropping to second and being far far ahead at the top and then dropping the ball completely at the end of the season
@@mirkys130 5 poits is not far far ahead, lets not move the goalposts now your favourite team is in the run in lmao
Well the model gives a probability of less then 50% of winning the league, which means they are more likely not to win it. Don’t see how that can then be considered a bottlejob:)
If we’re being fair, at this stage there can’t be any bottlejobs. The simulations are ultimately just a bit of fun anyways. At the current stage, and seeing the sort of form City, Liverpool and Arsenal are in, I’d give City a 50% chance at winning it, and Liverpool and Arsenal about a 25% chance each.
Wow wrong 4,000 times
all this for man city to win it on the last day in 117th minute
This simulation's biggest flaw is not factoring Arsenal's annual 🍼 in the last 10 games of the season.
😐😐😐
From banter to bottle. I guess we are in the right trajectory to go top. Arsenal will be champion this or next season. If bottle both, top 4 mediocrity with players leaving left and right.
Man City have hard games such as Liverpool, while Liverpool have easy games.... like Man City.🙄
Spurs not in the title race... except. Their last 5 games of the season, they play city, arsenal, and Liverpool. Right now they are only 7 points behind the top. If they pull a masterclass and win all three of those games, it's not infeasible they will be up there, so the "could they?" question... Its not beyond the realms of possibility
😂😂😂
Dodgy haircut?
Dial all villas injuries into the computer then see what it says lol
@@Blacknig69 no... where did i say that? but we JUST lost Kamara to an ACL for the rest of the season, and Diego Carlos to a hamstring for at least 2 months. On top of not yet having Digne or Torres back in a game, ramsey and moreno still lacking after injury and mings and buendia not coming back till next season. 3 ACL injuries in one season. SO plug that into your simulation and smoke it. Taking those fresh injuries into account, would likely drastically affect how high they are predicting villa to finish.
I’ll kick your ass if Manchester City wins the league 😂
LMFAO ... thst didn't age well.
Seriously don’t you have anything better to do with your time ??? I can tell you now Liverpool will win (with no simulations).