I missed a timing mat once. I swerved to the side of the road to throw away an empty gel pack and didn't notice the mat until I was past it. I also once cut a corner (by like 2 feet) because I was distracted by another runner who had blood running down his face. Lots of things happen over 42.2 km. You have to be very aware at all times.
Unlikely that would flag you as potential cheater. People look for unusual splits or times as a flag so if your pace was relatively constant or fell throughout known time mats that's not worth flagging even if you missed one. It's only if you miss quite a few and your pace is unusually high between the mats then it becomes suspicious. That's when you look for other information usually race photos to see if they were actually on the course and also look for people who were near them at various points as people will generally run even pace so finish near others who were near them in photos. Cutting a corner by a few feet is cheating but it's inconsequential and only makes a few minutes difference if that which is just within random noise. Course cutting is blatantly cutting multiple miles usually on out and back sections and gives unusual splits.
I got second place for a small town race last year and was looking at the splits and pictures of the lady who got first. She was wearing a sweatshirt the whole race in 80 f degree temps. Turns out she finished the race early (not doing all laps) -- I blame the timing company for not catching this. After I complained, I then received 1st place and the ladies below me must have felt good getting a better place. Thank you for this interesting study!
I have done similar analytics, but the most interesting data I have found was taking individual Boston marathon finishing times and comparing those times to their qualifying times, then breaking them down by age group and country of origin. This yielded VERY interesting results with some countries having 100% of an age group running faster than the qualifying standard at Boston and other countries of similar sample size having less than 50% beat the standard in the Boston race.If you then isolate the runners that finished more than 1 hour slower than the qualifying time and research their qualification race and method of verification it becomes pretty clear the scale of cheating going on in certain countries. I am pretty sure that if I can figure this out Boston has too! Need to change the race verification process in the future. Its a shame many actual runners are having their Boston slots stolen by cheaters!
Actual results for "time qualifying" races crack me up. NY is supposed to be like 3:10 qualifying yet only like 10% of the finishers finish under 3:10 .Hilarious how many illegiimate runners are in these races...whether hiring mules or running downhill courses or whatnot. To me its pretty meaningless getting a "qualifying time" Ho hum. Only thing I care about is how I'm performing vs my prior results and expected fitness level. Besides I don't run in the cheat shoes which are worth like 10-30 minutes lol. So ridicuious what is legal in running. Literally a spring plate in the shoe is legal. Most of these runners would be 20 minutes slower if forced to run barefoot on grass without their cheat shoes to spring them along. Reminds me of the cheat swimsuits from 15 years ago in swimming.
As a coach, this makes me so angry. My athletes work so hard to get stronger and faster while others just cheat their way through what is meant to be a life changing experience.
make sure you coach purpose, too - it's why most speedrunners for video games celebrate their PBs, with a WR being a secondary accomplishment. you can't ever take away your own personal best by cheating.
There was a guy who disappeared behind a portapotty and later we found out he descended down a manhole and followed a narrow sewage pipe a few hundred meters only reemerge behind another portapotty cutting 5 miles off the course.
@@dimitar297 That's determination. I condone every cheating but I have to admit that I have some respect for getting and walking though a narrow, smelly sewage pipe..
shoutout to all the WALL runners!! we see you (i am you) and we dont give up!!! we finishing that race no matter how bad them knees hurt or how bad your head says “yeah i dont want to do this at all”
An "average" runner is a 3:19 finish???? Also to say "back of the pack" didn't train is not true. Runners in the 5:30-6:30 finish times DO train. Often times much more challenging than a 3 hour marathoner - they are out there for training runs and races twice as long. I loved the analytics, but be fair to the other runners who aren't a 7:30 mile.
Thank you for this comment. I was a 5:30 finisher. Yes I slowed down at the end but I ran a consistent pace throughout the first 20 miles and didn’t hit a wall. I trained very consistently all spring/summer/fall. Kinda annoying to hear him say someone like me didn’t train 😒
I am not a big runner, and never ran more than 15k in one go, and I really don't understand why elite runners keep a certain pace (for example if there's a bridge why they use up more energy running up to keep the pace but then not running faster on the way down) but I don't think he meant to hurt anyone by calling a category whatever he called it. The point of the video is to show that looking at the data one can put runners in some categories based on the data pattern and if one particular runner's data looks like they belong to a category, except that in a few data points they are outliers, then it's worth to have a deeper look. Besides if you run and are not cheating then the name of the category you find yourself in is really not what should bother you after seeing this video. BTW I saw a woman a few months ago on youtube. She said that by the time she finished usually the organizers already take down some of the infrastructure. That really hurts to hear! Though it wasn't clear if those races had an official cut off time (or whatever it's called. In some races the organizers post in advance what time the runners have to beat at certain points and if they don't arrive on time then they can't/ shouldn't continue the race) so it might be that she wasn't aware of that or that she was and that's exactly the point she was making, that casual runners should be given more time, cause being slow doesn't meant they can't finish the distance.
I fell at the 38km and I was bleeding and all,but I kept on running. I can't believe people actually go that far just to cheat. This trip was not cheap for me.
Excellent research Willy. Enjoyed watching that one. We recently had an ultramarathon in the UK where a participant got a ride in a car. Slightly different but I’m sure it would be in the data.
They think anyone cares that they "qualified" for Boston. I'm more impressed by times/courses than "qualifiers". And its pretty obvious if a "friend"/acquaintance is lying about a time. Hint: If he's 200 pounds and has a dad bod and claims he ran sub-3:00, he lyin.
I’m sure it’s common in Chicago because of the course. I just ran it. I saw at least one runner come from way off the course onto the course and start running. He wasn’t coming from a porta potty.
at 14:17 you say the guy without the pictures is the cheater and the guy with the pictures is the bib mule, but you need to emphasise that he is ALSO a cheater
The NY marathon once rounded up suspected cheaters based on the time data, and sent letters (I believe anonymously) asking why they did it, and it turned out the suspected cheaters were often middle-aged men who felt life had cheated them. At least that is what I read in an old running magazine.
I always have my bib up at front, I almost lost a bib at the Berlin marathon, but I manage to pin it as I ran. They tell you at front not to cover your bib.
Great video. One thing I'd suggest for races is to include some metadata on all photos of a person (such as time and distance) of a photo. This would increase the number of data points we have for a runner from the 8 timing mats to also include any photos. ML models could potentially classify cheating risk categories for each runner (low, medium, high, & very high). For instance, someone who finished in 5:30 with consistent pacing throughout entire race and have photos of them on the course, but they missed one timing mat might be classified as medium risk. Someone who finished right under a BQ time for their age group + have extreme discrepancies in pacing to suggest course cutting would be classified as very high. One thing to keep in mind is just because a model flags someone does not mean they definitively cheated like you mentioned in the video. The goal would be to flag individuals, prioritize review, and then investigate. No model is going to have 100% precision. Having a situation where we could have a high recall + a manageable number of runners to review is best case scenario for the WMMs.
Yeah no I'm not signing up for a race that wants to publish regular pictures of me for their AI model. Get your priorities straight. That's so unhinged. The damage caused by an average person (non-elite/pro) cheating in a marathon is *not* enough to justify this.
Yep, maybe also there should be a whistleblower email address, where people who have this kind of data could send the data to the organizers and trigger an official inquiry. Maybe it wouldn't help others who finished after the cheater in that race, but disqualifying them from other future races does have at least some positive effect.
When looking at the data I was trying to find out why all those runners missed the timing mats. I looked into some of the runners who did missing timing mats. Looking at that specific runner who had all her timing splits missing was my starting point then looking at other runners pictures who had missing timing mats I noticed they shared similiar qualities. Wearing heavy clothing.
Some races have people wear a bib number AND a timing chip on their shoe/ankle, but I believe as of late the chip is attached to the back of the bib. Open question as to how well chips will work if under clothing, but a general takeaway should be that "missing a timing mat" could be due to covering clothing or a faulty chip.
I'm a bit perplexed by this. Most race bibs have a sensor/RFID chip affixed to the back. Berlin marathon, for one, still uses a timing chip that's attached to the shoe. In Chicago this year, there was a warning to have your bib visible and not covered, but I figured that was more related to photography and perhaps visual identification for fast runners/age group winners. In a big race like Chicago, where you're going to get more recreational runners, they may not know to have bibs visible. Also, a couple timing mats seemed kind of narrow this year, and it's possible that someone could inadvertently run around it. If heavy outer clothing interferes the bib RFID picking up the electromagnetic wave from the timing mats that energizes the runner's RFID chip, this is something that should be more heavily emphasized at races where cooler weather is anticipated. Great research!
@colinmcc8564 definitely missing timing mats could be due to faulty chips, person did not go over matt, and covered clothing. My premise why I put clothing as my reasoning was due to the fact after the first missing timing mats the runners times start getting registered. This data set is from 2018 Chicago Marathon. I am curious if I could find similiar results again.
@@3BrosRunning It would be interesting to apply the same type of analysis on Chicago 2021 because it was hot that year. I would not expect many runners to be wearing heavy clothing. If there were little or no runners who missed timing mats, then either less people cheated that year or the chips worked properly. I would lean towards the chips working properly given the no heavy clothing.
I am no Olympic runner but I do try hard in races. I had a 1:44 qualifying time for a race (that shall remain nameless) and got placed in Corral D for the start. The corrals only went from A to F so I was a bit surprised at starting so far back. The woman starting next to me had a 1:30 qualifying time and she was mad as hell at being in Corral D. Just looking at the people in front of us we knew that cheating must have been rife in the qualifiers for this event. There are certain physiques you can tell are never going to run a sub-2-hour half marathon (You know what they are). Sure enough, once the race started about 95% of the runners in the corrals ahead of us were never going to run a half in under 2:30 they jog-walked from the start line. We both ended up walking and jogging as we tried to zig-zag through the crowds. A frustrating and ruined event for me. I only got completely clear of the crowds at about the 8-mile mark. The only bright side for me was not only did I get a negative split time it was the first and only time I have ever finished a half marathon with a sub 20min 5k for the last 5k. It would be interesting to see how that would show up on your graph. 🤔 Event organizers everywhere need your program for every event they run to stop the things you highlight in this video. Cheating can ruin running for other people in an event.
I would guess that most of the slower runners in front of you in the faster corrals just cut into a faster corral and covered their bib, yes? At the Philly Half Marathon, they didn't check bib colors that carefully, and I remember mid-race passing a bunch of runners who were supposed to have started in a slower corral than mine.
@@mmj1342The event is very thorough about checking bibs and keeping people fenced in the right place. Very strict marshalls on every fence line. No corral jumping, unless you want to jump backward which is allowed. I think all the cheating went on in the provision of evidence to get the corral position. I have recently been to organized half marathon events with less than 50 runners where I know no women finished in front of me. But when the results were published there was a Susan who finished just in front of me and wasn't even the female winner! The guy in front of me didn't look like a Susan. Possibly hard to say these days though. If there is no prize for the event I don't think anybody checks.
I’m incredibly slow right now and have had injuries for over a year. It inspires me to train and get quicker. It means nothing to brag about something you didn’t earn. Very insecure people.
Thanks for doing this video…. Very interesting! I never dreamed that a person would cheat…. How can they sleep at night? Thanks again! Run on 🏃♀️ 🏃♂️
Could be a bib swap also happen accidently in groups, mostly the name stands at the number but i don't know if this is always the standard. i didnt run a lot races, but in all of them i registered with colleges and friends, and one of them is getting the numbers before the race and then they give it out to the group at raceday.
US Roadrunning has 5k to Half Marathon times - to help people train and fartlek is a good method and I use it, and that’s where I’m doing Chicago and a 5k leg of Denver in relay. My scratched super sprint triathlon entry had a great time for a 1.2k run for a lot of people when I ran 5 and biked over limit. I need to get my stopwatch right next time.
Not sure about Chicago, but when I did London there were 3 different start lines that converge after the 5km mark. It’s possible that someone could miss a mat or a particular mat on one of those starts could have failed.
I have some theories why people would cheat. 1) A semi-elite (close, but not good enough) trying to steal a win or Top 3 finish in the race or their age group. 2) Someone trying to fake running a race and finishing or getting a personal best for social media status. 3) Someone trying to get to Boston or another marathon requiring qualifying. 4) Someone trying it that is not properly trained for it or is out of shape or maybe suffered an injury. They may try to take a shortcut to get to the finish line and know that they aren't going to come even close to winning any type of awards anyway. I haven't ran a mini-marathon yet (first one at the end of this month) but ran many 5k and 10k's and one 15k. I saw a time of like 22:28 in one 10k and that was the only bogus time that I saw. It was a high school age boy on a Thanksgiving run and it was two laps. I think that he got confused and ran just one lap. He may have had an emergency or family need and needed to quit early too. He did get credit for winning the race at the awards ceremony (but was not present to receive the award), but he got DQ'ed in the final standings later when I checked that afternoon. Thankfully running has very few cheaters and only a handful of those cases affect age group or overall winners or major marathon qualifying.
Why hasn't a machine learning model been built to immediately issue a DQ for various classifications of cheaters? Corrections could be made for those cases with equipment error.
Something I was working on. To make an effective machine learning model I would need more data. I was able to scrape just one year of data but I think I would need 10 years or more to create an effective machine learning model. I have an algorithm I created based on certain parameters. I may release that video soon.
@@3BrosRunning Great video. In my opinion, I think 40,000 runners should be more than enough to create a decent model (though perhaps it would be good to have representation from other races if you really want it use it). Would love to see another video on how you went about it. Seems like a perfect use case for anomaly detection.
A lot of popular races have more demand than places. I've been unable to run before and given my place to someone else. They ran a much faster time than I could, but I'd never try and claim that time for myself. They got to run and enjoy the day and my entry fee wasn't wasted.
It's really about insuring that runners get shirts, medals, foods, etc... we can only estimate how many runners we will have before timelines for shirts and medals come close to the deadline. How to prevent this? sign up early and for our races we will normally let people transfer but for bigger races I can see the issue it can cause.
I don't get how having a sweatshirt or jacket over your bib would keep them from triggering the timer. Isn't it an RF device? Maybe it was just a defective device. But I still appreciate calling out the cheaters... mostly because their "success" cheapens the accomplishments of us honest runners.
To be fair running in Mexico City with the high altitude and poor air quality is absolute hell. I was coughing up blood after a run when I visited there
Super interesting video! Would love to see more of this! One question I have: How did you find the bib-swapper with the different name/sex? The only thing I can come up with is recognizing all photos of all runners, or looking at previous results for all runners (or just sheer luck), but all of these are not easy/cheap
As a runner myself I kind of know how people have cheated or the possible ways they could cheat. The photo one I feel with current technology is possible using facial recognition tech. I pretty much just scoured the images till I found one to use as an example. For my actual presentation I shared the name but for this I don't want to call anyone specific out. The bib mule one was one I knew existed and was super happy I was able to find some examples of it in the data and was able to showcase a "possible" cheater. I applied an algorithm to the Houston Marathon and wasn't able to find a bib mule.
I didn't miss timing mats but unfortunately did not get pictures of my race because my bib number was part covered by my running belt, so the AI things they use to generate your photos based on bib number didn't work for me. I was gutted haha but I'm sure I looked awful in the photos anyway....
Interesting video, thankfully it seems like the percentage of cheaters was very low. Seems like you could work with these bigger marathons to help them with cheat detection.
- when finally completed, would you provide the code / app / software to Others in order to enable Others to check the data of their completed races ? ( I once found the winner of an age-group in a marathon beeing a wheelchair-racer :-) ) - how do you receive and extract the full data efficiently / automatically from the results-webpage ?
What happens if you cut the course without missing any timing mats and wait for your average pace speed before carrying on to save energy for a fast finish.
I think if u do that it would take more work to cheat then it would be to just just run the race normally. There is this instance of a person who did something similiar during a trail race. They would hide in a Porta potty then run when it was appropriate and keep doing that till the finish. It made his times look legible until he was caught. Person used this technique to place in trail races.
Interesting, thanks. was also wondering if you could discern what percentage of people were "probably" cheating. Like others below - and as an engineer - not quite understanding how chips don't work through clothes.
Last year my running group's trainer told us that he has free bibs on a race. So i eent there. There were no free bibs, so i was very pissed, but i drove 40 minutes (and woke up 2 hours before i would've not for the race) so i decided to pay on the spot (highest price of course) At the end if the race i almost left for home, when i saw lots of people around some billboard. It turned out that they printed the results and put it on the wall. I managed to find myself and i couldn't believe my eyes! I was 3rd in my age group. What a surprise! This was my best 5k ever! Not only because of the real trophy 🏆 but it was also my PB. This is such a good feeling! I think it would be my best race even if i only had the PB, but it would be a shame not to know that i finished in the 3rd place, not to have a picture of me standing on the podium. That's why it's important to have tools that can detect cheaters in real time, so they can be disqualified before the results are announced. I heard that in the past there were cases when cheaters were caught and thus someone was awarded a medal (months) later, but those people who really deserved those medals were robbed the feeling and the picture they deserved
Missing every mat probably means they have a broken chip or something (or something covering it, like you found). More likely, some of the runners that missed two or three, but not all, were cheating. This is interesting. I would like to read your capstone or take a look at your notebook to see the code.
how do you differentiate between a cheater for the sake of competition and a cheater for the sake of they just don't care anymore and just want to be done.
I have to be satisfied all you have to do is walk this marathon for any medal, so not coming in on any timing mats at all is an issue that’s easy to fix
Great stuff, thanks. Sure, someone can miss a matt. It happens. But that and cheating are two very different things. I don not get why would someone display a medal they have not earned.
Fascinating! A great explanation and very "human" analysis. I only just run my second race at well over 50. The first with my adult daughter. We went through the finish together with essentially the same splits holding hand in the official photo finish. Yet, I can understand all of the that (from the warnings to not cover the bib) and to that short cut (in a 4 loop race, I looked at the finish and thought: why am I doing another loop , when there is a shortcut to the water and food right over there). Fortunately at my age and low level, there is no pressure whatsoever :-) I am looking forward to the 3rd, but paces might be slower as I am starting to fatigue ;-) I loved the cross though and just got a thrill from running reps alone on the tracks and being much faster than the VDOT tables (not sure that's so great, but I'll look into it)
Chicago's race configuration (essentially three out and backs in three different directions) is really ideal for those who want to cut the course. But you better know where the timing mats are (but apparently, ya can't check the results of all 47,000 runners).
Cracks me up the stupidity of people. Like anyone cares what a non-elite runner's time is or whether he finished a marathon. Big whoopie whether you did it in 4 or 6 hours. And then paying $200 to cut the course or cheat is even dumber. Personally I'm so anal I was running a 5K where the course had eroded away some brush from previous years and I still ran the longer distance of probably an extra 100 feet because I didn't want to have a short course time. I want to be able to compare to my previous times on the course as well and in general. I'm sure I was the only one that did that in that race. I also find it hilarious the amount of money people spend on these races to walk and jog them. Like you can't just go out on a saturday on your own and walk or jog a marathon without paying someone $200. I'll never get that. I've done that. Ran a marathon on my own and I do a lot of longer trail runs even ultras on my own. The only time I enter a race is when I will be pushing myself to my own limits and want to be pushed by others, otherwise I will just run the distance on my own. I always thought a mule was someone that ran using someone else's name/bib not someone that had 2 bibs (his own plus the person's). I would be down to run for someone as a mule. I think it's stupid that someone would pay all that money and have someone else run for him, but if someone is dumb enough to pay me to race, I'd do it.
I realise you’re only looking at instances of cheating in marathons. I’m sure you’re aware that cheating occurs in all races. Personally, I have been cheated out of a first place finish on three separate occasions in shorter races because of these despicable acts.
2019 i ran the marine corp marathon in heavy rain. I seen so many women get on rented ebikes and speed up the mat timing get off and go through find another bike. Insane bc the tracking that day wasnt working. The app tracker at least didn't work. I feel good i did the entire course, but it was an incredibly hard day. Insane amount of course cutting was going on
I’m going to start sprinting mid race to mess with data scientists.
I missed a timing mat once. I swerved to the side of the road to throw away an empty gel pack and didn't notice the mat until I was past it. I also once cut a corner (by like 2 feet) because I was distracted by another runner who had blood running down his face. Lots of things happen over 42.2 km. You have to be very aware at all times.
You dirty, dirty cheater
Unlikely that would flag you as potential cheater. People look for unusual splits or times as a flag so if your pace was relatively constant or fell throughout known time mats that's not worth flagging even if you missed one. It's only if you miss quite a few and your pace is unusually high between the mats then it becomes suspicious. That's when you look for other information usually race photos to see if they were actually on the course and also look for people who were near them at various points as people will generally run even pace so finish near others who were near them in photos.
Cutting a corner by a few feet is cheating but it's inconsequential and only makes a few minutes difference if that which is just within random noise. Course cutting is blatantly cutting multiple miles usually on out and back sections and gives unusual splits.
This is like how I envision my 7-year-old son talking to the priest during reconciliation.
It is the responsibility for every runner to hit the timing mat. No excuses.
@@mikepeligrodid making this comment make you feel better?
I got second place for a small town race last year and was looking at the splits and pictures of the lady who got first. She was wearing a sweatshirt the whole race in 80 f degree temps. Turns out she finished the race early (not doing all laps) -- I blame the timing company for not catching this. After I complained, I then received 1st place and the ladies below me must have felt good getting a better place. Thank you for this interesting study!
I love that he call someone running a 6min mile in a marathon an “average runner” 😂😢
3:19 is 7:35 per mile, wym
That's an elite pace.
😂I thought the same. I guess average is relative.
6min kilometer not mile. 6min km would average out as a 4 hour 13 min marathon - definitely not average but I understand what he means
2:39 the stat posted is incorrect as that was 11% of people who intended to do a future marathon stopped running altogether, not the marathon DNF rate
I have done similar analytics, but the most interesting data I have found was taking individual Boston marathon finishing times and comparing those times to their qualifying times, then breaking them down by age group and country of origin. This yielded VERY interesting results with some countries having 100% of an age group running faster than the qualifying standard at Boston and other countries of similar sample size having less than 50% beat the standard in the Boston race.If you then isolate the runners that finished more than 1 hour slower than the qualifying time and research their qualification race and method of verification it becomes pretty clear the scale of cheating going on in certain countries. I am pretty sure that if I can figure this out Boston has too! Need to change the race verification process in the future. Its a shame many actual runners are having their Boston slots stolen by cheaters!
True although boston marathon is not exactly flat, however 1hour slower definetly needs good reasoning.
Actual results for "time qualifying" races crack me up. NY is supposed to be like 3:10 qualifying yet only like 10% of the finishers finish under 3:10 .Hilarious how many illegiimate runners are in these races...whether hiring mules or running downhill courses or whatnot. To me its pretty meaningless getting a "qualifying time" Ho hum. Only thing I care about is how I'm performing vs my prior results and expected fitness level. Besides I don't run in the cheat shoes which are worth like 10-30 minutes lol. So ridicuious what is legal in running. Literally a spring plate in the shoe is legal. Most of these runners would be 20 minutes slower if forced to run barefoot on grass without their cheat shoes to spring them along. Reminds me of the cheat swimsuits from 15 years ago in swimming.
You mean China, right?
As a coach, this makes me so angry. My athletes work so hard to get stronger and faster while others just cheat their way through what is meant to be a life changing experience.
make sure you coach purpose, too - it's why most speedrunners for video games celebrate their PBs, with a WR being a secondary accomplishment. you can't ever take away your own personal best by cheating.
Ultimately, the ones who cheat in marathons also cheat in other areas of their life too. A poor excuse for living. ... Keep up the good work, Willie.
There was a guy who disappeared behind a portapotty and later we found out he descended down a manhole and followed a narrow sewage pipe a few hundred meters only reemerge behind another portapotty cutting 5 miles off the course.
@@dimitar297 That's determination. I condone every cheating but I have to admit that I have some respect for getting and walking though a narrow, smelly sewage pipe..
shoutout to all the WALL runners!! we see you (i am you) and we dont give up!!! we finishing that race no matter how bad them knees hurt or how bad your head says “yeah i dont want to do this at all”
Hahaha as soon as he talked about the wall I was like “yep been there”
An "average" runner is a 3:19 finish???? Also to say "back of the pack" didn't train is not true. Runners in the 5:30-6:30 finish times DO train. Often times much more challenging than a 3 hour marathoner - they are out there for training runs and races twice as long. I loved the analytics, but be fair to the other runners who aren't a 7:30 mile.
Thank you for this comment. I was a 5:30 finisher. Yes I slowed down at the end but I ran a consistent pace throughout the first 20 miles and didn’t hit a wall. I trained very consistently all spring/summer/fall. Kinda annoying to hear him say someone like me didn’t train 😒
I am not a big runner, and never ran more than 15k in one go, and I really don't understand why elite runners keep a certain pace (for example if there's a bridge why they use up more energy running up to keep the pace but then not running faster on the way down) but I don't think he meant to hurt anyone by calling a category whatever he called it. The point of the video is to show that looking at the data one can put runners in some categories based on the data pattern and if one particular runner's data looks like they belong to a category, except that in a few data points they are outliers, then it's worth to have a deeper look.
Besides if you run and are not cheating then the name of the category you find yourself in is really not what should bother you after seeing this video.
BTW I saw a woman a few months ago on youtube. She said that by the time she finished usually the organizers already take down some of the infrastructure. That really hurts to hear! Though it wasn't clear if those races had an official cut off time (or whatever it's called. In some races the organizers post in advance what time the runners have to beat at certain points and if they don't arrive on time then they can't/ shouldn't continue the race) so it might be that she wasn't aware of that or that she was and that's exactly the point she was making, that casual runners should be given more time, cause being slow doesn't meant they can't finish the distance.
I fell at the 38km and I was bleeding and all,but I kept on running. I can't believe people actually go that far just to cheat. This trip was not cheap for me.
Are you the one that distracted the top comment guy?
Not easy or cheap, us from Houston working class,💯
tableau says "3:19 is avarage runner" makes me feel bad 😅
Per KM not per Mile!
he means the marathon end time@@Tomcat304
@@Tomcat304 3h 19 was the finish time for that line. i cna only dream of goign that fast
I had to train like crazy to get a 1:45 half marathon time. Can't imagine what I'd have to do to hit that for a marathon.
Excellent research Willy. Enjoyed watching that one. We recently had an ultramarathon in the UK where a participant got a ride in a car. Slightly different but I’m sure it would be in the data.
Thank you. I may go over my cheating algorithm in a video.
As a runner and a software developer myself, I really enjoyed this video
Thank you for doing this! It makes honest runners like myself feel better.
You're so welcome!
Classic example of using data analytics/Tableau skills along with your running SME to solve real issues. Love it.
Why would you cheat just so you can be the slowest runner in your age/gender category at Boston?
They think anyone cares that they "qualified" for Boston. I'm more impressed by times/courses than "qualifiers". And its pretty obvious if a "friend"/acquaintance is lying about a time. Hint: If he's 200 pounds and has a dad bod and claims he ran sub-3:00, he lyin.
I’m sure it’s common in Chicago because of the course. I just ran it. I saw at least one runner come from way off the course onto the course and start running. He wasn’t coming from a porta potty.
at 14:17 you say the guy without the pictures is the cheater and the guy with the pictures is the bib mule, but you need to emphasise that he is ALSO a cheater
Here in the Philippines, our timing sensor chips are tied at the shoelaces to avoid the clothing covering problem
can a person swap shoelaces ? asking for a friend
Very interesting to see how some people cheat (themselves)!
The NY marathon once rounded up suspected cheaters based on the time data, and sent letters (I believe anonymously) asking why they did it, and it turned out the suspected cheaters were often middle-aged men who felt life had cheated them. At least that is what I read in an old running magazine.
I always have my bib up at front, I almost lost a bib at the Berlin marathon, but I manage to pin it as I ran. They tell you at front not to cover your bib.
i am pretty sure the 15,000 people who finished in front of me were cheating
Great video. One thing I'd suggest for races is to include some metadata on all photos of a person (such as time and distance) of a photo. This would increase the number of data points we have for a runner from the 8 timing mats to also include any photos.
ML models could potentially classify cheating risk categories for each runner (low, medium, high, & very high). For instance, someone who finished in 5:30 with consistent pacing throughout entire race and have photos of them on the course, but they missed one timing mat might be classified as medium risk. Someone who finished right under a BQ time for their age group + have extreme discrepancies in pacing to suggest course cutting would be classified as very high.
One thing to keep in mind is just because a model flags someone does not mean they definitively cheated like you mentioned in the video. The goal would be to flag individuals, prioritize review, and then investigate. No model is going to have 100% precision. Having a situation where we could have a high recall + a manageable number of runners to review is best case scenario for the WMMs.
Yeah no I'm not signing up for a race that wants to publish regular pictures of me for their AI model. Get your priorities straight. That's so unhinged. The damage caused by an average person (non-elite/pro) cheating in a marathon is *not* enough to justify this.
As a competitive runner and data analyst at my "day job", I found this video fascinating. :)
Really interesting video! Thanks for putting it together, and for the effort you put into detecting cheaters.
Great video. How does covering the bib with a jacket miss the mat? Are they not using RF? Obviously bib mules don't have that issue.
This^^ pretty much every race these days uses chip timing which is going to pick up the mat crossing even if you have a jacket on
I know on all the trail races I do they tell us the bib needs to be on the outside layer of clothing to make sure it gets picked up
First marathon in 2 weeks!!!! Wish me luck
You got this!
GO GO GO!! In those last 10km just remember... if it hurts to run and it hurts to walk, you may as well run!!
How'd it go?
Love this analysis. Would love to see the analysis part in more detail!
What a fun project! I'm a data nerd too
Maybe put air tags (that correspond to their bib number) on their shoes to prevent some cheating?
I think this is a great kind of content. It would be nice to see more of this “finding cheaters” in marathons 😂
I agree!
This was a fantastic video! Thank you so much!!
Yep, maybe also there should be a whistleblower email address, where people who have this kind of data could send the data to the organizers and trigger an official inquiry. Maybe it wouldn't help others who finished after the cheater in that race, but disqualifying them from other future races does have at least some positive effect.
How is it that times aren’t being recorded because a runner has a sweater over bib? I thought the chip records when they run over a mat.
When looking at the data I was trying to find out why all those runners missed the timing mats. I looked into some of the runners who did missing timing mats. Looking at that specific runner who had all her timing splits missing was my starting point then looking at other runners pictures who had missing timing mats I noticed they shared similiar qualities. Wearing heavy clothing.
Some races have people wear a bib number AND a timing chip on their shoe/ankle, but I believe as of late the chip is attached to the back of the bib. Open question as to how well chips will work if under clothing, but a general takeaway should be that "missing a timing mat" could be due to covering clothing or a faulty chip.
I'm a bit perplexed by this. Most race bibs have a sensor/RFID chip affixed to the back. Berlin marathon, for one, still uses a timing chip that's attached to the shoe. In Chicago this year, there was a warning to have your bib visible and not covered, but I figured that was more related to photography and perhaps visual identification for fast runners/age group winners. In a big race like Chicago, where you're going to get more recreational runners, they may not know to have bibs visible. Also, a couple timing mats seemed kind of narrow this year, and it's possible that someone could inadvertently run around it. If heavy outer clothing interferes the bib RFID picking up the electromagnetic wave from the timing mats that energizes the runner's RFID chip, this is something that should be more heavily emphasized at races where cooler weather is anticipated. Great research!
@colinmcc8564 definitely missing timing mats could be due to faulty chips, person did not go over matt, and covered clothing. My premise why I put clothing as my reasoning was due to the fact after the first missing timing mats the runners times start getting registered. This data set is from 2018 Chicago Marathon. I am curious if I could find similiar results again.
@@3BrosRunning It would be interesting to apply the same type of analysis on Chicago 2021 because it was hot that year. I would not expect many runners to be wearing heavy clothing. If there were little or no runners who missed timing mats, then either less people cheated that year or the chips worked properly. I would lean towards the chips working properly given the no heavy clothing.
Super well done ~ and yes, interesting and should be very useful for organisers and as a disincentive for those contemplating to cheat.
It’s very important to call out everyone
I am no Olympic runner but I do try hard in races. I had a 1:44 qualifying time for a race (that shall remain nameless) and got placed in Corral D for the start. The corrals only went from A to F so I was a bit surprised at starting so far back. The woman starting next to me had a 1:30 qualifying time and she was mad as hell at being in Corral D. Just looking at the people in front of us we knew that cheating must have been rife in the qualifiers for this event. There are certain physiques you can tell are never going to run a sub-2-hour half marathon (You know what they are).
Sure enough, once the race started about 95% of the runners in the corrals ahead of us were never going to run a half in under 2:30 they jog-walked from the start line. We both ended up walking and jogging as we tried to zig-zag through the crowds.
A frustrating and ruined event for me. I only got completely clear of the crowds at about the 8-mile mark. The only bright side for me was not only did I get a negative split time it was the first and only time I have ever finished a half marathon with a sub 20min 5k for the last 5k. It would be interesting to see how that would show up on your graph. 🤔
Event organizers everywhere need your program for every event they run to stop the things you highlight in this video. Cheating can ruin running for other people in an event.
If you can ran 5 k under 20 min, you should be a 1:30 HM runner....
@@yongdongyou3128 I wish I could maintain that pace for a half marathon.
I would guess that most of the slower runners in front of you in the faster corrals just cut into a faster corral and covered their bib, yes? At the Philly Half Marathon, they didn't check bib colors that carefully, and I remember mid-race passing a bunch of runners who were supposed to have started in a slower corral than mine.
@@mmj1342The event is very thorough about checking bibs and keeping people fenced in the right place. Very strict marshalls on every fence line. No corral jumping, unless you want to jump backward which is allowed. I think all the cheating went on in the provision of evidence to get the corral position. I have recently been to organized half marathon events with less than 50 runners where I know no women finished in front of me. But when the results were published there was a Susan who finished just in front of me and wasn't even the female winner! The guy in front of me didn't look like a Susan. Possibly hard to say these days though. If there is no prize for the event I don't think anybody checks.
Thank you for doing this and sharing. Def interested in the code video as well.
I’m incredibly slow right now and have had injuries for over a year. It inspires me to train and get quicker. It means nothing to brag about something you didn’t earn. Very insecure people.
One time I did not start a race, but my bib was in my bag and I passed near the finishing timing mat around the time other runners were finishing😅
Thanks for doing this video…. Very interesting! I never dreamed that a person would cheat…. How can they sleep at night? Thanks again! Run on 🏃♀️ 🏃♂️
I’ve ran a marathon with a jacket over my bib, i didn’t miss a single timing mat?
I ran in that marathon twice. On the way to work. You kind of have to merge and cross on an angle, all while keeping pace with everyone running.
Could be a bib swap also happen accidently in groups, mostly the name stands at the number but i don't know if this is always the standard. i didnt run a lot races, but in all of them i registered with colleges and friends, and one of them is getting the numbers before the race and then they give it out to the group at raceday.
US Roadrunning has 5k to Half Marathon times - to help people train and fartlek is a good method and I use it, and that’s where I’m doing Chicago and a 5k leg of Denver in relay. My scratched super sprint triathlon entry had a great time for a 1.2k run for a lot of people when I ran 5 and biked over limit. I need to get my stopwatch right next time.
You're doing the Lord's work. Love to see it.
Not sure about Chicago, but when I did London there were 3 different start lines that converge after the 5km mark. It’s possible that someone could miss a mat or a particular mat on one of those starts could have failed.
This is awesome! Thank you. I love running races and enjoy doing my best.
The lady's timing chip wasn't working or was misplaced
This was an awesome video. I’m in Data Analytics/Science and an ultra runner
well ultra runners have plenty of time for data analysis during the run.. do you take your laptop on the runs?
Thank you ! Very interesting video.
I have some theories why people would cheat. 1) A semi-elite (close, but not good enough) trying to steal a win or Top 3 finish in the race or their age group. 2) Someone trying to fake running a race and finishing or getting a personal best for social media status. 3) Someone trying to get to Boston or another marathon requiring qualifying. 4) Someone trying it that is not properly trained for it or is out of shape or maybe suffered an injury. They may try to take a shortcut to get to the finish line and know that they aren't going to come even close to winning any type of awards anyway.
I haven't ran a mini-marathon yet (first one at the end of this month) but ran many 5k and 10k's and one 15k. I saw a time of like 22:28 in one 10k and that was the only bogus time that I saw. It was a high school age boy on a Thanksgiving run and it was two laps. I think that he got confused and ran just one lap. He may have had an emergency or family need and needed to quit early too. He did get credit for winning the race at the awards ceremony (but was not present to receive the award), but he got DQ'ed in the final standings later when I checked that afternoon. Thankfully running has very few cheaters and only a handful of those cases affect age group or overall winners or major marathon qualifying.
You should absolutely call the cheaters out and expose them.
Why hasn't a machine learning model been built to immediately issue a DQ for various classifications of cheaters? Corrections could be made for those cases with equipment error.
Something I was working on. To make an effective machine learning model I would need more data. I was able to scrape just one year of data but I think I would need 10 years or more to create an effective machine learning model. I have an algorithm I created based on certain parameters. I may release that video soon.
@@3BrosRunning that would be wonderful to see. I have a data science background myself.
@@rashaadratliff-brown2734love it. It's interesting how much you can find by looking at data.
@@3BrosRunning Great video. In my opinion, I think 40,000 runners should be more than enough to create a decent model (though perhaps it would be good to have representation from other races if you really want it use it). Would love to see another video on how you went about it. Seems like a perfect use case for anomaly detection.
A lot of popular races have more demand than places. I've been unable to run before and given my place to someone else. They ran a much faster time than I could, but I'd never try and claim that time for myself. They got to run and enjoy the day and my entry fee wasn't wasted.
It's really about insuring that runners get shirts, medals, foods, etc... we can only estimate how many runners we will have before timelines for shirts and medals come close to the deadline. How to prevent this? sign up early and for our races we will normally let people transfer but for bigger races I can see the issue it can cause.
I don't get how having a sweatshirt or jacket over your bib would keep them from triggering the timer. Isn't it an RF device? Maybe it was just a defective device. But I still appreciate calling out the cheaters... mostly because their "success" cheapens the accomplishments of us honest runners.
Great video . Not that many cheats re % of runners who ran it. Take a look at the Mexico Marathon this year 11,000 people were caught cheating 😳 😅
To be fair running in Mexico City with the high altitude and poor air quality is absolute hell. I was coughing up blood after a run when I visited there
@@SuperFuzzyDunlop I understand that I ran the half in Mexico city nothing justifies cheating no sir egocentric bastards r to b ashamed publicly
This video rocks. Great research
Super interesting video! Would love to see more of this! One question I have: How did you find the bib-swapper with the different name/sex? The only thing I can come up with is recognizing all photos of all runners, or looking at previous results for all runners (or just sheer luck), but all of these are not easy/cheap
As a runner myself I kind of know how people have cheated or the possible ways they could cheat. The photo one I feel with current technology is possible using facial recognition tech. I pretty much just scoured the images till I found one to use as an example. For my actual presentation I shared the name but for this I don't want to call anyone specific out. The bib mule one was one I knew existed and was super happy I was able to find some examples of it in the data and was able to showcase a "possible" cheater. I applied an algorithm to the Houston Marathon and wasn't able to find a bib mule.
I didn't miss timing mats but unfortunately did not get pictures of my race because my bib number was part covered by my running belt, so the AI things they use to generate your photos based on bib number didn't work for me. I was gutted haha but I'm sure I looked awful in the photos anyway....
Great video. Good logical process.
Appreciate it
Interesting video, thankfully it seems like the percentage of cheaters was very low. Seems like you could work with these bigger marathons to help them with cheat detection.
The marathons don't care if they have cheaters they are money making enterprises.
Very interesting. Nice work!
such an amazing analysis 🤯
- when finally completed, would you provide the code / app / software to Others in order to enable Others to check the data of their completed races ? ( I once found the winner of an age-group in a marathon beeing a wheelchair-racer :-) )
- how do you receive and extract the full data efficiently / automatically from the results-webpage ?
What happens if you cut the course without missing any timing mats and wait for your average pace speed before carrying on to save energy for a fast finish.
I think if u do that it would take more work to cheat then it would be to just just run the race normally. There is this instance of a person who did something similiar during a trail race. They would hide in a Porta potty then run when it was appropriate and keep doing that till the finish. It made his times look legible until he was caught. Person used this technique to place in trail races.
My dream and goal run Chicag Marathon (my home town), and the Boston.
This is fascinating. Well done
Appreciate it. Subscribe to follow along on our journey to put on races
Interesting, thanks. was also wondering if you could discern what percentage of people were "probably" cheating. Like others below - and as an engineer - not quite understanding how chips don't work through clothes.
My bib worked in a few races covered by windbreaker. It was in the 30s!
@@frankshorter449wow! You ran in the 1930-s and are still around? If that is not cheating, then I don't know what is :)
This was so cool to watch!
Wow, really cool analysis
Last year my running group's trainer told us that he has free bibs on a race. So i eent there. There were no free bibs, so i was very pissed, but i drove 40 minutes (and woke up 2 hours before i would've not for the race) so i decided to pay on the spot (highest price of course) At the end if the race i almost left for home, when i saw lots of people around some billboard. It turned out that they printed the results and put it on the wall. I managed to find myself and i couldn't believe my eyes! I was 3rd in my age group. What a surprise! This was my best 5k ever! Not only because of the real trophy 🏆 but it was also my PB. This is such a good feeling! I think it would be my best race even if i only had the PB, but it would be a shame not to know that i finished in the 3rd place, not to have a picture of me standing on the podium.
That's why it's important to have tools that can detect cheaters in real time, so they can be disqualified before the results are announced. I heard that in the past there were cases when cheaters were caught and thus someone was awarded a medal (months) later, but those people who really deserved those medals were robbed the feeling and the picture they deserved
love the video! so cool!
If it’s not on Strava then it didn’t happen 😂 🤣😜
Missing every mat probably means they have a broken chip or something (or something covering it, like you found). More likely, some of the runners that missed two or three, but not all, were cheating.
This is interesting. I would like to read your capstone or take a look at your notebook to see the code.
Good point. How many mats would Rosie Ruez have skipped when she cheated at Boston in 1980? I think she jumped in the race in the final four miles?
My sister brags how she finished. I would like to see the how you would analyze her performance in chicago marathon
First name starts with lo last name ends with rose
Interesting video. You look like Daniel Levy btw
how do you differentiate between a cheater for the sake of competition and a cheater for the sake of they just don't care anymore and just want to be done.
Cheaters actually cheat themselves.
So true. Knowing you did your best should be enough.
Nice! Cool project.
I have to be satisfied all you have to do is walk this marathon for any medal, so not coming in on any timing mats at all is an issue that’s easy to fix
Ahhhh, the Windy City, my home town.
yes!
Great stuff, thanks. Sure, someone can miss a matt. It happens. But that and cheating are two very different things. I don not get why would someone display a medal they have not earned.
loved this video
Fascinating! A great explanation and very "human" analysis. I only just run my second race at well over 50. The first with my adult daughter. We went through the finish together with essentially the same splits holding hand in the official photo finish. Yet, I can understand all of the that (from the warnings to not cover the bib) and to that short cut (in a 4 loop race, I looked at the finish and thought: why am I doing another loop , when there is a shortcut to the water and food right over there). Fortunately at my age and low level, there is no pressure whatsoever :-) I am looking forward to the 3rd, but paces might be slower as I am starting to fatigue ;-) I loved the cross though and just got a thrill from running reps alone on the tracks and being much faster than the VDOT tables (not sure that's so great, but I'll look into it)
Do any of the races put GPS devices on the runners?
really interesting
Thank you
You are doing the Lord's work 🙌
Cheaters are pathetic😂
Interesting.
Very!
Cool vid.
😮 The Chicago the Naughty City.
thought it was windy
Chicago's race configuration (essentially three out and backs in three different directions) is really ideal for those who want to cut the course. But you better know where the timing mats are (but apparently, ya can't check the results of all 47,000 runners).
Are you telling me Chicago marathon doesn't have a chip in the bib? Why would they need to SEE it?
Bib mules. Backup evidence
I don't buy the first explanation about the bib covered. If the bib has a chip, it doesn't matter... 🤷🏼♂️
Cracks me up the stupidity of people. Like anyone cares what a non-elite runner's time is or whether he finished a marathon. Big whoopie whether you did it in 4 or 6 hours. And then paying $200 to cut the course or cheat is even dumber.
Personally I'm so anal I was running a 5K where the course had eroded away some brush from previous years and I still ran the longer distance of probably an extra 100 feet because I didn't want to have a short course time. I want to be able to compare to my previous times on the course as well and in general. I'm sure I was the only one that did that in that race.
I also find it hilarious the amount of money people spend on these races to walk and jog them. Like you can't just go out on a saturday on your own and walk or jog a marathon without paying someone $200. I'll never get that. I've done that. Ran a marathon on my own and I do a lot of longer trail runs even ultras on my own. The only time I enter a race is when I will be pushing myself to my own limits and want to be pushed by others, otherwise I will just run the distance on my own.
I always thought a mule was someone that ran using someone else's name/bib not someone that had 2 bibs (his own plus the person's). I would be down to run for someone as a mule. I think it's stupid that someone would pay all that money and have someone else run for him, but if someone is dumb enough to pay me to race, I'd do it.
Hard to believe that clothing obstructs the chip. You sure that is the case? If so it would be thousands of runners.
I realise you’re only looking at instances of cheating in marathons. I’m sure you’re aware that cheating occurs in all races. Personally, I have been cheated out of a first place finish on three separate occasions in shorter races because of these despicable acts.
2:16 HEY YOU CHEATED>! LMAO...jk jk.
2019 i ran the marine corp marathon in heavy rain. I seen so many women get on rented ebikes and speed up the mat timing get off and go through find another bike. Insane bc the tracking that day wasnt working. The app tracker at least didn't work. I feel good i did the entire course, but it was an incredibly hard day. Insane amount of course cutting was going on
Makes you wonder why they entered a running race, right?
What? All the runners were marines? If that's true, it's a shame. Not that it is much better if all the cheaters were not marines just humans...
It's the people's marathon so it's open. I highly doubt it was even military. There's cheaters in every races. This was notable bc of the conditions
Show us the algorithm :)
11:06 be bold start cold.