Soccermatics: could a Premier League team one day be managed by a mathematician?

Oxford Mathematics Public Lectures: David Sumpter - Soccermatics: could a Premier League team one day be managed by a mathematician?
What do you need to win the Premier League? Money? Sure. Good players? Yup. A great manager? It helps. Mathematics? Really? 100%.
David Sumpter will explain. Fans of Barcelona and Paul Pogba might (or might not) be surprised.
2022 update. David now runs a whole Soccermatics course based on his lectures:
soccermatics.readthedocs.io/en/latest/

[@jo87sip29]

Good mathematicians make good assistant coaches. But being a head coach takes more than statistics and analysis.

[@jas2800]

I agree. Mostly the job of the main coach is to talk at press conferences and to motivate the players.

[@egoxagony4623]

@@jas2800 and a mathematician can;t do that. It should be simple to motivate players. By just researching motivational speeches and creating a statistical bias towards certain motivational speeches than others.

[@eliaskv932]

@@egoxagony4623 Lol. It certainly isn’t that easy.

[@ilydevonte4764]

@@eliaskv932 he’s leaning towards being book smart a little too much

[@mahmoudwahbeh963]

@@egoxagony4623 you can’t research every situation that will ever happen. Coaches have to have high emotional intelligence to handle different players

[@bocckoka]

I've heard a Hungarian mathematician (Szemerédi Endre) talk about football strategies, and I immediately thought that this guy has more idea about football strategies than any Hungarian coach, and he should be managing our national team.

[@paulojacobsilva3018]

I'm afraid you were just impressed by someone who never really coached anything. Maybe he sounded smart, impressive. The reality of being head coach is a lot more complex and requires soft skills, managing injuries, phases, egoes. It requires experience. A mathematician, no matter how brilliant, is not a substitute to someone who actually does coach. Coaching is not easy. But the public has the impressions that it is somehow - it's an illusion trick.

[@ratedpending]

@@paulojacobsilva3018 yeah a mathematician could be a great assistant coach, but in order to be a head coach they 100% would have to have football experience

[@mattiabeltramo9120]

Unfortunately for him, football it's another thing. In sports and life there are the cathegories, there is who can and who can't.

[@bakhodiryakubov3981]

@@ratedpendingJose Mourinho was not a professional player, yes he was a translator but that's not a football experience. If you argue that he learned football from being close to managers, mathematicians also can do that - assist manager and then become a head coach

[@paxdriver]

This video aged well. Leeds did get promoted 👍 good call. And losing mane did hurt Liverpool massively. Most importantly though, long shots lead to disorientation in the box and putting keeper out of position - meaning these data don't account for opportunities created by long or sharp angle shots that don't go in the net. Rebounds and deflection are very common ways for a ball to pop out to an open striker in that prime scoring position in the 6yd box.

[@danielp4528]

not really. you dont get that many rebounds from long shots. either player is offside or defender gets to it first

[@celestialrex551]

we replaced mane . he declined

[@manee2412]

@@celestialrex551 cope

[@sss63763]

@@celestialrex551lmfao how deluded can someone be

[@delmanpronto9374]

the point is that statistics are infinite. the number of variables are infinite. there's no end to how detailed the analytics could be. soon enough, we deal with data overload. it's a very complex world, that's why there's only one magnus carlsen, one alex ferguson, one guardiola, etc.

[@racarth1]

Interesting how he sometimes slips into speaking Swedish, e.g. 27:43: "till exempel" instead of "for example"

[@Gameboob]

What a great presentation. It's a shame that they didn't upload the slides. Can barely see any of the data

[@OneMor3Win]

I used to be rank 1 in Fifa for 07, WC 10, and 15, not calling myself a mathematician, but I’d apply a mentality to figure out the most favorable probability for every single situation. I’d have a very average start for over 100-200 games.. eventually something would click in my head as I was playing and I’d just have a solid understanding of what was the highest probable pass/shot for every possible situation, even when defending, I’d force oppositions into angles to take the tempting least probable chances on purpose. Instead of sometimes playing pretty and safe football. Id intentionally brute force my way as often as I could, the more I did this, I’d learn from each turnover what was ineffective, I’d slowly become more and more efficient converting chances/ timing skills or shots etc. I even caught on fast, skipping celebrations/cutscenes immediately/not making any subs/pauses decreased the likelihood of a sudden rage quit dramatically.. perhaps this wasn’t smart as wins helped with ranking, but I enjoyed chasing the record gpg. By the time I ranked up to #1 and stopped playing 15, I had 5.5 goals per game average, a full goal clear of the second highest.. and the 2nd lowest goals conceded ratios of the top 100 players. There’s also different short periods of scripting that would intentionally hurt and help you.. small context clues in player agility/acceleration would give it a way to help you adjust your approach during those moments. I never looked at Fifa as a game in which you play for fun sadly.. it was clearly mostly all about taking/ timing the most likely and efficient path as often as possible.. FIFA’s predictability/tuning/execution is drastically less consistent now and has a rng element to every action so I stopped try Harding after fifa 17.. but I’m ironically actually having more fun throwing games and playing casually whenever I load it up… but yeah just saying a lot of this logic can be applied to certain video games easily.

[@ALEXLAD94]

The best player ratings thing is ridiculous - of course players in top teams will be in that list. When they pass to teammates, they have better first touch = better passer rating When they defend, their teammates help press = better defensive rating When they have chances, it’s from better weighted passes = better shots on goal

[@pslanez]

If you look at one of the greatest managers, Alex Ferguson, the thing he was best at was his people skills. He knew how to get the best out of his players and the refs. It'd make sense to have a mathematician on staff as an advisor but not as the manager.

[@danielp4528]

yeah person skills and money and the best players

[@notfunny007]

@@danielp4528 He created the best players and generated the money

[@joelm7621]

@@notfunny007 well said

[@NotfromDateline]

You can have both

[@aryanspecialtruthonly.2311]

I disagree with the prejudice of Mourinho He’s extremely tactical and wouldn’t be surprised if he’s using deep analytical methods He uses height and strength of players to his advantage for example which is basic math

[@OxfordMathematics]

Actually later in the lecture David does give Mourinho a bit of credit (reluctantly as a Liverpool fan!)

[@thomasellis445]

@El_Glu statistically Pogba has been a very inconsistant player, mathemtaics has to be consistant ;)

[@Akill379]

Essa palestra tem insights interessantes, a ciência ainda é bem pouco aplicada no futebol brasileiro e as estatísticas tendem a ser subestimadas como meio de avaliar um jogador por aqui. Mas com os aplicativos de estatísticas essa tendência vem mudando no Brasil, antigamente a matemática por aqui só era mencionada quando envolvia as chances matemáticas de um time ser rebaixado para a segunda divisão. Eu acredito que a matemática pode fazer pelo futebol o mesmo que fez pelo basebol com o advento do Moneyball.

[@sakibshadman1448]

He lost me when he said - "Paul Pogba defends very well"😭

[@alimamulma3sum14]

In Juce he used to but in Man U he didn't for some reason.

[@sakibshadman1448]

@@alimamulma3sum14 Mate in Juve Pirlo was 6 and pogba was advanced 8. Juve mostly played back 5 and attacking players hardly tracked back.

[@joelm7621]

@@alimamulma3sum14 because Juve actually had tactics

[@andrasszabo1570]

@@alimamulma3sum14 He wasn't defending well at Juve. He began as a #8 alongside Vidal, but that didn't protect Pirlo enough in the most difficult matches. In the knockout rounds of the CL Marchisio had to come in to play there in central midfield, Pogba ended up as #10

[@grigornenov3151]

Ottmar Hitzfeld was a trained mathematician and one of the most successful coaches in the history of football.

[@lnhart7157]

He isn't, he is a trained teacher for sports and mathematics.

[@michaelmuller4356]

@@lnhart7157 So teachers of mathematics did not learn about mathematics in detail? What country are you from where this is the case? Teachers don't (necessarily) do research, but they are trained in mathematics.

[@lnhart7157]

@@michaelmuller4356 I'm from Germany, presumably the same country as you, and no, teachers don't learn mathematics in detail here. They learn math, sure, but at a lower level than engineers, let alone mathematicians. Nobody would call a math teacher a "trained mathematician". They're trained ... teachers.

[@michaelmuller4356]

@@lnhart7157 Where? Here you first learn about the topic in detail and then pedagogics and didactics on top. They even hire Experts without pedagogical training.

[@petarmilic6419]

Hammarby finished third in the Allsvenskan, one point shy of winners Djurgården. Looks like it works

[@u.v.s.5583]

Being one point SHY? That's actually another consequence of using mathematics. Mathematicians are often shy in general.

[@PoppaDocRocks]

Yea there is a ton of mathematics involved but there is also a lot of emotional mgmt, man mgmt, communication, individual psychology, team/group psychology, etc…

[@steviervl8731]

Now 2 years where’s Leeds United ? This is amazing

[@prakashbhanwariya6386]

Excellent

[@charlietian4023]

I've been saying this for a long time. Baseball was taken over by stats even earlier because of its discrete nature. Basketball recently too had a wave of analytics. I think (American) football does well with this similar type of thinking and passing network too

[@andrazstrmcnik2331]

yes and all of those sports are unwatchable now

[@charlietian4023]

@@andrazstrmcnik2331 that's just Luddite bias though. Players playing optimally is anything but unwatchable. It was always just as unwatchable

[@zihanyuan3413]

A very interesting video, especially for Liverpool fans!

[@ErtjonMecka]

that was awesome. a movie should be based on this :)

[@mapper7310]

not football, but Moneyball is about how data was applied to baseball and ended up with wild success

[@Anonymous13448]

this should have several million views

[@cityzens634]

2 flaws in this thinking 1 Rajeem Stirling is an instinct player, if he has time to think in pressure situations he usually misses. A lot of his goals come in games where the team is already comfortably winning and he doesn’t have the pressure to score. 2 A team is only as good as it’s defense , if you are leaking goals at one end then scoring goals will have a lessening effect on the result.

[@proflame4767]

These aren’t flaws ...?

[@noahhughes127]

You might be correct about "pressure", but this can be parametrised and taken into account in-line with the current "thinking" and approach to the problem. Also, the second "flaw" was taken into account; for example, in the Markov model defensive actions were weighted.

[@kalfch1]

In France, there is Christian Gourcuff, father of french footballer Yohan Gourcuff, who is a maths teacher and was manager of many Ligue 1 teams. He is considered as a theoretician of football

[@ALEXLAD94]

Re shooting outside the box - you get more chances to do it than having a point blank shot from straight in front. Also if you stop shooting from distance, defenders will just sit back and block up space in the box as they don’t neeed to close down longer shots. 18-22 yards out with a slight angle are still easy money if you can curve it into far corner.

[@michaelmuller4356]

Yes, expected goal model should take into account the skill of the player / the likelihood a player takes a shot at all. Even if you get most defenders in a 15% goal probability position and force them to shoot, I highly doubt the overall conversion rate is that high. On the other hand, they know this, so they rather pass the ball than try shooting.

[@Akiak7]

18:45 its not just randomness... have you considered that you can be a good shooter or a bad shooter? precision is variable and so is shot power???

[@aaravsharma9430]

Love this !!!

[@averagehauptsatzenjoyer]

What does the purple colour depict in the horizontal histogram shown at 25:22? It is not in the legend on the slide, and I did not hear an explanation of it by David during talk.

[@GodofStories]

Especially now. It's definitely yes. Once you take all the data to a DL/ML model. You can moneyball your way into a championship, eventually.

[@___Q-bot]

i wonder where did he get the dataset for player movement? using on-site camera and openCV? or if such data is freely available for download?

[@amir9178]

I would love to see these statistics and examples used for someone who heavily abuses the box such as Erling Haaland.

[@kirillacharya343]

This is quite interesting, though telling the player that it was a right decision to pass across in my opinion is not correct. We talk about probabilities here, and the model does not imply the player's passing characteristics, the chemistry between the players and many other factors. Yes, the probability of the successful across pass is more, yet the aim is to score a goal. And maybe passing down to the striker is more possible to score a goal. Footballers aren't machines, they have only a second max to decide where to pass (if talking about the situation near the box). There are a lot of things to think of, and obviously having this data one can improve the team performance, but anyways there are too many variables.

[@nicholaskim5187]

This is so sick

[@thlee3]

is ball telemetry used with grading a player? shot/pass power? and isnt there gps on the ball now?

[@just_bee9482]

A system in which Pogba defends well is very dodgy

[@johngafa3480]

At what point does the winning of a league championship, like the premier league, become impossible for the team or teams which are trying to keep up with the leading team ? I'm sure the pundits have a formula. What is it?

[@samtorr3947]

It's somewhat easy: Assume that Team A (The leading team) does not win another match after the current matchday. It will be impossible for the other team to win when the points available if Team B (The chasing team) win all their games < the gap between Team A and Team B

[@u.v.s.5583]

@@paftrsickl9339 It is an algorithm with finite number of steps. I don't see how such algorithm might be different from a formula. 3*N > dP, where dP is point differential, N is remaining games for the chasing team

[@samtorr3947]

@PaFtrSickl a formula isn't necessarily symbols however, if you can't connect the dots in your head: For team A to win early: A-B>R×3. Where: A is team A's points. B is team B's points. R is the remaining games.

[@kibetronoh2376]

Roberto Carlos scoring from the corner quadrant 0°.

[@mikailconstantbilyamin1832]

Error!

[@antoinecovet5681]

if that's the futur of football, it's quite frightenning... But hopefully if Barça are losing 8-2 while using these models,it might mean that football is more than that

[@michaelmuller4356]

@JÜRGEN KLOPP: Call that guy already.

[@sss63763]

Football is more about instincts and on the spot tho

[@aynrandom3004]

A scout or advisor perhaps? reminds me of the movie moneyball

[@sastaffa]

"Money Ball'....statistics and evaluation

[@parijatsaha3977]

UNAI EMERY - GOOD EBENING. Can someone SUMMARIZE please?

[@thewalkingjoke3843]

I don't think analyzing soccer to that degree would give any useful practical tools. It would make sense in simpler sports, but in soccer there are just too many variables. The focus should just be on your own roster and the enemy roster, and simple principles and tactics that are based on human weaknesses and strengths.

[@advancewarstournamentseries]

Ok, but this all based on maximizing odds for "perfect play", the thing about it is that against actually skilled opponents you can't play safe and win, you have to take risks, like that pass you said he was right in to passing to the guy that got mad, maybe it wasn't the best pass, but if you're losing against a better team and they are playing well, you won't score anyway. I think more aggressive risk taking should be added to this model or otherwise it will stale quite quickly, if everyteam is playing "perfectly" then the team with the best talent always wins

[@Hieillua]

Quick answer: no Because there's also still a human aspect. Man management is super important too.

[@nathanmingle7]

Alphazero in chess changed my mind on all this shit. There is no one play which will give you a scoring chance. You have to make the defense react to different plays and exploit the weakness that they show while their on their heels. So it just takes a better knowledge of the position and instinct. They're *

[@akshayanthivakaran9634]

Man said , help liverpool get promoted 🤣🤣🤣🤣

[@Magihyun]

Pep Guardiola is your best example.

[@clash44]

No managing people requires knowing people and having the soft skills to engage. Math is helpful for analysis though.

[@oliverforway3868]

I think it's important that you have a football mind as well which allows you to interpret the data you're given correctly. Anyone can gather data but it's how you interpret it to a football game. I doubt a super mathematician with no knowledge of Football could make a great manager.

[@rah-mown6392]

My curve ball is literally the y axis (it doesn’t curve it just goes straight someone please help me I can’t curve the ball 😭)

[@francescos7361]

Great Pogba , the real Batigol .

[@RoyalBlue1905]

On what basis does he mock Mourinho? Proper Liverpool bias cuz Mou destroyed Gerrard's dream. If anything, it is Mourinho who is excellent with stats and tactics. Blatant lie in the opening.

[@tesla5297]

Calm down Chelshit fan. Mourinho ain't nothing after RM stint.

[@pher8902]

He apologizes at the end lol

[@TheGarfield1337]

Ottmar Hitzfeld was a mathematician

[@mm1k3y]

Possibly they could get by. They'd never produce world class talent from the inside and they would be so dull and boring. Football is not about stats. Ronaldinho is always my example for this. Roberto Soldado has more career goals than Ronaldinho but anyone with any passion for football knows who is the most talented.

[@tastytoast4576]

Liverpool should’ve listened. We’re fighting for our life this season 😭

[@SafE29]

the lecturer is actually a liverpool fan too!

[@johnfarell7471]

This guy definitely likes blue lock

[@davenalunat1433]

Maybe, but there is a reason why coaches in football are called "managers".

[@sdsdsdsdsddssdsd4040]

So easy 😎😎😎😎😎

[@easy-going6097]

17 :34 ، اين انتم يا عرب ؟!

[@yacinecabba363]

هنا ...

[@easy-going6097]

الحمدلله '

[@marwafouith5786]

نعم

[@theoofth]

This is very interesting however, there are so many intangibles in football! And there has to be a human element which you cant account for. If it wasn't human I think bielsa would win everything.

[@vsubhuti]

It is all obvious to me , passing from back to counter attack

[@jcoz1]

Diaosegoubeihoutongni! Cao

[@egonjohnssonaberg8909]

Åh Hammarby!

[@rabbyhasan4507]

Is this the guy who predicted Brazil would win the world cup ?

[@5MinuteMedicinePearls]

Judging by your predictions for this world cup, i would say no🤣

[@akiiis1439]

Its footballmatics..

[@somejinggg]

I hate math but-

[@mikelam9253]

rub it in Manchester United fans … got me chuckling

[@delmanpronto9374]

yes, because the mathematician will be able to tell if a player is in good form or not, feeling confident on a pass/shot/cross or not, determine how a player will respond to a hard tackle, what equation he has with a fellow teammate, whether the injury to the ankle the week before affects his performance on the night... every passing day, i find the world is trying to "quantify" things, turning everything into a "statistic" or putting a dollar value on it.... life is a lot more complicated than that. this cycle of putting a value/number on everything is getting a little boring. these professors should stop dreaming and do something more useful with their time.

[@mahirsudad2311]

That is the day football dies

[@parthibbhattacharya1308]

Just an opinion. I do agree that mathematicians can become good coaches. But some of your arguments in this talk are really biased, which means that your book must also be biased.

[@TheRealCristiann]

No you can't , soccer is a fun sport you cant calculate fun passes you can only calculate numbers !

[@just_bee9482]

This Markov chain model is rubbish. Pogba defending well . Are you kidding me 🤣

[@sharukhrahman5304]

Who's here after the World Cup 2022 prediction failure 🤲

[@kingrobert7246]

Were going to use a fourier transform laplace formation Players: What?

[@aelnenaey]

So the data basically shows that the probability of scoring a goal increases as you get closer to the goal? You don't have to be a mathematician to figure that out. But thank you Oxford University for this groundbreaking research.

[@novideostoday]

Nerds never played a single match in real life so what can we expect. They have never touched the ball outside FIFA 😁

[@Andrew-rc3vh]

This is one of the reasons this country is so damn stupid. The top-rated mathematicians of the country burn their time on games. They waste themselves, where say they could be using their mathematics to run the NHS more efficiently. You know - solve useful problems like we once did before we mentally grew down rather than grew up.

[@joelm7621]

Why did they remove the downvote button?! Using the word Soccer and Premier League in one sentence - shameless! What a disgrace

[@sophocles1198]

KZbin removed visible dislikes in 2021. I miss them. The creators can still seem them.

[@rc....]

YNWA :)

[@gideoncheung8731]

answer: no.

[@saki1333]

The answer is no

[@tj_1260]

Akk

[@bob-yt8co]

first mathematicians think they are artists, now this...

[@pakeka6175]

🤣

[@epicmarschmallow5049]

At the highest level mathematics requires creativity and inspiration, and the end result is often beautiful. Study complex analysis for instance, and tell me mathematics isn't beautiful

[@kingrobert7246]

@@epicmarschmallow5049 it isn’t beautiful. Screw maths

[@u.sbanban]

football isnt just science, its mostly art

[@mazt7]

If they still call it soccer, then no....

[@manee2412]

its called foofball first of all

[@letstellourselvesthetruth2432]

Hell no 😂

[@Polimuni]

Short answer: No. 🤷‍♂

[@thomasmac8563]

17 minutes to say anything of note

[@xavisanchez7522]

Never.and it is called FOOTBALL 😂

[@bytrixx6901]

._.XD

[@sachindey3242]

The simple answer is "NO". Thank you for reading.

[@TernaryM01]

Should be called "Footballmatics" instead. This game really has nothing to do with sucking.

[@markarmage3776]

Yeah, the guy doesn't understand the game very much. Reducing extremely complex situations where countless variables come into play to a simplistic model that is obviously false is mathematically sacrilegious. All models are wrong, but this guy's model isn't even close to being useful. Football, Basketball, Hockey, they're not like baseball, far more complex with far more variables, infinitely harder to make an acceptable model.

[@paulojacobsilva3018]

the players are humans. So, no. the end

[@norbulepcha5930]

Ok haaland

[@asvas5036]

>soccer

[@jamesbolivardigriz8252]

this isn't even good statistics, let alone good management...

[@jamesbolivardigriz8252]

why stop at a one continuous angular parameter model of the probability of scoring when you could be really "applied mathematical" and reduce it to a binary parameter: whether one of the little numbers on the top left of the screen is incremented shortly after the shot. just think how much meaningful complexity you could elide then...

[@jamesbolivardigriz8252]

considering nominating this work for the annual prize recognising laudable contributions in the vein of the ancient applied mathematician Ignobilis of Cretes

[@jamesbolivardigriz8252]

also conspicuously fails to reference prior art by esteemed applied mirthmatician Harry Enfield (et. chums) circa. 01990 kzbin.info/www/bejne/eKjOZGupp6Z2qKM

[@u.v.s.5583]

In reference to your user name I wish to refer you to a place known as the hell in that particular fantasy universe.

[@terelaupan963]

this is getting really boring

[@terelaupan963]

boring boring borrrrrrrrrrrrrrring

[@Beastly707]

Maybe if football players where robots this shit would work...