This discussion is now closed.

Check out other Related discussions

The Manchester United Thread XIX

Original post by TheDefiniteArticle

football is a game

therefore, football is reducible to numbers and any attempt at pure qualitative analysis is rather missing the point

if you think you can reduce football simply to numbers and algorithms, you aren't a real supporter of the game :smile:

Reply 6641

Rk2k14

17

Numbers mean nothing.
Whoever turns up on the day will win.

Reply 6642

TheDefiniteArticle

20

Original post by Lúcio

if you think you can reduce football simply to numbers and algorithms, you aren't a real supporter of the game :smile:

Oh I watch football because I think it's fun, and no, I'm not a real supporter of the game, I don't see the point in supporting a local team etc etc

my point is that most results boil down to expectation with some measure of variance, it's quite possible for a manager to be unlucky for multiple seasons in a row (thus I'm sceptical about the skills involved in football management). basically anything to do with results (including player/team performance stats) we can deal with more effectively with maths than asking some trumped-up 'expert' like any ex-footballer ever for their opinion

the match of the day analysis would be worth watching if it was done by statisticians and game theorists

Original post by Lúcio

LVG has a good record against top teams this season

We just played our main rivals at home needing to score 3 goals, created **** all in open play and made 3 defensive substitutions

If you take out the FA Cup games against lower league sides we've won something like 2 of our last 15 away games

Reply 6644

SirAlexFerguson

14

Original post by TheDefiniteArticle

How has the notion of expected value not filtered through to football fans yet...

If we have 20% chance of beating City and 30% chance of drawing, our expected value in terms of points is 0.2*3 + 0.3*1 = 0.9, so one point ahead of us with a game in hand would be an advantage. If on the other hand, we altered those numbers to 30% and 40%, our expected value is 0.3*3 + 0.4*1 = 1.3, which would lead to the opposite conclusion.

https://www.youtube.com/watch?v=MgCzKQJyWdk

You remind me of this advertisement - wait till the "the professor" part :smile:

Reply 6645

doobucks

16

Original post by TheDefiniteArticle

Oh I watch football because I think it's fun, and no, I'm not a real supporter of the game, I don't see the point in supporting a local team etc etc

my point is that most results boil down to expectation with some measure of variance, it's quite possible for a manager to be unlucky for multiple seasons in a row (thus I'm sceptical about the skills involved in football management). basically anything to do with results (including player/team performance stats) we can deal with more effectively with maths than asking some trumped-up 'expert' like any ex-footballer ever for their opinion

the match of the day analysis would be worth watching if it was done by statisticians and game theorists

Well yeah there is place for statisticians in football but the kind of mathematics you are using to back your point is utterly ridiculous. I don't think I need to go into why expected value is ridiculous but I will do anyway.

Firstly if United have a 30% chance of winning and a 40% chance of drawing, City will therefore have the exact same expected value as you lot, therefore not the 'opposite conclusion'

Lets not get into the fact that you are plucking probabilities from thin air and as these are singular events where the probability is ever changing due to the dynamic situation of a football match and almost unquantifiable, such a thing is impossible to put into numbers. Have you factored the chance of a dodgy ref, the chance of a red card(which depending on the teams approach, whether they are playing home and away which has a fairly obvious affect on a refereeing performance)

It is possible to qualitatively see a teams chances of a performance, by seeing their previous results, their style of play, the strength of their players, their fitness etc which again you cannot quantify.

(edited 8 years ago)

Reply 6646

TheDefiniteArticle

20

Original post by jam277

Well yeah there is place for statisticians in football but the kind of mathematics you are using to back your point is utterly ridiculous. I don't think I need to go into why expected value is ridiculous.

Firstly if United have a 30% chance of winning and a 40% chance of drawing, City will therefore have the exact same expected value as you lot, therefore not the 'opposite conclusion'

obviously I'm using very basic maths to illustrate a point, I don't exactly expect everyone ITT to be comfortable with Bayesian concepts or whatever (which would also be applicable to football for obvious reasons)

It is the 'opposite conclusion' as to whether being a point ahead with a game in hand is a preferable situation or not

Reply 6647

TheDefiniteArticle

20

Original post by jam277

Lets not get into the fact that you are plucking probabilities from thin air and as these are singular events where the probability is ever changing due to the dynamic situation of a football match and almost unquantifiable, such a thing is impossible to put into numbers. Have you factored the chance of a dodgy ref, the chance of a red card(which depending on the teams approach, whether they are playing home and away which has a fairly obvious affect on a refereeing performance)

It is possible to qualitatively see a teams chances of a performance, by seeing their previous results, their style of play, the strength of their players, their fitness etc which again you cannot quantify.

of course I'm plucking probabilities from thin air, I don't expect it to be perfectly quantifiable, but we can probably, with access to the correct data, isolate what's relevant and get a pretty good guess... and of course it's dynamic, but I'm talking about taking a snapshot at a single point on an ex ante basis... we can factor in dodgy refs etc with access to enough data

'qualitatively' -> goes on to mention a bunch of quantitative factors... or at least all of which can be meaningfully proxied through other quantitative factors

also, I'm not denying that one can qualitatively get some impression, or use qualitative factors to gauge quant variables which we either don't know or can't effectively take into account for some other reason, but i do believe that it holds that attempting to gain any useful insight on a purely qualitative basis is completely absurd

Reply 6648

doobucks

16

Original post by TheDefiniteArticle

obviously I'm using very basic maths to illustrate a point, I don't exactly expect everyone ITT to be comfortable with Bayesian concepts or whatever (which would also be applicable to football for obvious reasons)

It is the 'opposite conclusion' as to whether being a point ahead with a game in hand is a preferable situation or not

Ok then use proper maths because you have not illustrated anything with that maths because you have plucked those probabilities from thin air.

It is not the opposite conclusion because firstly United do not have a game in hand over City(who they are challenging for 4th) and they are facing them tomorrow therefore are using up their game in hand(over City their challengers for the top 4)

Reply 6649

TheDefiniteArticle

20

Original post by jam277

Ok then use proper maths because you have not illustrated anything with that maths because you have plucked those probabilities from thin air.

It is not the opposite conclusion because firstly United do not have a game in hand over City(who they are challenging for 4th) and they are facing them tomorrow therefore are using up their game in hand(over City their challengers for the top 4)

i'm talking hypothetically, i don't even know who's where in the table atm (except I know Leicester are top) so I definitely don't know the actual stats, I just like it when united win, and I think there's more interesting/useful ways to talk about things than 'oooh they've been playing poorly lately' (I haven't seen any actual analyses of the relevance of form as a predictor of future performance, which is definitely something we can isolate)

abstract argument is far more interesting than the *******s football people normally talk, who really gives a **** who manages a team next year

Reply 6650

TheDefiniteArticle

20

Original post by SirAlexFerguson

https://www.youtube.com/watch?v=MgCzKQJyWdk

You remind me of this advertisement - wait till the "the professor" part :smile:

haha, tbf I think people who know their stats (as in the maths) are the only people who can realistically gain an edge over bookies, but realistically the types of people in this advert are all the types of thickos who Ladbrokes will be greedily rubbing their hands at (especially mr gut feeling who's probably down tens of thousands on sport betting lifetime)

Reply 6651

Mo De Lemone

1

Diego Torres reporting for El Pais that United and Mou have signed a pre-agreement for Jose to become our next manager; http://deportes.elpais.com/deportes/2016/03/19/actualidad/1458411442_468100.html?id_externo_rsoc=TW_CM

Reply 6652

doobucks

16

Original post by TheDefiniteArticle

of course I'm plucking probabilities from thin air, I don't expect it to be perfectly quantifiable, but we can probably, with access to the correct data, isolate what's relevant and get a pretty good guess... and of course it's dynamic, but I'm talking about taking a snapshot at a single point on an ex ante basis... we can factor in dodgy refs etc with access to enough data

But how can you take a snapshot of something that is dynamic and subject to change to such a large degree over 90mins?

'qualitatively' -> goes on to mention a bunch of quantitative factors... or at least all of which can be meaningfully proxied through other quantitative factors

also, I'm not denying that one can qualitatively get some impression, or use qualitative factors to gauge quant variables which we either don't know or can't effectively take into account for some other reason, but i do believe that it holds that attempting to gain any useful insight on a purely qualitative basis is completely absurd

How do you determine a teams chances of a red card in a specific game, especially since some players are going to be more prone to a red card than others and you also have to determine the team selection used(because certain players are more likely to get a red than other players), how sway?

The type of data you require to do such a quantitative analysis is so extensive that there is no point doing it, because it is ridiculously subject to change.

Honestly bro leave it.

Original post by TheDefiniteArticle

i'm talking hypothetically, i don't even know who's where in the table atm (except I know Leicester are top) so I definitely don't know the actual stats,

Then don't use something like a 'game in hand' as facts to back up your point when it's clear that a draw is a very unfavourable result for United even if the expected value you then used is gearing United for a draw.

I just like it when united win, and I think there's more interesting/useful ways to talk about things than 'oooh they've been playing poorly lately' (I haven't seen any actual analyses of the relevance of form as a predictor of future performance, which is definitely something we can isolate)

Well so do I. But individual factors means that it is very hard to quantify this stuff accurately.Betting companies e.g. utilise the actual gamblers flow of money to formulate their opinion on odds e.g. and while some more meaningful statistics would be useful, it is very hard to have an accurate conclusion on a teams chances of winning a game in the way you are saying.

abstract argument is far more interesting than the *******s football people normally talk, who really gives a **** who manages a team next year

Yeah so basically you don't watch football or really have a care for but makes an opinion on it and predicting it.I'm out.

Reply 6653

Dirtybit

OP

20

There is really no scope for statisticians in football, I've done the research. There are way too many variables that may be effected. The guy obviously doesn't watch football.

Teams like Leicester who cruising the league right now are testament to this.

(edited 8 years ago)

Reply 6654

Lúcio

19

Original post by Dirtybit

There is really no scope for statisticians in football, I've done the research.

Teams like Leicester who cruising the league right now are testament to this.

This.

That's exactly why we all love the game so much tbh.

Already refuted dis.

Original post by jam277

But how can you take a snapshot of something that is dynamic and subject to change to such a large degree over 90mins?

Because at any given point in time, that probability is static. P(team A win) is obviously much lower immediately before they score than immediately after, but at any point in time, it is static.

How do you determine a teams chances of a red card in a specific game, especially since some players are going to be more prone to a red card than others and you also have to determine the team selection used(because certain players are more likely to get a red than other players), how sway?

We can assess the probability of each player being selected in both teams, assess the probability that each player gets sent off based on their responses to the other team/referee and the referee's statistics, and sum from that. Naturally there are more factors involved in each part, I'm not denying it's very complex.

The type of data you require to do such a quantitative analysis is so extensive that there is no point doing it, because it is ridiculously subject to change.

I'm mostly talking about things which would require writing an algorithm, but nevertheless we can as humans draw some conclusions.

Then don't use something like a 'game in hand' as facts to back up your point when it's clear that a draw is a very unfavourable result for United even if the expected value you then used is gearing United for a draw.Well so do I. But individual factors means that it is very hard to quantify this stuff accurately.Betting companies e.g. utilise the actual gamblers flow of money to formulate their opinion on odds e.g. and while some more meaningful statistics would be useful, it is very hard to have an accurate conclusion on a teams chances of winning a game in the way you are saying.Yeah so basically you don't watch football or really have a care for but makes an opinion on it and predicting it.I'm out.

Opinions are fun. Saying 'did you see the game last night oh my god what was Jones doing' and everyone grunting in agreement is really, really dull. I don't care enough to inform myself to a greater extent about football, but it irks me when I see people who care about football completely ignore data.

Reply 6657

Dirtybit

OP

20

Original post by TheDefiniteArticle

We can assess the probability of each player being selected in both teams, assess the probability that each player gets sent off based on their responses to the other team/referee and the referee's statistics, and sum from that. Naturally there are more factors involved in each part, I'm not denying it's very complex.

First of all you don't need to create algorithms for this lmao. Second of all the only use of getting the actual probabilities would be for betting and that's all.

Reply 6658

TheDefiniteArticle

20

Original post by Dirtybit

First of all you don't need to create algorithms for this lmao. Second of all the only use of getting the actual probabilities would be for betting and that's all.

Well, I mean, each of these will be more nuanced (you'll probably need to do some Bayesian stuff to get a truer figure for a lot of these stats, and there will be other factors I haven't mentioned). The point of getting the probabilities will be making accurate predictions, and prediction seems to be what goes on most of the time ITT.