Sabermetrics applying to Soccer

[Oliver Anderson]

That is a great point that is lost in all this how Robinson remains Tottenham number 1.

Our biggest critism of Martin Jol is his blind playing of underperforming players the last few season becuase like any sport it can seriously effect a team. His continuing playing of Robinson, Assou-Ekotto, Hossam Ghaly, Taino etc cost tottenham over six goals in terms of their goal difference last season.

From our studies a quick and dirty measure is that 1 goal is worth approximately 1 point - therefore Tottenham lost 6-7 points last year from playing players performing below the league-average standard for their position. Had Jol simily played average performers or switched to better or other options soon instead of continuing to play poorly performing players Tottenham might have finished higher in the table.

As for Lehamann and other goals keepers all players that played in the league last year are analysed and assessed with our measures in the book - as well as the teams. Also all players have their stats predicted for the upcoming season from our player projection model (SIMprofiler).

In terms of Lehmann he has started the season making some big errors. Last year he was 10th in our stop percentage rating and had a percentage about 3.5% above-average. So not the best, but still a respectable keeper.

Post it on the site and spread the word if you like.

[The Jitty Slitter]

That is a great point that is lost in all this how Robinson remains Tottenham number 1.

Our biggest critism of Martin Jol is his blind playing of underperforming players the last few season becuase like any sport it can seriously effect a team. His continuing playing of Robinson, Assou-Ekotto, Hossam Ghaly, Taino etc cost tottenham over six goals in terms of their goal difference last season.

From our studies a quick and dirty measure is that 1 goal is worth approximately 1 point - therefore Tottenham lost 6-7 points last year from playing players performing below the league-average standard for their position. Had Jol simily played average performers or switched to better or other options soon instead of continuing to play poorly performing players Tottenham might have finished higher in the table.

As for Lehamann and other goals keepers all players that played in the league last year are analysed and assessed with our measures in the book - as well as the teams. Also all players have their stats predicted for the upcoming season from our player projection model (SIMprofiler).

In terms of Lehmann he has started the season making some big errors. Last year he was 10th in our stop percentage rating and had a percentage about 3.5% above-average. So not the best, but still a respectable keeper.

Post it on the site and spread the word if you like.

I am going to buy your book.

[Oliver Anderson]

Just to let everyone know I have posted the up-to-date Premier League player stats on the site and there are a few more articles on there now including a look at Sir Alex Ferguson and Rafa Benitez's rotation policies.

[revelationx]

Just to let everyone know I have posted the up-to-date Premier League player stats on the site and there are a few more articles on there now including a look at Sir Alex Ferguson and Rafa Benitez's rotation policies.

Another fantastic article, Oliver.

My rotation watch thread in the Liverpool forum is showing Prem lineup changes as the season progresses to show that all the big teams rotate and not just Rafa. For some reason the media seem obsessed with Rafa and his changes while completely overlook or ignore the other teams when they do it.

[The Jitty Slitter]

I ordered your book but it's not arrived yet!

[Oliver Anderson]

Amazon or off the site?

Amazon are being a bit slow at the mo with their distribution. Loads of new books apparently.

They'll get it to you soon.

[The Jitty Slitter]

off the site using paypal on the 5th of sept?

[Baysider]

Another fantastic article, Oliver.

My rotation watch thread in the Liverpool forum is showing Prem lineup changes as the season progresses to show that all the big teams rotate and not just Rafa. For some reason the media seem obsessed with Rafa and his changes while completely overlook or ignore the other teams when they do it.

Yeah, I noticed that the top 4 teams were all in the top 5 in terms of rotations. I could think of two explanations for that. One is that they play more games and so they have to rotate. The other is that their reserves are of higher quality so that the cost of rotating is less.

[Oliver Anderson]

Hi Jitty Slitter,

I've asked the guys from the site that send out the books and they assure me that they sent your copy out on 6th September.

So it should be with you soon

[Oliver Anderson]

Hi baysider,

We did loads of research into rotation last year and found that rotation is squad specific. Your reasons for the big teams are spot on. Thay have to rotate if they are to compete on all fronts.

One of my favourite stats on the subject was that over 2 years (2004/05 and 2005/06) Liverpool played almost a full Premiership worth of games (34 games) more than Tottenham and Tottenham would arguably be the first team included in a Premier League big-five after Chelsea, Liverpool, Man United and Arsenal.

[tk421]

Oliver,

How do you collect your stats, especially for stuff like assists? Do you have people go through each game manually? Or do you take the stat lines somehow?

Do you see your methods being able to be applied to the MLS? Or better yet leagues that are less mined for talent?

[Oliver Anderson]

Hi there tk421,

The answer to your last two questions is a definite yes.
The methods could quite easily be applied to other leagues and we have been doing some work in that area this year. There are a lot of leagues out there that are unmined and these types of methods definitely help the mining process.

The vast majority of data is charted by my self, family, friends and colleagues to specifications we set up. The real thing we are trying to do though is present the data properly and not just in standard total or average form

[muppet77]

hi there Oliver. I ordered your book the other day and got it this morning.
what a cracking read! well done.

how do you adjust the points estimate for each team?
you say it relates to the value of opposition and the venue the game is played at?

[Presario]

I got his book too the other day and it is the best footie book I've read in a while.

My favourite bit was the age profiling stuff that says players peak in the EPL earlier than most people think.


As for your question muppet77, I think he adjusts the estimated points by opposition and venue using his estimated points model and adjusts them all to a level playing field. So if a team plays Chelsea away that might be a 10% harder fixture because its Chelsea and a further 5% harder on top of that because its away.

So if the model said a team deserved 2 points in the above contest adjusting it because it was a 15% harder fixture than average would make the adjusted figure 2.3 points.

Thats how I understood it. He is probably easier to contact via the e-mail address on his site if you want more info.

[JoeYorkshire86]

Evening all, nothing like arriving at the party four years late.

For Christmas I received Moneyball, which I found to be a great read. It followed on from reading Clive Woodward's book that looked at stats and the like. My overall feeling was that statistics will never (and should never) replace judgement, but that it's important to be able to back up decisions with evidence. Received wisdom is a very dangerous thing. The On-base % discovery was very interesting to me for example, but then I was disappointed to read Billy Beane asking for players over the phone who he'd only read about minutes earlier.

That prompted me to buy Oliver Anderson's book as advertised here. Clearly there are associations with Bill James' early work here ;)

I find the book interesting though it's mostly a reference guide than anything. In all honesty, I found the introductory information and the concluding essays the most interesting. If I were to level some criticism at the book, it would be that the player profiles don't relate too heavily to the stats. Oliver does a fantastic job with the data, but then to talk about a player like anyone down the pub is to do himself a dis-service.

However, he promises it'll be an annual tome and it is sure to evolve. I think there are many people such as myself who know little to nothing about stats (GCSE Maths for me, cheers), but I like to find answers to the unanswerable.

I also think it's great there are so many knowledgable folk in this thread, but sad that it has died a little.

I would suggest that the folks here develop a little project for the group to work on, to see if we can at least go some way to answering one of the many questions. Collaboration seems key in this day and age.

Keep up the good work guys and let's hope this thread gets a revival of sorts.

[Presario]

Woh Joe,

i think you are doing the book a dis-service. I went back and read over the player comments for six teams last night and not a single one didn't reference some stat either in the table or another he collects.

I think you've misunderstood something when he says above or below average he is not talking about his opinion he is talking about what the stats show above. It is far from 'pub' talk it is what the stats say.

I work with community stats every day and this is how we prepare our stat reports - reference those shown and add some soft data. It is exactly what is discussed in Moneyball - add cold hard stats to scouting opinion and you have something twice as powerful.

Read any baseball statistical scounting guide after Moneyball and this is exactly what they do. This is the first I've seen for soccer and its a great initial stab.

As for reviving this thread your spot on.

[TimB4Last]

I knew there had to be a BS thread ...

Probably deserves (and surely will get) its own thread, but I want to park an article here because it suggests a very specific idea regarding player evaluation. I joked about the SPARQ announcement earlier in the thread, but it seems certain that the push for precise statistical information will continue to grow.

http://football.guardian.co.uk/News_Story/0,,2274932,00.html

....

The main difficulty the pair have faced is that while a sport such as baseball can be easily broken down into individual events - a pitch, a hit, a catch - football is a much more fluid game, which makes it more difficult to isolate cause and effect. Nevertheless, {Billy} Beane {yes, that Billy Beane} seems determined. "Say you have an approach to signing players and you get it right 30% of the time and then you can discover a different approach that allows you to do the right thing 35% of the time, you have then created a 5% arbitrage. I don't know if we can do that for soccer but I want to find out."

....

"You don't need statistics to spot the real great players or the really bad ones. The trick is to take the players between those two extremes and identify which are the best ones," said the Match Analysis company president, Mark Brunkhart, a fervent believer that the methods Beane popularised in baseball will soon be widespread in football. ...
....

Four areas where Sabermetrics scientifically identify the best footballers

1. Number of touches

A measurement of how often a player is involved on the ball

What it reveals Player's fitness level, the number of times he gets into a position to receive the ball and team-mates' willingness to pass to him

2. Shot creation

The number of times a player participates in a possession leading to a shot (both on target and off)

Reveals The attacking effectiveness of a player, especially attacking midfielders and forwards. Measures ability to balance ball retention with creating scoring chances

3. Ability to retain the ball

A measurement of the probability that the next player who touches the ball will be a member of your team

Reveals Contribution of players who are less directly involved in attack

4. Balls won per 90 minutes

Measures defensive effectiveness

Reveals Attacking players' willingness to defend; defenders' ability to tackle, intercept passes and position themselves well

[Ed-D]

so how can I get these stats for MLS?

[tripvm]

Has anyone heard if Oliver is going to produce his Football Review book again this year?

[JoeYorkshire86]

Has anyone heard if Oliver is going to produce his Football Review book again this year?

Yes, and the answer is not right now. Without wanting to speak for him he has been busy and therefore not done a new book, but may do in the future.

Whilst I'm here, I thought I'd make a suggestion for anyone interested in some collaboration.

A useful two features of Google Docs spreadsheets are:
1) for several people to collaborate on one document
2) the ability to "import" tables and data from websites (HTML & XML) and RSS feeds

What this means is that it could be possible to automatically combine all the disparate sources of football statistics into one single spreadsheet, which can be used as people see fit. It can be self-updating and exported into different formats beyond a spreadsheet file.

Anyone interested?