What's new

Damien Comolli on 'moneyball' in football

JerryGarcia

Dark star crashes...
May 18, 2006
8,694
16,028

cjsimba

Well-Known Member
Dec 5, 2006
2,623
9,599
Dempsey is our classic 'moneyball' player/signing. Not the most talented player in terms of technical ability but he does regularly score important goals. And goals win games.
 

We'llWinThePrem

Well-Known Member
Oct 28, 2010
2,853
4,821
I remember some ITK (JJ?) mentioning that we have people that study stats and one of the reasons we wanted Remy? last year was because he scored the most headed goals in Europe.
 

JerryGarcia

Dark star crashes...
May 18, 2006
8,694
16,028
Yeah good call on Dempsey, he fits the bill on that. I don't know if he's a player we were following for a while or not?
 

wozzahotspurs

Active Member
Jan 19, 2007
585
335
Moneyball doesnt always work! a great example of this is Stuart Downing.............

Create the most chances in the league. Liverpool buy him & he got what 0 assists 1 goal in 38 games or something like that.
 

michaelden

Knight of the Fat Fanny
Aug 13, 2004
26,418
21,763
Moneyball doesnt always work! a great example of this is Stuart Downing.............

Create the most chances in the league. Liverpool buy him & he got what 0 assists 1 goal in 38 games or something like that.


i thought it was 0 and 0 in 38 games
 

sloth

Well-Known Member
Mar 7, 2005
9,018
6,900
Just found this on newsnow and thought it was worth a thread. Comolli talks a bit about his time at Spurs and how we used statistical analysis when we were building the squad. I don't know what Franco Baldini does in terms of stats, I've not heard anything specific mentioned, though we don't seem to be targeting moneyball type players anymore either, maybe we've moved past that stage.


I have a sports quant, which means we use statistics to identify in-running market inefficiencies and bet accordingly.

I'm not a statistician, but judging from that interview I think the Commolli's completely misunderstood what can and cannot be said about a player or a team using statistics.

The first thing to say is that were you to put confidence bands on statistical analysis of the typical numbers of player interventions in a season which you think are important (e.g. completed passes in the opposition third) I think they'd be wide, meaning it would be difficult to know whether what you thought you were seeing was significant.

Second, to try and normalise those statistics such that the quality of the team, the make-up of the team, the tactics, formations, coaching, time together etc are factored out, is almost certainly impossible given the limited size of the data-set. The difference in sports like Baseball or like Tennis is that you're essentially mainly measuring a duel between two individuals (batter and pitcher in Baseball), and this extraneous stuff can be more easily ignored.

Third, as a non-statistician with several years of working in the field I can tell you firstly how easy it is to get led astray by what you think is a strong signal but which nine times out of ten turns out to be noise, and secondly how easily statisticians get led astray too; we think of them as scientists, but they often don't have huge insight into what they're analysing, they also want to prove their worth, and so they also often make leaps which cannot be stood-up by the facts.

Fourth, in my game there's no hiding place if you're wrong, if the model's no good you'll lose money and keep on losing money, however in Commolli's world it's much more about humans and within the human decision making process at a club it's very easy to hide the truth from one's self and therefore from others. This means that there's no easy, straight-forward way to learn from errors, to understand when you've got it right or wrong, and which bit you got right or wrong.

All of which means that whereas I'm sure Commolli would like to align what he does with science, in fact what he's practising is a kind of medieval sports alchemy, with all the hocus pocus bullshit and credibility that goes with it!
 

JerryGarcia

Dark star crashes...
May 18, 2006
8,694
16,028
I think it's fair to say Comolli had mixed results during his time at the club, it's also difficult to pinpoint exactly which deals were his and his alone. It would be interesting to hear from our resident Liverpool fan to see if DC left any kind of legacy behind there. We have a better squad than we did before he came here but Harry had to rebalance it when he took over.

Good post sloth, I can see why it would be easy to think you've found some hidden gem. Especially when he can "cross like Beckam".
 

stonecolddeanaustin

Well-Known Member
Aug 27, 2011
1,631
2,607
This moneyball thing stemmed from baseball didn't it. I can easily see how it could apply to that game because to form a winning team is a fairly simple process. You need pitchers who can strike out batsmen, batters who can get to first base or score home runs and fielders that can field. If you have all that then you'll have a successful team.

There's a lot more to a football team than the sum of its parts. Darren Bent was a classic moneyball signing in that he was probably signed for the sole reason that he scored goals. If anyone had actually watched him play properly they'd realise that he was and still is a very unintelligent footballer that can't hold up the ball, has a severe lack of vision and can in fact do nothing other than score goals. We all realised that after watching him play for about 4 games and Villa seemed to realise that too now. A player like that was never going to fit into our team at that time.

The way Dortmund smashed both Real Madrid and Man City in the Champions League this year is a perfect example of how moneyball cannot be applied to football.

Bugger, just seen that Sloth has pretty much explained all this in a much more coherent and thorough way. Well I've come this far so I'm posting anyway.
 
Top