Jump to content

Derby - Moneyball at it's finest


Recommended Posts

I've had a fascination with moneyball since reading it years ago (if you have not its a great read) ..

I have since always tried to translate this into FM as that is my "style" or "philosophy" if you will -

Just quick sample of me being proud of this seasons turnover:

dpbmva.png

4tai9x.png

72gak2.png

Link to post
Share on other sites

How did you apply Moneyball to football? What statistical analysis did you use to identify undervalued players? What statistics did you choose to ignore? How did you search for the players?

Or did you just buy cheap and sell high?

Link to post
Share on other sites

Hi Cavenagh -

Anyone in my squad can be had for the right price. I make sure not to get attached to a player (which surprisingly is not easy) and look at them as commodities.

To sell .. if I hear someone is interested in a player I will offer him to clubs for X amount ..usually twice their player value. Then I cut it back each time I offer 500k or so until I get a bid I am comfortable with.

This has the chance of unsettling the player so its risk/reward.

For searching/scouting I use the Searching the Stats - kpi method from : http://thehighertempopress.com/

I do not rely on attributes ... I look for players who perform .. The only attribute I kinda require in my tactic is work rate so I use that as the base line.

I am 7th after 27 games because it took time with my player turnover to get my tactic fluid.... We are on an 8 game unbeaten streak right now.

5W 3D

Link to post
Share on other sites

LOL - Bennet is a fairy striker who won't put in the work I require in my tactic upfront.

He B*tiched and whined and was upsetting team morale so I cut him loose ..

Will I pay for it in the future? Maybe

But for how I like my team to play he is useless to me...plus I could not get any freakin offers for him.

Link to post
Share on other sites

"Moneyball" doesn't really work with football for a lot of reasons, most of them to do with the very different rules governing salaries and transfers in baseball (and also the existence of the luxury tax, amateur and rule 5 drafts, etc.). Also, the object isn't to sell a lot of players and replace them cheaply; it's to narrow the gap between rich and poor clubs by identifying distortions in the market where some players are over- or undervalued.

What's your wage budget within the division?

Link to post
Share on other sites

I thought 'Moneyball' was more to do with attributes and stats, rather than buying cheap and selling high.

My approach is to analyze player attributes in an external spreadsheets. I have quite a few players whose star ratings are pretty low, but they have all the neccessary numbers to do a good job.

snip.jpg

Link to post
Share on other sites

Doesn't moneyball work so well in baseball because of the heavy reliance on standalone stats? So a player who is good at one thing can directly affect the outcome of a game. Whereas in football that's a lot more of a grey area.

I thought 'Moneyball' was more to do with attributes and stats, rather than buying cheap and selling high.

My approach is to analyze player attributes in an external spreadsheets. I have quite a few players whose star ratings are pretty low, but they have all the neccessary numbers to do a good job.

snip.jpg

The way I understood it, it kind of goes hand in hand. You analyse players using stats to pick out players who are underpriced. All about getting more bang for your buck, maximising what you get for your money using statistics. Or something like that.

Link to post
Share on other sites
Doesn't moneyball work so well in baseball because of the heavy reliance on standalone stats? So a player who is good at one thing can directly affect the outcome of a game. Whereas in football that's a lot more of a grey area.

The way I understood it, it kind of goes hand in hand. You analyse players using stats to pick out players who are underpriced. All about getting more bang for your buck, maximising what you get for your money using statistics. Or something like that.

It seems it all boils down to picking up good deals and making a competitive team despite weaker finances, as the very specific baseball style analysis isn't as useful in football. In FM it's quite easy, sell high, and pick up quality players from the biggest teams, as they usually over stock and don't play some excellent players. Leaving them transfer listed for really nice deals. The only real analysis that matters in FM is picking players with the right attributes for the roles to fit your tactics.

FC Midtjyllandare using some kind of moneyball/statistical analysis to power their team apparently, I heard they even send the manager data as the match goes on which he uses to adjust tactics and make subs.

Link to post
Share on other sites

Yes BoxtoBox you have it correct

When I took over Derby they were heavily in debt and over budget with 40mil due in loans and wasting away in the 18th spot.

1 seasons later we are top mid table fighting for a playoff spot and our finances are looking fantastic.

I expect next season to look for a true promotion push.

It's all about not getting attached to players, being very very patient and finding hidden gems for low money while selling any player you can for the highest return.

Current finance screen: (a little over 1 season ago Derby was about 40 million in debt, way over wage budget, and 0 transfer budget)

14kjy9j.png

Link to post
Share on other sites

OMFG

What an overstatement of a thread title. :thdn:

Was hoping to get some insight, but that's plain nothing :( Oh, and still trying to figure at which point in time there was a -40m balance (or are you really talking of debt, which wouldn't matter?)... -10m is the lowest I can see...

Well done on your transfer policy, that's what many people on here do.

Link to post
Share on other sites
Yes BoxtoBox you have it correct

When I took over Derby they were heavily in debt and over budget with 40mil due in loans and wasting away in the 18th spot.

1 seasons later we are top mid table fighting for a playoff spot and our finances are looking fantastic.

I expect next season to look for a true promotion push.

It's all about not getting attached to players, being very very patient and finding hidden gems for low money while selling any player you can for the highest return.

Current finance screen: (a little over 1 season ago Derby was about 40 million in debt, way over wage budget, and 0 transfer budget)

14kjy9j.png

On FM15, I took newly promoted Portsmouth in 24th place in the Championship, 8 points behind all teams in November. I guided them to 12th the first season and won the Championship the following season and then qualified for the Champions League my first season in the EPL by signing free agents and transfer listed players and loaning players ;) Derby is a team that should be expected to compete for promotion so 7th place is not so impressive and a rebuilding job wasn't exactly needed....7th place hardly justifies it either.

Link to post
Share on other sites

Indeed, this is more like buying bargains and and selling on... Not really seeing moneyball in action..

I think a more realistic approach would be to read up on midtjylland and brentford for inspiration, not the moneyball approach of baseball.

Link to post
Share on other sites
The basic idea of moneyball is finding inefficiencies (probably the purest expression of this in FM is playing whichever formation happens to break that year's match engine). If you're interested, check out The Extra Two Percent by Jonah Keri.

The concept of Moneyball is not simply "be frugal". Like Sunstrikuuu says it's about finding inefficiencies. Billy Beane didn't just go out and buy cheap, but serviceable players. He would scout a player often times for one specific attribute. He would find value in a player where others found none, most of the time emphasizing statistics he felt were most crucial, not whatever the conventional wisdom was.

For example, OBP (on base percentage) is not the flashiest of baseball stats, but it's often undervalued in favor of "impact" stats such as power numbers, contact hitting, athleticism, etc. Often times he would look for that sole statistic, a player's ability to simply get on base, above all others.

This would be hard to replicate in FM. There's much less emphasis on statistics in football and that's not counting when you get into baseball sabermetrics.

I agree that the best way to recreate this approach would be to find the most effective tactical system and scout only for that formation or style of play. Maybe you know that your center backs needn't have any pace, they could even be statues, as long as they have solid aerial presence and strength in the tackle. This would then lead to you buying cheaper (and slower players) maybe few others would consider. Again, it's tougher in FM because pretty much any way you slice it everyone is looking for the same attributes at every position.

Link to post
Share on other sites

Just to pile on here, but since a sizable chunk of a player's FM value is his CA and PA, I think a simple way to communicate the essence of "moneyball" into FM is to target players with relatively low CA and PA (and, therefore, cheaper) but with a good attribute spread for their defined role inside of a defined tactical system. i.e. look for those 2-3 star-rated players that end up performing above that rating.

Link to post
Share on other sites

IMO football will be evolving its statistical models to better reflect the value of individual players inside of their teams systems. Its a new science and will take time to develop. Stay tuned for further developments.

Link to post
Share on other sites
I think a more realistic approach would be to read up on midtjylland and brentford for inspiration, not the moneyball approach of baseball.

+1

My theory is that Matthew Benham's "football computer" is really the FM database... ;)

Link to post
Share on other sites

As others have said, the Moneyball approach focuses on more than just buying bargains and selling at a profit. It's about finding value in statistics that others do not value, or selling overvalued players who perform in a way that others value highly. E.g. Striker who scores a lot of goals, high average rating but doesn't tackle in a high-press style of football.

Beane also used to throw in players he didn't really like, but knew others did like, on games which didn't matter too much in order to boost the statistics that others found important and then trade them to clubs for players they didn't want, plus a draft pick or two. In FM that is quite possible, talented player in a role you'll never use, but you have a lot less throw-away matches in English leagues than somewhere like Brazil. Beane also isn't that much of a wheeler-dealer, he isn't a Harry Redknapp-type.

I think Moneyball can be used in FM, especially with the wealth of numbers available. But you need to find what the Ai managers value as being more important than anything else - average rating seems to be one thing, my players with a higher average rating quickly get scouted out so it seems they value that above (say) assists made. You could have a solid defender, with a crap average rating, because he doesn't go beyond the penalty box and intercepts more than he makes key tackles. He'll possibly be undervalued because of his lower average rating, but will suit your needs. He'd be a player for you to buy from the club for as little as possible.

Personally, Simon Kuper & Stefan Szymanski's Soccernomics book has a much more football-specific approach to statistic models but even they admit that, though, that while success is selling dear and having a positive net spend - this has little correlation with team performance.

Link to post
Share on other sites
Just to pile on here, but since a sizable chunk of a player's FM value is his CA and PA, I think a simple way to communicate the essence of "moneyball" into FM is to target players with relatively low CA and PA (and, therefore, cheaper) but with a good attribute spread for their defined role inside of a defined tactical system. i.e. look for those 2-3 star-rated players that end up performing above that rating.

This is more like it, low CA, but a good spread of attributes.. just the essentials for the role.

Link to post
Share on other sites
This is more like it, low CA, but a good spread of attributes.. just the essentials for the role.

Either that, or being able to locate the specific statistics that makes for perfect use in your specific system, say i.e. a player with a very high score in tackles won, which is what you need for him to do, but really lousy at shots on target, but that doesn't matter for his role (extreme examples, I know, but y'all get the point).

Saying that getting players cheap and selling them for a profit, and then naming the thread "moneyball at it's finest" really isn't the way to go..

Link to post
Share on other sites
IMO football will be evolving its statistical models to better reflect the value of individual players inside of their teams systems. Its a new science and will take time to develop. Stay tuned for further developments.

One of the most useful metrics they use in baseball now is WAR (win above replacement player). It's more of a cumulative stat, incorporating every aspect of the player's game. It produces a number of wins that the team has had with him on the field as opposed to a replacement player. It factors in hitting, fielding, baserunning, or pitching so is really a good overall view of the player's value to the team. Would be interesting to see WAR in football.

I wonder what a player like Messi's WAR would be? I'm guessing not terribly high since Barca seem to be able to play extremely well without him. Maybe when Luis Suarez on Liverpool would be a good example of a high football WAR since he was an absolute superstar on maybe not the best team overall.

Link to post
Share on other sites
Either that, or being able to locate the specific statistics that makes for perfect use in your specific system, say i.e. a player with a very high score in tackles won, which is what you need for him to do, but really lousy at shots on target, but that doesn't matter for his role (extreme examples, I know, but y'all get the point).

Saying that getting players cheap and selling them for a profit, and then naming the thread "moneyball at it's finest" really isn't the way to go..

That's what I mean... low enough CA to keep a players rep down and keep him off the radar as it were.. but attributes exactly where they need to be.

Link to post
Share on other sites
That's what I mean... low enough CA to keep a players rep down and keep him off the radar as it were.. but attributes exactly where they need to be.

Or perhaps strikers who have weak finishing, but score a lot due to pace and "likes to round keeper", probably not a great example as pace is always a hot commodity regardless.

Link to post
Share on other sites
What is this need for people to seemingly overcomplicating the game?

Sadly it goes on in football too (to a point), especially in England. Too many coaches/managers seem to follow the latest fad instead of making their own decisions.

Professional sports are big business. Maximizing your return on investment and finding market inefficiencies is smart business.

Link to post
Share on other sites
What is this need for people to seemingly overcomplicating the game?

Sadly it goes on in football too (to a point), especially in England. Too many coaches/managers seem to follow the latest fad instead of making their own decisions.

There is nothing "overcomplicated" in finding new ways to evaluate players. Stats will never surpass watching a player in a match, but sometimes a wide variety of statistics to choose from can show patterns in a player's game that a scout might not notice over a shorter span of scouting the player.

It's merely a tool, that's all. Some rely on them more, some rely on them less. I'll never understand why people are against the use of advanced statistics as a tool. Give me one way in which the use of statistics in evaluating players has changed football for the worse?

Link to post
Share on other sites
Sorry to burst your bubble but a lot of people are reporting crazy finances and transfer budgets from year 2 on. Its a bug. It's too easy to make money.

What part of it is a bug? the new deal for tv coming in or?

Link to post
Share on other sites
Either that, or being able to locate the specific statistics that makes for perfect use in your specific system, say i.e. a player with a very high score in tackles won, which is what you need for him to do, but really lousy at shots on target, but that doesn't matter for his role (extreme examples, I know, but y'all get the point).

Saying that getting players cheap and selling them for a profit, and then naming the thread "moneyball at it's finest" really isn't the way to go..

The main problem I find playing with "stats only" is that you have to turn every league's detail level you may want to scout to "All"; which can slow the game down to a crawl even with just the big 5 leagues going. If you don't do this, then the stats you see are all but complete fabrications.

Link to post
Share on other sites
What part of it is a bug? the new deal for tv coming in or?

I don't know why it happens, I'm not the developer. Just saying that a lot of people are reporting crazy finances and budgets in the first few years. My save at CP I spent 50 million on players in year 3 and my finances are still secure.

Link to post
Share on other sites
I don't know why it happens, I'm not the developer. Just saying that a lot of people are reporting crazy finances and budgets in the first few years. My save at CP I spent 50 million on players in year 3 and my finances are still secure.

Well, considering the new TV deal gives the bottom-placed Prem club around 100m for the season, the entire Prem should be financially secure. It would be rather unrealistic if Prem clubs weren't getting huge money once that deal kicks in.

If people managing in, say, The Netherlands, are getting massive budgets, that might be a bit more disconcerting.

Link to post
Share on other sites
I don't know why it happens, I'm not the developer. Just saying that a lot of people are reporting crazy finances and budgets in the first few years. My save at CP I spent 50 million on players in year 3 and my finances are still secure.

CP meaning Crystal Palace? with the new TV deal?

Link to post
Share on other sites
One of the most useful metrics they use in baseball now is WAR (win above replacement player). It's more of a cumulative stat, incorporating every aspect of the player's game. It produces a number of wins that the team has had with him on the field as opposed to a replacement player. It factors in hitting, fielding, baserunning, or pitching so is really a good overall view of the player's value to the team. Would be interesting to see WAR in football.

I wonder what a player like Messi's WAR would be? I'm guessing not terribly high since Barca seem to be able to play extremely well without him. Maybe when Luis Suarez on Liverpool would be a good example of a high football WAR since he was an absolute superstar on maybe not the best team overall.

There is something like that, it's an algorithm called "Goalimpact".

Link to post
Share on other sites
There is something like that, it's an algorithm called "Goalimpact".

Just read up on it and while an interesting stat it is quite different from WAR. Rather than combining lots of individual stats it starts from the "top down" and looks at overall results. I'd equate it more to basketball's Real Plus-Minus (RPM) where end results while the player is on the court are its only factor.

These are definitely useful. Say I'm managing Arsenal and I wonder what are our results with Coquelin on the pitch. I'd go to Goalimpact.

WAR doesn't use results while a given player is in the game, it's more a comparison of value between a low cost replacement player and the player in question.

IMO this is more detailed and would be fascinating to see the WAR of big name players. The numbers behind it are quite complicated, but if some "replacement player" standard were put in place (pass percentage 60%, tackles won percentage 50%, along those lines) I think it could certainly be done in football. Of course it'd have to use different metrics for defenders/midfielders/forwards and GKs.

Link to post
Share on other sites
The main problem I find playing with "stats only" is that you have to turn every league's detail level you may want to scout to "All"; which can slow the game down to a crawl even with just the big 5 leagues going. If you don't do this, then the stats you see are all but complete fabrications.

Completely correct! Also a lot of attributes hunters forget to include footedness in their calculations.

Link to post
Share on other sites

Having a good amount of down time at work today I've looked closer at the possibility of a WAR type statistic in football. I think it's doable and am wondering if there's any statisticians (or anyone really) who'd like to take a look at this possibility. Just PM me

Link to post
Share on other sites
Completely correct! Also a lot of attributes hunters forget to include footedness in their calculations.

Perhaps a moneyball-ish consideration. One footed players. Less useful because of their one footedness, but possible to build a system to exploit their skills and minimize their weaknesses. The advantage being, two 145 ca players, given one is Right Only, and the other Either Footed, the Right only will have more CA invested in other attributes.

Link to post
Share on other sites
Perhaps a moneyball-ish consideration. One footed players. Less useful because of their one footedness, but possible to build a system to exploit their skills and minimize their weaknesses. The advantage being, two 145 ca players, given one is Right Only, and the other Either Footed, the Right only will have more CA invested in other attributes.

Or put another way: one 145 ca player either footed or a 125 ca player right only. There is a good chance that the vast majority of the 20 point difference will be tied to his weak foot (it's a huge chunk of the CA calculation), and the either footed guy will be (everything else being equal) more expensive to buy. So it's the manager's call on whether or not the "market" (i.e. FM value calculation) is inefficiently assigning too much value for that skill and the manager can get at least equal performance out of the cheaper asset.

Link to post
Share on other sites
Or put another way: one 145 ca player either footed or a 125 ca player right only. There is a good chance that the vast majority of the 20 point difference will be tied to his weak foot (it's a huge chunk of the CA calculation), and the either footed guy will be (everything else being equal) more expensive to buy. So it's the manager's call on whether or not the "market" (i.e. FM value calculation) is inefficiently assigning too much value for that skill and the manager can get at least equal performance out of the cheaper asset.

That's a far better explanation of what I was trying to say, thank you.

Of course, we shouldn't know raw CA numbers, but going by ratings & value vs attributes & footedness you can usually see what you're looking for.

Link to post
Share on other sites
That's a far better explanation of what I was trying to say, thank you.

Of course, we shouldn't know raw CA numbers, but going by ratings & value vs attributes & footedness you can usually see what you're looking for.

Right. In practice, it'd be roughly the same as the either foot guy being rated 4 stars and the other guy being rated 3.

Having said all this: there is a likely chance that the market is efficient and, therefore, the value being assigned to the extra skills on his weak foot is demonstrated in the match engine.

Link to post
Share on other sites

This would be hard to replicate in FM. There's much less emphasis on statistics in football and that's not counting when you get into baseball sabermetrics.

The fact there's less emphasis on stats at the moment makes moneyballall the more profitable if you can pull it off. Moneyball has gotten harder in baseball because everyone is using stats now. Because stat usage in soccer is so primitive by comparison there's likely massive inefficiencies to exploit still that have already been exploited in baseball.

Link to post
Share on other sites
The fact there's less emphasis on stats at the moment makes moneyballall the more profitable if you can pull it off. Moneyball has gotten harder in baseball because everyone is using stats now. Because stat usage in soccer is so primitive by comparison there's likely massive inefficiencies to exploit still that have already been exploited in baseball.

I totally agree. The same civil war that still rages in baseball between traditionalists and sabermaticheads is coming to a sport near you.

Link to post
Share on other sites
The fact there's less emphasis on stats at the moment makes moneyballall the more profitable if you can pull it off. Moneyball has gotten harder in baseball because everyone is using stats now. Because stat usage in soccer is so primitive by comparison there's likely massive inefficiencies to exploit still that have already been exploited in baseball.

two games are just so fundamentaly different it will never work. why do you think stats in football are still basic while majority of other sports it is so developed? football is just too unpredictable, too much driven by random events, uncontrollable... that stats tell you very little.

baseball is very static game compared to football and major part of it is one v one battles easily bracketed by the stats. football is everything else.

Link to post
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...