FM19 | Moneyball | Part 1

I start a Masters Degree in Data Analytics this September, and recently I’ve been looking at how I can use Analytics to impact my favourite game, Football Manager. Cleon recently tweeted about Moneyball, and how no one has replicated it correctly when it comes to FM.

Now, I had heard of Moneyball, and knew it involved using Analytics to scout players, but I didn’t know the true extent of Moneyball, or where the definition came from. So, I watched the film.

Seeing all those numbers and how they impacted the Oakland A’s made the maths nerd inside me very excited. I love analytics, and I love football and this felt like the perfect way to combine them using my favourite game.

So, what is Moneyball? Well the general idea is to use data and statistics, rather than a generic scouting network, to find undervalued or overlooked players who can be purchased for lower than their market value. Billy Beane was the first to utilise this technique to great effect in baseball as General Manager of Oakland Athletics.

In 2001 the A’s lost the Division Series to the New York Yankees (Football equivalent of play-offs (Kinda)) after winning the first two games of the five game series. They then lost their best three players to free agency. I’d compare this to Leicester’s title winning side losing the Premier League on the final day, then losing Mahrez, Vardy and Kante in the following transfer window.

The following season, due to budget cuts from the owner, the A’s had the lowest wage budget in the league. Billy Beane reasoned that if you had the lowest budget and you did the same as everyone else, you’d finish bottom of the league. And so, out of necessity, invention. Moneyball was born.

Billy Beane, along with Paul DePodesta, used Sabermetrics (the study of baseball using statistics) to source cheap replacements for the stars they had lost. Scott Hatteberg, a former pitcher for the Red Sox was brought in. He had ruptured a nerve in his elbow, and would never throw again but was signed because his on base percentage was very good. David Justice, a great player, but considered past his best, was brought in from the Yankees. The A’s were happy to overlook perceived flaws, trusting the statistics and believing in their principles. Hatteberg went on to hit the home run that would win them their 20th game in a row, an American League record.

So, can this translate to football? Well, it turns out, not so easily. Data is used a lot in scouting and performance management (more on that in the video below if you’re interested), but it can be quite difficult to execute, especially when using it to sign players. Football is a much more dynamic game than baseball. With roles so clearly defined (pitcher, first base etc), it’s easy to see which statistics are relevant. With football, it’s not so easy. However, I have a plan!

Implementing Moneyball in FM

A lot of people interpret Moneyball as buying low and selling high. Whilst you must be willing to let players go, Moneyball is more about finding undervalued statistics and players before anyone else. It’s about using statistics to find players who have more value than the market would suggest. It’s about exploiting market trends to get the most for your money.

So, how do I aim to implement this. I’ve recently moved from Uniao to Crystal Palace, who finished 17th last season (22/23) with one of the lowest wage budgets in the league. Palace seemed like the perfect place to implement Moneyball. With the lowest expectations, there can be no failure.

I have a few pointers to follow, based on the A’s and other sports teams who have been inspired by Moneyball & Soccernomics over the years, such as Southampton. The plan is:

  • Get Scout Reports from my Scouting Team
  • Separate players into shortlists for each position
  • Use Stats to whittle down Scouted players
  • Have one or two replacements lined up for each position

It’s important to mention that for each position I am looking at different Key Performance Indicators (KPIs). I will explain this in more detail later on when I discuss my tactics, but essentially the stats I am looking for are dependant on the player.

I will also be attempting to exploit market trends by signing:

  • Bosman Players
  • Transfer Listed Players
  • Players with low Release Clauses
  • Old / Injury Prone / Unproven Players
  • Players from Lower Leagues / Outside top five European Leagues

Key Performance Indicators

I’ve opted for a Structured 3-5-2 Formation, partially because I want to try it and partially because I can give each position a key role within the tactic. As such, I’ve split my players into 6 main positions.

  • Goalkeepers
  • Centre-Backs
  • Wing-Backs
  • Central Midfielders
  • Number 10
  • Strikers

Each position has different KPIs based on what I want each role to do. For example, my central midfielders need to be comfortable on the ball, able to win the ball back and recycle possession. Therefore, I want players with good passing, tackling, interceptions etc.

After deciding on 10 KPIs for each position, I sorted my already scouted players into separate Shortlists, one for each position. I’ve got a few months to scout players for each shortlist before the January transfer window hits. Hopefully by then I will have several options to choose from in each position, should I need players for my team.

I’ve set up a spreadsheet to record my KPIs so I can compare and analyse data on each player as we approach the January transfer window. We have to be willing to lose players, and when we do, we’ll be bringing in replacements following my rules.

It’s important to remember that this is not a profit making exercise. We are looking to exploit the market and look at the statistics to find undervalued players. We won’t be stockpiling under-23’s with the intention of selling them on at a later date without ever playing for the club. We’re aiming for the best functioning team for the lowest possible outlay.

Summer Signings

Before I decided to start this Moneyball save, I signed a few players following my rules without knowing. In future, I’ll be basing all signings on the numbers, rather than attributes.

  • Transfer Listed (Req)
  • Signed For: £7.5M
  • Market Value: £15M
  • Release Clause
  • Signed For: £6M
  • Market Value: £8M
  • Free Agent
  • Signed For: Free
  • Market Value: £24M
  • Lower League
  • Signed For: £17M
  • Market Value: £19.25M
  • Unproven / Lower Leagues (Played a handful of games in PL then sold to Rapid Wien)
  • Signed For: £11M
  • Market Value: £15M

In my next post, we’ll follow my progress through the January transfer window as I line up replacements for my squad, should they be poached by other teams in the league.

Moneyball Series:

Other articles you may enjoy:


23 thoughts on “FM19 | Moneyball | Part 1

  1. James says:

    A couple things –

    “Moneyball” is based on a system of statistical analysis that is highly reductive. Basically, what Sabermetrics does is attempt to eschew any statistics that cannot be reduced to what essentially amounts to 1 vs 1 outcomes. Specifically, strikeouts, walks, and home runs all boil down to pitcher vs hitter. Nearly all of Sabermetrics statistical metrics are derivatives of these duels.

    Football has no analogous statistical metrics, other than, perhaps, penalty kicks, which are not useful for extrapolation with respect to football in general.

    Where are your statistical comparisons? I can see you’re trying to find value in the transfer market by buying low, but, as you explicitly state, that is not moneyball. As stated in your piece, David Justice was a washed-up has been when the A’s signed him. But a statistical analysis of his hitting suggested he was worth more than the market suggested.

    Without any examples of statistical comparisons, your project is utterly opaque. Kalinic, for instance, played seven senior matches over five seasons before joining you. Wober played 20 in three seasons. No meaningful statistical analyses could have been done on those data sets.

    I’m not trying to give you a hard time, I just don’t see how statistical analysis, which is the keystone of the moneyball approach, is being implemented in any way whatsoever. It just looks like the standard opening for a FM save with a bottom half team.


  2. James says:

    Oops. I jumped the gun. I didn’t understand that the analysis would start during the season. A reread clarified.

    That makes complete sense.

    Sorry for the premature windbaggery.

    I’ll be tracking this project. I’m obviously interested.

    Liked by 1 person

  3. Eliseo Avramides says:

    I really like this idea. First, I’d recommend reading the book too, it is fantastic.

    I’ll be following this journey


  4. OlivierL says:

    Did u thought about doing this project with the ‘no attributes idea from Cleon / skin’ ?
    I’m thinking about starting one (no attributes and picking my team/squad and transfers with certain kpi’s.


    • fmvars says:

      I did, but chose not to simply because I want to see whether a player can play a position based on stats and not attributes. If I can’t see the players attributes, I won’t know whether he is well suited to the position or not. A player with bad attributes for a role who is doing really well in it would be very interesting to see


  5. Olivier Landman says:

    I wonder how u are going to update the kpi’s for a potential target when seasons pass. save every season in a different save/file so u get access to older stats?

    are u going to keep stats/kpi from your own players ?


  6. jellico73 says:

    I’m wondering if data imported into an R based dbase might be easier to work with. Excel is nice but it does have limitations when trying to show data graphically, especially some of the attribute related areas IMO.

    Looking forward to the next article!



    • fmvars says:

      Do you think that would be worth doing? I have no experience with R but I will be learning it soon so might as well get a head start.

      Thanks for the feedback, appreciate it


      • jellico73 says:

        I think so. Especially if your doing player comparisons across several categories, like this :

        I’m pretty sure that’s done in R, there’s actual data packages already made for such work if I read correctly. Excel is great for some stuff, but I think if your looking for a better visual representation of player data, R is the way to go. I could well be wrong though. 😃. I am going to take a look into it as well, be ause I have that block of free time between 2 and 4am open…


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s