Digging Through Data – The Moneyball Approach

0
128
Tempo Tactics

We love Football Manager because it provides limitless access to almost every professional and even semi-professional footballer on the planet. We can get a snapshot of their current attributes, their historical stats such as goals and appearances, and we can predict their future ability. No other database (that you can get for between £20-30 on Steam), provides this level of detail.

But for all its beauty, FM is still catching up to the data revolution which is taking over football. It has collaborated with STATS and various other companies to assist with its game analysis features. These are useful, but it can be hard to extrapolate the number of shots from various angles in a single game, to measure a team/players performance over the course of a season.

Most frustratingly, once a season has finished, in-depth player stats such as their passing success, dribbles per game and yellow cards are consigned to individual player histories, and can’t be viewed as part of a shortlist, for example. This can make it harder to compare players effectively, and may result in you signing a player, when a much more capable option has been left undiscovered.

To try and get a hold on this, I will be running through a brief series where I use real data analytics to help me sign young, talented players, and try and build a successful team.

Step 1: Decide my Tactics

The first team for something like this has to be to get a tactical system in place. As some clubs in real life are realising, recruitment without a clear strategy leads to giving long term deals to over 30’s and allowing David Luiz to remain at your club at all.

To keep things simple, I will be recruiting players for a 4-1-2-3, an easy to use Tiki-Taka tactic that generally provides a good balance between defence and attack.

Step 2: Create my Shortlist

So, the basis of the save is that I remain unemployed throughout the first season, to allow players to build up some in-game statistics that I can then analyse. Once the season is completed, I find a club and start signing the most effective players, according to the data.

My aim was to take a lower-end Premier League job at the end of the season, then signing younger players from the Championship. As a result, I created a shortlist for each position in my tactical plan, and filled them with the top 20 most valuable players under-25.

Step 3: Find a Club

Following their tenth placed finish, Leicester were in need of a manager, and came to me (I was set as a former international footballer etc just to make this step easier).

£30m in the budget is solid, but nowhere near enough. First job at the new club was to begin the offload process, offering out most of the top players. For the purpose of this save, I want to exclusively use the players I signed, where possible, so Tielemans, Praet , Perez and co were not needed.

Step 4 and Beyond: Analyse Players and Make the Signings

So, here we are. We have our club, money and shortlist, time for some analysis.

Starting with goalkeepers, naturally, creating data visualisations immediately outlines the players who are most interest. The plot below shows players’ “goals conceded per 90” against the total number of clean sheets.

Those at the top of the plot have the most clean sheets, but as is shown by the coloured key, those are also the players with the most appearances made. This is why I have a lot of interest in Meslier, the Leeds loanee from Lorient. Despite making relatively few appearances (17 in all comps), he kept 9 clean sheets, an excellent return.

This then also manifests itself in the form of goals conceded per 90, where Meslier is the only player to have conceded fewer goals than games played.

The below graph then indicates the other players that are of interest to me in goal. George Long of Hull was one of two players to achieve 16 clean sheets, but he did so while needing to make many more saves. This suggests he’s a better stopper than his counter-part, David Raya of Brentford (of course Brentford would get a mention here), so may also be worth a bid.

I have made approaches for both Meslier and Long, as both would command a transfer fee of less than £6m. Raya would cost me upwards of £10m, which, when looking at the stats suggests I would be overpaying for his value.

In deciding between Meslier and Long (can’t be getting both), this is where the modern side of football recruitment kicks in. Any recruitment based purely on scouting/player style will not work. Similarly, recruitment based purely on data will not work. It’s the combination of the two that yields most positive results.

As a result, I will be prioritising Long over Meslier, as Long has a specific skillset that Meslier lacks: kicking. As my keeper will be a sweeper, football ability is key, and that may tip my decision towards him. While Meslier may have a much lower conceding rate, I need my keeper to do more than just stop shots.

The next article will feature our recruitment of new defenders and their data, as we continue to build a low-value, high quality team.