Part I: Does Fibra correlate with player performance data?
We're simulating a year in FM22 in order to take a look at the data from the top five loaded leagues around Europe (that's England, France, Germany, Italy and Spain). Let's be frank about this, a correlation should absolutely exist. If not, what's the point in attributes? Admittedly, there could be my misunderstanding about the effectiveness and/or legitimacy of certain attributes within things 'under the hood' of the match engine. There is also the chance that those warriors are being set out by the AI and held back from fighting their way to success in less aggressive tactical instructions. So, a simulation isn't a deciding factor in determining whether or not fibra correlates with player performance data…but it's a good start.
Simulating a season will give me a partial answer. From there, I'm going to look at taking on management of a team that scores highly in fibra and further instill more of it into their side via player recruitment. I'll also be playing a system which, in my view, will suit the players and I'd be hoping to see fibra 'perform' in the repeated exercise of drawing out the numbers from Football Manager after season two.
So, to recap:
Sim a season (2021/22) of European football. See if players' fibra attributes scores correlate with player performance. Pick a side to manage for Season 2.
Manage a top-level club for the 2022/23 season, recruit or utilise existing fibra and shape the team to make use of it.
Let's have a look by first defining what makes my fibra attribute score (FAS). I'm keeping it simple, and purely mentally focussed*:
Aggression - a highly aggressive player is more likely to tackle hard and leave his mark on a player.
Bravery - a brave player is more willing to put his body on the line and engage in physical duels.
Determination - a highly determined player is more likely to help the team fight back from losing positions or improve his own poor performance during matches.
Teamwork - a player with a high teamwork attribute will follow tactical instructions and complement the attacking/defensive units of the team.
Work Rate - a player with a high work rate will exert more of his physical capability during a match.
Five attributes meaning I am rating players out of 100, therefore giving attributes equal weighting…simply because I don't know the exactness of what makes the most difference in the match engine**. Oddly enough, I don't even want to know…I think I like the mystery around Football Manager and it certainly keeps me coming back each year.
With FAS I am able to see if a higher score correlates with a number of player metrics that could be crucial to my tactical approach, particularly towards the defensive side of the game (although winning headers can definitely be a route to goal for us too):
Yellow Cards per Tackle
Tackles Won Per 90
HDRS A - Headers Attempted
AER A/90 - Aerial Challenge Attempts per 90
HDRS - Headers Won
HDRS W/90 - Headers Won per 90
HDR % - Headers Won ratio
K TCK - Key Tackles
TCK - Tackles per game
TCK R - Tackle Completion Ratio
ITC - Overall number of interceptions
INT/90 - Interceptions per 90
Distance Covered per 90
*Disclaimer: Some/All of these metrics will be influenced by other technical/mental and physical attributes e.g. Interceptions affected more by Anticipation, Concentration and Positioning. But I at least want to see if there is a FAS correlation.
**Another disclaimer is that you could just as easily build a tactic/style first and then evaluate which attributes are making the most impact on certain metrics, using regression analysis. The best ever example I have seen of this in FM circles is when FM Tahiti applied it on his nine (!) years of player data when playing a 442. What I am getting at here is that my study is still very much basic/and from a top-level with pre-defined assumptions of my fibra attributes. Certainly no 9 years of player data performance here; although maybe that’s one for the future 😃
Season 2021/22
So, I loaded the leagues on save creation and went on a one-year holiday meaning the 2021/22 processed in the Full Match Engine (the engine that would have been used if a human player was playing in). As mentioned, we have the top 5 leagues in Football Manager loaded (as rated by FM). Only two of the 2021/22 league winners match reality: PSG in France and Bayern in Germany. Diverging from reality we saw titles wins from Barcelona in Spain, Liverpool in England and Juventus in Italy. A mirrored Champions League Final took place though, only for Liverpool to be crowned the winners in this alternate reality, beating Madrid 3-1 after extra-time. YNWA.
But enough of alternate timelines, how does fibra correlate to the in-game statistics mentioned a moment ago? At season's end, I exported every player from the top leagues who had played 1,000 minutes or more. This equates to circa 1,500 players. I first eliminated the hundred or so Goalkeepers that had played over 1,000 minutes (their role on the pitch is so vastly different that I did not want them to distort the analysis). That being said, fibra is probably an underrated quality in GK…especially in terms of Aggression and Bravery in terms of one vs ones or aerial battles. But let’s save that analysis for a rainy day.
With a revised count of 1,350 players I was able to total up their FAS and then compare that score alongside the different player metrics using a scatter plot. I have already mentioned the many disclaimers with this kind of high-level study, but some observations can be taken:
Minutes Played – A slightly trivial one to start with, but there was a positive correlation of FAS and minutes played throughout the season. There could be many reasons for this, and fibra could be just the tip of the iceberg in terms of AI decision making when fielding an XI…but the higher the fibra, the more minutes you could expect to play (generally).
Tackles won per 90 – there was a slight positive correlation. An aggressive and brave player is probably going to win the 50:50s against less aggressive players, this could be the explanation.
Distance per 90 – eeekkk a negative correlation. At first this surprised me, but then I thought more about it. An explanation for this is that a significant number of fibra players are central defenders (20% of players with a FAS score were natural at CB). This is an outfield position that covers less distance per game, so this would bring any positive correlation down.
What’s interesting is that a positive correlation exists when you exclude all central defenders and look at the remaining 75% of outfield players (as shown below).
Headers per 90 – positive correlation. Bravery will influence the player going for aerial duels, this makes sense to me. Also, many of the high scoring FAS players are more than likely to be used in central roles; the likelihood of engaging in aerial duels (and winning headers per 90) from those central areas of the pitch is significantly increased.
Interceptions per 90 – slight positive correlation, but not enough to certainly say FAS has a huge influence. This supports my preconceived opinion that interceptions are more likely to be influenced by the likes of Anticipation, Concentration and Positioning. Again, further study needs to be done here.