First of all thanks for sharing your project, looks very cool!
I built something very similar for FM24 a few weeks ago, but I have EULA-concerns which is why I don't wanna release it. I vibecoded the UI, so don't criticize me on that, I hate frontend development and it's usable enough for me š It is also trimmed for my save with FC Salzburg, so it has some issues if you'd want to use it with other clubs, but they're easily fixable. I also have to fix language inconsistencies (English/German) the vibecoding produced.
I'm running a lightweight database in the background and thinking of switching to something that can save historic data (e.g. player attribute changes) without losing to much performance. I've set it up, so it auto detects if the in-game date changes. It then reloads the database and notifies me, if there are newgens that meet certain criteria (e.g. potential rating).
My rating is based on the attribute testing and @GeorgeFloydOverdosed's findings. It also uses the footedness for the relevant positions. (that's the A-Rating) Then there's the R-Rating which also takes position proficiency into account. I also have matchday ratings: 1) M factors in fatigue, consistency, morale etc. by giving penalties on the R rating (like testing showed) 2) Mi also factors in important matches by giving the M-rating further penalties 3) Mb is the maximum M rating a player could reach if fluctuating attributes (morale, fatigue, ...) were perfect
Then I added potential ratings. For the calculation I used age and the best training schedules found by @harvestgreen22. I then built a prediction model which tries to predict the maximum CA a player can hit and what his stats would look like with that training schedule. I also give out an achievability score from 0 to 1, as I use an optimistic approach. 1) P-rating assumes a player's position proficiency doesn't change 2) PĀ°-rating applies penalties for retraining players' positions. I added this, so I can see if it's worth it to retrain my prospects, because sometimes retraining e.g. a DM as a CB will give a massive hit to their attributes, as their RCA is higher than their CA. I built a CA calculator for that, which gets pretty much the same results as FMScout. 3) Pb-rating which is the Mb-rating but just based on P instead of R. But I'm not happy with that, as especially consistency can grow quite a bit for young players.
As I didn't use the setpiece attributes in the rating calculation I added extra setpiece ratings, but they're just my own assumptions, as there's little testing data on that.
I'm not fully happy with the ratings in general, as it almost always suggests to retrain strikers to wingers, as most newgens get higher rating there.
Also my best-XI calculator uses my hardcoded tactic and player positions.
But results have been very good over 6 seasons and I've found some low CA players that perform exceptionally well.
I'm also working on performance and consistency as some club/nation data points are all over the place in the RAM or I just haven't found the right pointers yet.
Maybe I have inspired you for some updates, I haven't looked at FM26 yet, but I'd assume it's quite a bit easier to get the data compared to FM24, took me 2 weeks to get consistent data š Here are some screenshots: dashboard
predicted best future player
filters, with advanced filters you can combine multiple things and use all saved data points
I built something very similar for FM24 a few weeks ago, but I have EULA-concerns which is why I don't wanna release it. I vibecoded the UI, so don't criticize me on that, I hate frontend development and it's usable enough for me š
It is also trimmed for my save with FC Salzburg, so it has some issues if you'd want to use it with other clubs, but they're easily fixable. I also have to fix language inconsistencies (English/German) the vibecoding produced.
I'm running a lightweight database in the background and thinking of switching to something that can save historic data (e.g. player attribute changes) without losing to much performance.
I've set it up, so it auto detects if the in-game date changes. It then reloads the database and notifies me, if there are newgens that meet certain criteria (e.g. potential rating).
My rating is based on the attribute testing and @GeorgeFloydOverdosed's findings. It also uses the footedness for the relevant positions. (that's the A-Rating)
Then there's the R-Rating which also takes position proficiency into account.
I also have matchday ratings:
1) M factors in fatigue, consistency, morale etc. by giving penalties on the R rating (like testing showed)
2) Mi also factors in important matches by giving the M-rating further penalties
3) Mb is the maximum M rating a player could reach if fluctuating attributes (morale, fatigue, ...) were perfect
Then I added potential ratings.
For the calculation I used age and the best training schedules found by @harvestgreen22. I then built a prediction model which tries to predict the maximum CA a player can hit and what his stats would look like with that training schedule. I also give out an achievability score from 0 to 1, as I use an optimistic approach.
1) P-rating assumes a player's position proficiency doesn't change
2) PĀ°-rating applies penalties for retraining players' positions. I added this, so I can see if it's worth it to retrain my prospects, because sometimes retraining e.g. a DM as a CB will give a massive hit to their attributes, as their RCA is higher than their CA. I built a CA calculator for that, which gets pretty much the same results as FMScout.
3) Pb-rating which is the Mb-rating but just based on P instead of R. But I'm not happy with that, as especially consistency can grow quite a bit for young players.
As I didn't use the setpiece attributes in the rating calculation I added extra setpiece ratings, but they're just my own assumptions, as there's little testing data on that.
I'm not fully happy with the ratings in general, as it almost always suggests to retrain strikers to wingers, as most newgens get higher rating there.
Also my best-XI calculator uses my hardcoded tactic and player positions.
But results have been very good over 6 seasons and I've found some low CA players that perform exceptionally well.
I'm also working on performance and consistency as some club/nation data points are all over the place in the RAM or I just haven't found the right pointers yet.
Maybe I have inspired you for some updates, I haven't looked at FM26 yet, but I'd assume it's quite a bit easier to get the data compared to FM24, took me 2 weeks to get consistent data š
Here are some screenshots:
dashboard
predicted best future player
filters, with advanced filters you can combine multiple things and use all saved data points
the best predicted players in my save
the best predicted (R-rating) XI of my club