s0rn said: hey, think I found a bug in with the FM26 outfield defaults. After uploading my CSV, only the GKs are getting actual scores, every outfield position (ST, AM, M, D/WB, etc.) is sitting at 0.00. Went to check Attribute Settings → FM26 → Outfield and every attribute across Technical, Mental and Physical is set to 0 points. That explains the 0.00 scores Expand
Was a strange one. It happened with the deployed version, but not local. Should be fixed now. Can you check again?
I updated "Getting Started" with the scoring changes. A tl;dr is on the Change Log page. For those that haven't yet, kind reminder to re-download the Player Database view.
WesleySantiago said: I'm using the new view (Fm 26), but nothing has changed. The positions appear normally in the data lab, but only six players are shown in the tactics setup. I also tried adding them manually, but there's no option.
That upload was very helpful, it helped me find a lot of bugs. The benefit of using real data vs synthetic data
The interim formula also seems to not be as OK as I had expected, had to redo the aggregation and all that. It wasn't wrong per se, but it was causing the Core4 Attributes (Acc, Pac, Jum, Dri) to have an even bigger impact on the final score than they already have.
So the overall scores might be a bit lower than before, and will favor all-rounders more.
I haven't fixed all the bugs yet, but I should have a release available this weekend.
Hi mate, what did you change compared to previously? Would be good to know! Thanks for all your work Expand
There's new attribute weights for FM26, which are derived with formulas from the new research data. And ofc toggles to move between them.
The points weights should be fine, but the goalsFor & goalsAgainst numbers were a bit iffy to determine. Which is why they are currently disabled in the calculations. So you may see them in the UI, but they do nothing.
This is all placeholder for now, while I work on a better way to extract the data from the excel and come up with a new formula.
WesleySantiago said: The program loads only some positions in the tactics section (team builder) and leaves others blank, even though all players have been loaded. Expand
FM24 or FM26? Squad or Player Database? By any chance, in the Data Analysis tab, do you see Position "Unknown" for some players?
I ask, because I updated the view in FM26 yesterday. The old "Player Database" view, used the column "Best Position", instead of "Position", so it wasn't converted correctly.
If so, you can either redownload the view from the page, or manually right-click & replace the column "Melhor Pos" with "Informações Gerais" -> "Posição"
Cheers! I uploaded an update with the FM26 weights. I can't vouch for correctness of the attributes, but for as far as I tested them in my save, they match expectations.
GeorgeFloydOverdosed said: From what I've read, the game's outcome is essentially pre-determined the moment you enter the match with your selected players, but that the outcome will be recalculated when you make substitutions. This still won't sway the outcome much, because each substitution is at best a ~10% difference of 9% of what's on the field. If you're going in with the right tactic already set, the rest doesn't seem to matter.
I think what is most beneficial is non-rotation tendency. You'll have noticed that AI managers rarely, if ever, rotate their players. They keep running them even if they're exhausted, and it is perhaps for the two following reasons. First, low match fitness is significantly more detrimental than low condition, and almost impossible to keep atop of if a player is not playing almost every game of the season. Second, the high morale produced by a run of good form seems to moderately increase the chance of winning the next game, but this also snowballs over the season. Morale appears to return to a neutral baseline fairly quickly, so rotation interrupts and prevents good form from occurring.
Additionally, it's been deduced that giving young players a handful of matches each season or subbing them on in the dying minutes does little to nothing for their development. They need at least ~15 full matches per season at a minimum for development, otherwise they're better off playing friendlies in the reserves.
I realized that at this point, no assistant manager attribute is essential, and so I put this theory to the test. FM24, Manchester City, with a top preset tactic and set routines. Assistant manager was assigned all relevant roles, including setting training.
To reduce the effect of the player variability through the season, and also see if a different effect was seen in an inferior/underdog team, I tested using my 1 CA players Man City team. I created a perfect 200 PA assistant manager with 20 in almost all attributes, and a 1 PA one with the opposite (I even gave them 1/10 English proficiency). Same tactics were used. Simmed only to midseason, since I was running into problems with sacking after that point.
The difference between a perfect assistant manager and an abysmal one is statistically insignificant as to be indistinguishable from random chance variation of results. It may not be just a coincidence that the superior assistant manager does slightly better in both tests, but even if so, the difference between two realistic assistant manager options (stats of 12 vs 16 instead of 1 vs 20) would be minuscule.
Seems to me that one should choose an assistant manager who is a good coach or cheap. Expand
I did notice that it's better to sub players out based on their current match-rating than based on condition.
6.2 players seldom recover to get an 8.0 Replacing these bums with a fresh player that starts at 6.7 sometimes ends in a "super-sub" situation.
It's almost as if the the game rolls some dice for each player during the match on whether they will have a good match or a bad match, and then it spends the rest of the game working towards the outcome of the dice.
I also noticed that when I use Instant Result and sometimes check the player ratings at the end, I find that the the assistant did not sub out a player, and they end up with a 4.8 or other crazy-low rating.
If this hunch were correct (and I don't know how to prove that it is not), then I still wouldn't know whether a single sub rerolls the entire match outcome, or just decides the fate of this new sub. I'd lean on the latter. In either case, the aggregate result might tilt.
But even so... how does the game deal with: "oh, you were supposed to lose 1:2, but now you're supposed to win 3:2, but there's not enough time left to give you a penalty or a free-kick to justify the goal"
This is a whole lot of speculation on my end, with precious little to back any of it up.
My initial question was more in-line with: "Hey, what if we hired Marco Rose/Diego Simeone as an Assistant Manager. Would we win more than if we hire a random cheap AM from the LLM, who just happens to have godly Mentals, but horrible Coach Attributes?"
Lapidus said: When you play with Instant Result then your assistant manager does the team talks, substitutions and the shouts and if he's better than you in those areas then you'll get better results with Instant Result and if he's worse than you then you'll get worse results.
It's simple as that. Expand
Say, I wanted to cheese this to the max. Which attributes should my assistant manager bring to the table?
- Tactical Knowledge - Tactics (?) - Motivation - Shouts (?) - Judging Current Ability - Substitutions (?) - Determination - Just because it makes everything better?
From your experiance, the roles dont matter so much? Like if player is good as CM, it doesnt matter if it is playmaker, channel, attacking ? Expand
I'd argue that roles matter a lot, but their impact is relative to the tactic & team-composition (or match-engine bug abuse)
The calculator is meant to help you more as a tie-breaker. Who is better? Questions like: Haaland or Mbappe? According to the calculator, Haaland is better.
It will also help you find undervalued players to replace the Star players you're looking to offload.
The research on role-familiarity showed that familiary a significant impact on match performance. Thus, if you tried to play Virgil Van Dijk, as a Target Forward, he would underperform until he reached peak familiarity. A natural Nr. 9 would still outperform him though.
BaZuKa said: One question: in this case, with two very similar players, would you use the top 5 impact as the deciding factor?
Expand
I wasn't even considering using it before @Cassoh mentioned it. Now that it is there, I can understand the value. Perhaps he can provide more insight in how he's using this information.
===
My take on the two players: In your specific example I wouldn't look at the 'top 5 impact' first, but rather his age and positions. The price-tag difference is almost irrelevant.
Karim Konate is 19, and has an extra year of turbo-growth ahead of him. He also didn't "waste" CA points on learning multiple positions, which means a lot of that untapped PA can go into attribute growth, rather than versatility.
Maurits Kjaergaard however is plateauing.
If I wanted an AMC, I'd pick Kjaergaard, because well-roundedness (lower % in top 5 impact) seems more relevant to midfielders than strikers.
Strikers operate more in the edge-case space, so you'd want higher % in top 5 impact. Their off-the-ball needs to be higher than the defender's positioning. Their jumping, pace & acceleration needs to be higher than the defender's. All these would produce higher xG.
Given the type of goals I have seen being produced by strikerless, I'd lean towards a well-rounded midfielder with a high-score on long-shots, and the trait "shoots with power". Same seems to apply to wing-backs.
For wingers/inside forwards, I noticed a disproportionate number of non-penalty goals coming from crosses+headers. i.e.: left-flank crosses from byline and right IF heads it in.
These are just observations, and my opinions. I have no numbers or data to back any of this up.
To find a more definitive answer, perhaps create two separate saves, and vacation through an entire season. Write down the results, and compare which player performed better, and perhaps which player had the higher valuation at the end of the season.
CBP87 said: Hey @Possebrew I'm hoping you can help, I can't get BepInEx to work, I've dropped the required files in the relevant folder but when clicking on the fm.exe nothing happens apart from the game loading, my understanding is that a window should appear showing the BepInEx code working its magic, you wouldn't happen to have a solution to get this working would you? Thanks Expand
- Got full internationalization going now for FM24 & FM26 (that means all languages work) - Added Total 5 Impact as a column - various responsiveness features - pagination for the table, so it doesn't lag so bad when you have 9000+ entries uploaded
Working out the last kinks atm, might have a release ready tomorrow.
I'm afraid it's not that simple, but I understand that there's interest in internationalization. I will look into it this weekend. Takes some time to generate the necessary static data for normalization.
Cassoh said: Sure thing. So when you load your players in under Data Analysis, click a player and under the header Impact Analysis, at the very bottom, is 'Top 20% Impact'. I want to sort players by this percentile to see which players might be strong overall, because I’ve found some players very useful who haven’t had the highest total impact but were very good all-round players. Expand
I am still not sure I understand. That number tells you how much the top 20% attributes have contributed to the total score of the player.
For example, I have a player here with 576.14 Score. His pace, acceleration, stamina, dribbling and anticipation (5 out of 25 attributes) have accounted together for 81.8% of the 576.14 Score.
For another player, the top 5 attributes have been in order: pace, acceleration, anticipation, work rate & jumping reach. These 5 attributes accounted for 90.4% of his total 364.72 score.
Thus your explanation confuses me. You want to sort by players whose top 5 attributes accounted for most of their total score?
I mean, yes. A player with lower Top 20% impact, would be a more well-rounded player, given the same score. But he'd be more of a "jack-of-all-trades master-of-none" kind of guy.
I mean I suppose I can add a column that shows you this information. Will have to consider how I run this while uploading.
Ah, and the position names were the big deal-breaker. Each language used different names, so it made parsing a nightmare.
The problem arises, because the team I was using did not necessarily have each and every position in question, so sometimes you had these compounded position names that made everything horrible to narrow down. I always ended up missing one or the other position.
That is because I am using the "D/AM (RL)" notation for the "Team Builder"
After uploading my CSV, only the GKs are getting actual scores, every outfield position (ST, AM, M, D/WB, etc.) is sitting at 0.00.
Went to check Attribute Settings → FM26 → Outfield and every attribute across Technical, Mental and Physical is set to 0 points. That explains the 0.00 scores
Was a strange one. It happened with the deployed version, but not local.
Should be fixed now. Can you check again?
I updated "Getting Started" with the scoring changes. A tl;dr is on the Change Log page.
For those that haven't yet, kind reminder to re-download the Player Database view.
As usual, let me know if you find any bugs!
If you need the CSV: https://www.transfernow.net/dl/20260408bH6PQIYV
That upload was very helpful, it helped me find a lot of bugs. The benefit of using real data vs synthetic data
The interim formula also seems to not be as OK as I had expected, had to redo the aggregation and all that. It wasn't wrong per se, but it was causing the Core4 Attributes (Acc, Pac, Jum, Dri) to have an even bigger impact on the final score than they already have.
So the overall scores might be a bit lower than before, and will favor all-rounders more.
I haven't fixed all the bugs yet, but I should have a release available this weekend.
Thank you!
Hi mate, what did you change compared to previously? Would be good to know! Thanks for all your work
There's new attribute weights for FM26, which are derived with formulas from the new research data.
And ofc toggles to move between them.
The points weights should be fine, but the goalsFor & goalsAgainst numbers were a bit iffy to determine. Which is why they are currently disabled in the calculations. So you may see them in the UI, but they do nothing.
This is all placeholder for now, while I work on a better way to extract the data from the excel and come up with a new formula.
FM24 or FM26?
Squad or Player Database?
By any chance, in the Data Analysis tab, do you see Position "Unknown" for some players?
I ask, because I updated the view in FM26 yesterday.
The old "Player Database" view, used the column "Best Position", instead of "Position", so it wasn't converted correctly.
If so, you can either redownload the view from the page, or manually right-click & replace the column "Melhor Pos" with "Informações Gerais" -> "Posição"
https://fm-arena.com/thread/13508-training-is-fake-it-just-assigns-attributes-not-grows-attributes-results-based-on-a-large-number-of-tests/
Cheers! I uploaded an update with the FM26 weights. I can't vouch for correctness of the attributes, but for as far as I tested them in my save, they match expectations.
What specifically did not work?
I think what is most beneficial is non-rotation tendency. You'll have noticed that AI managers rarely, if ever, rotate their players. They keep running them even if they're exhausted, and it is perhaps for the two following reasons. First, low match fitness is significantly more detrimental than low condition, and almost impossible to keep atop of if a player is not playing almost every game of the season. Second, the high morale produced by a run of good form seems to moderately increase the chance of winning the next game, but this also snowballs over the season. Morale appears to return to a neutral baseline fairly quickly, so rotation interrupts and prevents good form from occurring.
Additionally, it's been deduced that giving young players a handful of matches each season or subbing them on in the dying minutes does little to nothing for their development. They need at least ~15 full matches per season at a minimum for development, otherwise they're better off playing friendlies in the reserves.
I realized that at this point, no assistant manager attribute is essential, and so I put this theory to the test. FM24, Manchester City, with a top preset tactic and set routines. Assistant manager was assigned all relevant roles, including setting training.
Default:
+122 105pts
+89 95pts
+82 92pts
+87 90pts
95.5 pts average
1 PA:
+106 100pts
+81 97pts
+95 96pts
+79 92pts
+83 88pts
94.6 pts average
To reduce the effect of the player variability through the season, and also see if a different effect was seen in an inferior/underdog team, I tested using my 1 CA players Man City team. I created a perfect 200 PA assistant manager with 20 in almost all attributes, and a 1 PA one with the opposite (I even gave them 1/10 English proficiency). Same tactics were used. Simmed only to midseason, since I was running into problems with sacking after that point.
200 PA:
+42 66pts 2nd
+13 48pts 5th
+1 48pts 8th
+10 40pts 8th
50.5 pts average
1 PA:
+27 57pts 2nd
+15 46pts 5th
+19 45pts 7th
+13 43pts 6th
47.75 pts average
The difference between a perfect assistant manager and an abysmal one is statistically insignificant as to be indistinguishable from random chance variation of results. It may not be just a coincidence that the superior assistant manager does slightly better in both tests, but even if so, the difference between two realistic assistant manager options (stats of 12 vs 16 instead of 1 vs 20) would be minuscule.
Seems to me that one should choose an assistant manager who is a good coach or cheap.
I did notice that it's better to sub players out based on their current match-rating than based on condition.
6.2 players seldom recover to get an 8.0
Replacing these bums with a fresh player that starts at 6.7 sometimes ends in a "super-sub" situation.
It's almost as if the the game rolls some dice for each player during the match on whether they will have a good match or a bad match, and then it spends the rest of the game working towards the outcome of the dice.
I also noticed that when I use Instant Result and sometimes check the player ratings at the end, I find that the the assistant did not sub out a player, and they end up with a 4.8 or other crazy-low rating.
If this hunch were correct (and I don't know how to prove that it is not), then I still wouldn't know whether a single sub rerolls the entire match outcome, or just decides the fate of this new sub. I'd lean on the latter. In either case, the aggregate result might tilt.
But even so... how does the game deal with: "oh, you were supposed to lose 1:2, but now you're supposed to win 3:2, but there's not enough time left to give you a penalty or a free-kick to justify the goal"
This is a whole lot of speculation on my end, with precious little to back any of it up.
My initial question was more in-line with: "Hey, what if we hired Marco Rose/Diego Simeone as an Assistant Manager. Would we win more than if we hire a random cheap AM from the LLM, who just happens to have godly Mentals, but horrible Coach Attributes?"
How do you pick Assistant Managers?
It's simple as that.
Say, I wanted to cheese this to the max.
Which attributes should my assistant manager bring to the table?
- Tactical Knowledge - Tactics (?)
- Motivation - Shouts (?)
- Judging Current Ability - Substitutions (?)
- Determination - Just because it makes everything better?
From your experiance, the roles dont matter so much? Like if player is good as CM, it doesnt matter if it is playmaker, channel, attacking ?
I'd argue that roles matter a lot, but their impact is relative to the tactic & team-composition (or match-engine bug abuse)
The calculator is meant to help you more as a tie-breaker. Who is better?
Questions like: Haaland or Mbappe? According to the calculator, Haaland is better.
It will also help you find undervalued players to replace the Star players you're looking to offload.
The research on role-familiarity showed that familiary a significant impact on match performance. Thus, if you tried to play Virgil Van Dijk, as a Target Forward, he would underperform until he reached peak familiarity. A natural Nr. 9 would still outperform him though.
BepInEx_linux_x64_5.4.23.5.zip
BepInEx_linux_x86_5.4.23.5.zip
BepInEx_macos_universal_5.4.23.5.zip
BepInEx_Patcher_5.4.23.5.zip
BepInEx_win_x64_5.4.23.5.zip
BepInEx_win_x86_5.4.23.5.zip
Source code
(zip)
Feb 8
Source code
(tar.gz)
Feb 8
BepInEx_win_x64_5.4.23.5.zip
Is this that one?
I'm guessing you're looking at the Github releases? You're on the wrong page.
On the exact link I posted, scroll down to "Artifacts"
I wasn't even considering using it before @Cassoh mentioned it. Now that it is there, I can understand the value.
Perhaps he can provide more insight in how he's using this information.
===
My take on the two players:
In your specific example I wouldn't look at the 'top 5 impact' first, but rather his age and positions. The price-tag difference is almost irrelevant.
Karim Konate is 19, and has an extra year of turbo-growth ahead of him. He also didn't "waste" CA points on learning multiple positions, which means a lot of that untapped PA can go into attribute growth, rather than versatility.
Maurits Kjaergaard however is plateauing.
If I wanted an AMC, I'd pick Kjaergaard, because well-roundedness (lower % in top 5 impact) seems more relevant to midfielders than strikers.
Strikers operate more in the edge-case space, so you'd want higher % in top 5 impact. Their off-the-ball needs to be higher than the defender's positioning. Their jumping, pace & acceleration needs to be higher than the defender's. All these would produce higher xG.
Given the type of goals I have seen being produced by strikerless, I'd lean towards a well-rounded midfielder with a high-score on long-shots, and the trait "shoots with power". Same seems to apply to wing-backs.
For wingers/inside forwards, I noticed a disproportionate number of non-penalty goals coming from crosses+headers. i.e.: left-flank crosses from byline and right IF heads it in.
These are just observations, and my opinions. I have no numbers or data to back any of this up.
To find a more definitive answer, perhaps create two separate saves, and vacation through an entire season. Write down the results, and compare which player performed better, and perhaps which player had the higher valuation at the end of the season.
If you find any bugs/issues, please let me know!
Can you check if this one works for you?
https://builds.bepinex.dev/projects/bepinex_be
Grab this one:
BepInEx Unity (IL2CPP) for Windows (x64) games
- Got full internationalization going now for FM24 & FM26 (that means all languages work)
- Added Total 5 Impact as a column
- various responsiveness features
- pagination for the table, so it doesn't lag so bad when you have 9000+ entries uploaded
Working out the last kinks atm, might have a release ready tomorrow.
['ÂGE', 'MONTANT TRANSFERT', 'COR', 'CEN', 'FIN', 'CTR', 'TÊT', 'TIR', 'T LG', 'MAR', 'PEN', 'ANT', 'CRG', 'SGF', 'CTN', 'DÉC', 'DÉT', 'INS', 'LDR', 'APL', 'PLA', 'VIS', 'VOL', 'ACC', 'AGI', 'DÉT', 'PHY', 'VIT', 'END', 'SRF', 'BOX', 'TSP', 'REL', 'PUI', 'DÉT', 'COM', 'EXC', 'PBL', '1C1'],
I'm afraid it's not that simple, but I understand that there's interest in internationalization.
I will look into it this weekend. Takes some time to generate the necessary static data for normalization.
I am still not sure I understand.
That number tells you how much the top 20% attributes have contributed to the total score of the player.
For example, I have a player here with 576.14 Score. His pace, acceleration, stamina, dribbling and anticipation (5 out of 25 attributes) have accounted together for 81.8% of the 576.14 Score.
For another player, the top 5 attributes have been in order: pace, acceleration, anticipation, work rate & jumping reach. These 5 attributes accounted for 90.4% of his total 364.72 score.
Thus your explanation confuses me.
You want to sort by players whose top 5 attributes accounted for most of their total score?
I mean, yes. A player with lower Top 20% impact, would be a more well-rounded player, given the same score. But he'd be more of a "jack-of-all-trades master-of-none" kind of guy.
I mean I suppose I can add a column that shows you this information. Will have to consider how I run this while uploading.
I am not entirely sure I understand what you are asking for.
Could you please elaborate?
And assuming you obtained this feature, could you describe what you hope it will do for you, which you are unable to do at this time?
The problem arises, because the team I was using did not necessarily have each and every position in question, so sometimes you had these compounded position names that made everything horrible to narrow down. I always ended up missing one or the other position.
That is because I am using the "D/AM (RL)" notation for the "Team Builder"
===
This for example was the German mapping:
// Additional German mappings
'OM (RL)': 'AM (RL)', // Offensiver Mittelfeldspieler (Rechts Links) - Attacking Midfielder (Right Left)
'V/AM (R)': 'D/AM (R)', // Verteidiger/Attacking Midfielder (Rechts) - Defender/Attacking Midfielder (Right)
'V/OM (R)': 'D/AM (R)', // Verteidiger/Offensiver Mittelfeldspieler (Rechts) - Defender/Attacking Midfielder (Right)
'V/M (L)': 'D/M (L)', // Verteidiger/Mittelfeldspieler (Links) - Defender/Midfielder (Left)
'OM (LZ)': 'AM (LC)', // Offensiver Mittelfeldspieler (Linker Zentrum) - Attacking Midfielder (Left Center)
'V (RL)': 'D (RL)', // Verteidiger (Rechts Links) - Defender (Right Left)