Testing players inside plug-and-play tactics proves the tactic is OP, not that attributes work like a magical spreadsheet. Expand It has been tested with default tactics.
Angelski said: High Pace looks great on paper, but role fit, mental attributes, and relative league quality matter far more in practice. Expand Roles do not exist. Even positions seem to have none or at most few attribute differences between each other, except for GK and position proficiency of course.
Edit: On reflection it might be an exaggeration to say that roles do not exist. But not by much. I guess you could point out that in meta tactics, roles are a part of it, and this not my domain. I guess what I really mean to say is that roles do not affect what attributes a player should get. So roles may have some tactical value, but you are clearly talking about getting players with attributes to fit roles, not tactical differences themselves.
Angelski said: Case in point: I used a striker with just 13–14 Pace/Acc in the Bulgarian league (where average CA starts around 95 and below). He scored 50+ goals a season, won 2 Ballon d'Ors, and helped me reach the Champions League every year—even winning it once. I eventually followed the popular advice to replace him with a 19–20 Pace/Acceleration player, and the speed replacement performed far worse overall simply because he didn't fit my tactic or have the same high Off the Ball and Composure. Expand The first part doesn't contradict my findings. In fact it is fully in line with my template which you can view on page 1, which shows 13-14 pace/acc is a viable minimum for the English Premier League, as exemplified by players such as Harry Kane and Robert Lewandowski.
The second part mistakes the causes. Unless you're simply playing a player completely out of position, tactics make use of all players the same way. Off the ball is a completely useless attribute that has been tested to have no effect, not just by me, but by both HarvestGreen and the FM Arena testing.
Angelski said: With FM26 introducing distinct in-possession and out-of-possession roles, the devs clearly made squad and tactical fit far more critical than just hoarding fast players to break the engine. It's about how a player fits your system across both phases, not how high their physical stats can grow. Expand I don't test FM26, but those who have tested it - HarvestGreen and FM Arena - have found that there are only a few minor differences in the effect of attributes, and that the new OOP thing is pretty much placebo. Again I haven't tested FM26 myself, but if the attributes effect remain largely the same as others have found, then 2/3rds of attributes remain completely useless and hoarding fast players to break the engine remains the way to go.. unless of course you prefer the more realistic and successful method, which is using my weights and following my templates. If anything, FM26 appears even more broken to me, as I notice all the top tactics are the same tactics that lack strikers and I've heard it said that they are, or at least were, more effective than Knap tactics in FM24. Maybe I'm wrong on that though, I don't play FM26 because it's hot garbage.
Angelski said: But I still don't believe that just stacking 5 attributes at 18–20 makes a player great. CA is capped at 200, so a player can't have half their stats maxed out and expect to be a complete monster without balance. Expand You should check out the thread where I have used a team of 1 CA players to consistently win the Premier League.
That was many months ago. Since then I've come up with a more realistic formula: No high pace/acc, no unrealistic attributes like say 16 dribbling, and a total of just ~90 CA on average. The template is on page 1, and yes, you can find similar players in the actual starting database to emulate the template - it would cost only a few mil to find and buy the right players straight off the bat to win the Premier League.
I don't know where you're getting this idea that I'm using attributes at 18-20. The template has no attribute above 15, and it's what I've been using in all my testing for ages now:
Angelski said: Even when precision is minimized down to 1–1.5%, you can never fully eliminate RNG. Focusing strictly on raw physical players just isn't the right way to look at the game. Expand Can't speak for you or anyone else, but 1% margin of error sounds good enough to me.
Please update yourself on the 'raw physical players' claim.
keithb said: You "test" using the best tactics and set pieces? You can get Darmstadt to win the bundesliga in the first season with that setup. Probably the double. You also take man city out of the picture. It's such an easy test, it barely proves anything. You don't even realise half the stuff you do is flawed. It's not even amateur.
Osimhen would easily outscore and get a better rating than Maeda. Expand Darmstardt players have ~120 CA. The players I have added to Man City are ~90 CA. Most people use a top tactic I assume. But otherwise the 4 teams test (80%, 78%, 76%, 74%) suffices as proof of validity of the weightings whatever you take exception to, as no custom tactic, set pieces, nor custom players were used (completely default tactics, with squads of real players).
You are not even familiar with what you are arguing against.
I want to say one more thing on the randomness point.
I tested teams in one league together consisting of 80%, 78%, 76%, and 74% rated players primarily because I wanted to speed the verification process up, but I also had some other reasons in mind. One was to test the 'its too random' theory.
It's one thing to say that Haaland finishes 1st to 4th reliably as a sole player in a largely controlled team. What about when you have a whole squad of real players that's both more variable and realistic - what happens to these small % distinctions then?
You can go back a few pages and see the results, but basically it turned out that even under these conditions, the results were reliably predictable. More than even I thought they would be to be honest. I thought it's probably going to be impossible to distinguish below ~5% difference, but there were clear distinctions even with just 2% Genie Scout difference.
The randomness was seemingly too overwhelming for you to distinguish factors, because you were assessing by the wrong factors. You were looking at league points and passing ability, rather than league position and anticipation, so overall it appeared too random to you to pin down.
Yarema said: "We see that work rate reduces in-match condition % identically to stamina. So stamina cannot be more important than work rate, or at least not more than ~20% more important say."
And you have the weights at 18 and 12 respectively? This is why I always ask questions because you come out with statements like this where it's one way for a few days then it completely flips, but you are 100% certain each time. You had one set of weights one day, I found them strange and enquire about the reasoning behind them, then you completely switch the system with massive changes and somehow it's my fault for asking. The "truth" shouldn't switch around so much, sure there is some inaccuracy and with more data we can hone in better.
I generally like that you are pursuing these questions, but the attitude "I am right and everyone else is wrong and stupid" is kind of offputting, especially when you do 180 as often as you do. Like do the tests and lets discuss if it makes sense, what are the limitations, can you approach it differently etc. You've found some interesting stuff that is unfortunately the victim of the level of discussion in here. Expand I could be like those guys who put out one file, call it a day, and reap the patreon/'buy me a coffee' money for the next 2-3 years off various 'Ultimate FM26 Weightings Megapack' releases while vigorously defending the unalterable truth of my findings.
The god honest truth is that my releases are haphazard and change drastically because I actually put time and effort into finding new things and don't let myself fall into the trap of getting my knickers in a knot over 'b-but what will people think of me now'.
I don't know where you're getting this work rate 18 and stamina 12 thing from. The current weights have 63 work rate, 94 stamina.
Yarema said: Like do the tests and lets discuss if it makes sense, what are the limitations, can you approach it differently etc. Expand That is literally what I've been doing. It's people such as yourself and keithb who have been trying to interrupt and overwhelm things by constantly posting the same tired complaints of 'you're wrong.. you're just wrong, because I know you're wrong, okay?' over and over and over again. From my perspective, more posts in the thread draw more attention to it, and allows for a kind of dialetical process where I can address normal user's critiques that aren't voiced and also realize certain things I may have done mistakenly or glossed over on occasion. And it's a damn good feeling on those days when I can suddenly present solid data to blow the BS claims out of the water.
Here's a suggestion: Perhaps stop acting like you're the last thing standing between me and human decency, and instead start doing some testing of your own to either support your claims or otherwise complement my findings. Or if you're lazy or uninterested, which is perfectly fine, simply stop insisting on tired outdated BS like 'technicals still matter somewhat' or 'attributes are mostly linear'.
keithb said: You rated genie scout default ratings higher than several ratings files created by people in the last 2-4 years lol. Expand Yes, including one or two of my own you'll note.
It's literally just measuring the % disparity between two players where there shouldn't be one, and saying which has the largest or smallest error.
Testing shows that on average, Maeda finishes 3 position, Osimhen finishes 3.833 position. Yet your weights show Osimhen 88.50% FS 88.37% TS, Maeda 79.80% FS 69.06% TS.
That's embarrassing for someone who insists nothing has changed since they learned 'everything' 2 or 3 years ago. Or at least I'm embarrassed for you.
treath said: we want to know who is the best player, what if liverpool has a awesome season in a simulation and my position got lower than in another simulation, what this have to do with alisson be better thank ter stegen?
it's just randomness, that not prove nothing nothing nothing about atributes...c'mon einstein scientist (never wrong) get your scientific method together and don't do this bs
You are Just misguidind people for fun Expand I didn't mention it, because I thought the data self-evidently refuted the ol' 🤪 iTsAlLrAnDoMaNyWaY 🤪 argument.
As an aside, I'm enjoying these text color options.
We have heard it claimed for a long time that you cannot even measure, let alone employ, various factors in the game, because the inherent volatility is supposedly too great.
Yet observe the results I recorded: When I test Haaland, I don't have him suddenly finishing 9th or 16th. His lowest result in 6 samples taken is 4th. Doesn't seem very random to me. And it's precise enough to even distinguish clearly between him and Mbappe and various other contenders, who finish 2.5 or 3 instead of 1.666.
Now there is the valid point of seasonal variation throwing a spanner into the works, because realistically we're playing for each season, not waiting around for it to eventually get it right in season 2 or 3, even if season 2, 3, 4, 5, 6, 7 are all correct.
But I have calculated that variation and taken into account in assessing what level of certainty we can reasonably hold. I've stated before that I asked ChatGPT to combine this seasonal volatility, and the the current degree of statistical error, and it said we can be reasonably confident that to the point of +/- 7.8%. In other words, when I'm looking at GS for the best ST, I can have reasonable confidence that the best ST is going to between 83% and 90%, rather than 67% and 90% for GS default weights, or 60% and 90% for keithb's weights. I don't know if this sounds useless to you, but it doesn't to me!
keithb said: You keep saying what others have said or thought and you're always wrong haha. Like Yarema said above. I set stamina higher for positions that lose more condition in matches. Thats it. But you still haven't worked out why its important.............. no surprise there.
More ground breaking results everyone; speed is important for centre backs as well as strikers!!
And there's more;
" 2/3rds of attributes continue to have zero effect on final season position" Who bloody knew!!
What are you even talking about when you say "The trouble with the 'pace/acc is everything' explanation is that it failed to account for players such as Kane and Lewandowski" What does that even mean?!?! They are accounted for lol.
More than half the stuff you say is just from your fantasy world. Its a complete fiction that you've created and then talk about. Its not real.
In your last testing you said jumping reach was super important for every position and now you've set it quite low. Before you said Mbappe was shit hahah. Now he's your 2nd best player lol. And lets not forget you said I was pissing in my own face for saying stamina was important haha. You said you didn't need it at all and now you say its 80%!! What level of reality are we supposed to be operating on? Because I cant operate on yours.
Keepers dont need technique
Forwards dont need concentration
Wingers and 10's can have jumping reach 1.
Centre backs dont need high stamina. Expand Rattling off a bunch of falsehoods doesn't make them true.
Unfortunately for you, EBFM's uploads on Youtube still exist there and can be referred to.
We see that work rate reduces in-match condition % identically to stamina. So stamina cannot be more important than work rate, or at least not more than ~20% more important say.
But the real showstopper to your argument is that in the video Max also states 'a player will use 15-20% OPC per half' while showing a match in progress that shows players that started with identical condition % and attributes, and the DL/DR condition % is indistinguishable from other positions.
That said, anecdotally I know that in standard play, fullbacks tend to use a bit more condition % than the other players. I suspect this is due to the 'best' fullbacks also tending to have high work rate. This is a kind of self-fulfilling prophecy: People use GS weightings that select for high work rate on fullbacks, so they end up using more condition %, requiring higher stamina, and therefore favoring stamina too. But very high work rate on fullbacks isn't necessary to begin with. Testing shows that work rate above ~11-13 is largely superfluous.
We also see that 1 > 20 sta is equivalent to 100% vs 97% condition difference when starting match. We know that bad natural fitness will result in an inevitable condition difference of at least that, yet you have natural fitness weighted at 0.
In regards to jumping reach, it remains important for every player. I don't know how you consider '46' weighting low. It was in fact viable at lower values of ~20-30, but I thought 46 was more appropriate and it aligned fine. I haven't yet examined if extra jump reach is necessary on DC, which is an important remainder matter to clarify.
In your weightings, Vinicius Junior is no.1 FS, Mbappe is 2nd beating Haaland, while you have to scroll down to find Maeda below a 32-year-old colombian at Orlando City. Comparable analysis showed that your weightings have worse variance than the Genie Scout default values from a decade ago. The question I have is not how did you come up with those ratings, but how did you manage to get them to do worse than even the Genie Scout default??
Yarema said: I don't think it was stated that a GK is only 20% of an outfield player.
Why you keep using league position as the end point still baffles me. Yes it has less variance because there are only 20 possible spots compared to lets say 100 possible points. You have to go really out of your way not to be in a +-3 places range. Additionally there is a lot more volatility in the middle of the table than at very top or bottom.
There was decent data for GKs from FM Arena, Harvestgreen, Orion and probably others and they mostly match with what you found so it's not exactly groundbreaking. Still valuable additional data from a different approach. Expand Yeah.. look, I've heard so much rubbish from you that you're going in the keithb category for me now.
I feel like I've given you a fair go.
I responded to your claims several times with solid evidence to the contrary, whether it was about technical attributes (aside from drib) mattering at least somewhat as you insist they do when they don't, or your claim that attributes aside from work rate are 'pretty much linear', but you just ignored my responses - even when I reminded you that I had already responded to your claims.
It's understandable to not be in the know, or double down on something you've always believed in. I did that with personality distribution in FM. But I conceded the point after looking into it. You just seem intent to repeat the same spiel of 'you're wrong' over and over, and making claims that are easily refuted with a simple lookup of what I'm referring to:
harvestgreen22 said: 2.The goalkeeper is the least important person among the 11 players. The verification method is to select any one of the player on the field, change the given attribute, and observe the extent of the impact on the winning rate. The impact of the goalkeeper on the winning rate is roughly only one quarter or one fifth of that of the any one players on the field. Expand
'There was decent data for GKs from FM Arena, HarvestGreen, Orion, and probably others' - Where is it? I've addressed the HarvestGreen data in the post you're directly responding to. C'mon man.
I will address your other point for the benefit of those who are being fair-minded and just want to know the truth of how FM works.
As you state, I use league position to assess because it has much less variance than league points. It is also true that there is will be more position volatility in the mid-table than the top 6, because the quality gaps between the top 6 are likely wider than the mid-table. However, the implicit idea here that league position only seems more reliable because it compresses the volatility is misleading - what actually happens is that league position is the first outcome, and league points seems simply tacked on as a narrative for whatever reason (it is at least consistent with this hypothesis). So for instance, a team will finish 2nd most times, but one season it might get 64 points, the next it will be 84 points. One season it will be 2nd, 14 points behind, the next it's 2nd only on goal difference. And this is not just some miracle oddity, I've observed this myself literally hundreds of times. Hence, position is a much more accurate/less volatile measure of performance - and how can you argue the league position you finish in (i.e. 1st), isn't the primary goal of fiddling around with attribute distribution? Who wants to reach 80 league points and finish 7th, rather than finish 1st regardless of the league points tally?
For whatever reason it seems to me that people take FM Arena's testing rather than HarvestGreen's as a starting point, so I'll begin with that:
So we see here that stamina is moderately, not highly, important. Specifically it is 29.7% of either acceleration or pace.
Two important things to note are that points is used a measurement rather than league position, and that we are measuring from 8 > 20 stamina. It's not clear what the makeup of the team or opposition is, but at least it's giving us a fair idea of what to expect in any case.
We see here in HarvestGreen's latest result for stamina that is it 27.1% of acceleration or pace going from 1 > 18.
For 6 > 18 it is 14.1% of acceleration or pace. In other words, stamina is never critical, but becomes even less important beyond '6'. He also has 1 > 6 results showing this.
In his initial attribute test shown here, stamina 10 > 20 is 30.9% of acceleration, which in my view is the most important result in regards to application to the Premier League.
In keithb's GS weightings, stamina is weighted at 60 for DLR/WBLR/DM/MC, 20 for AMLR/ST, and 0 for DC.
What this no doubt reflects is the now outdated myth that different positions require different sets of attributes. This is not an article of faith for me; I myself used to believe in, and initially employed, weightings that varied per position. Partly this was because each position has different CA weightings for attributes, and so high stamina for low CA cost on fullbacks seemed like a boon. I suspect that the myth has been generally perpetuated throughout the years with the CA weightings as a primary impetus, and that keithb isn't even aware of this - he seems to just think that stamina is what's necessary to win, simple as that. My mind was changed by the test results.
What my test results found, and seemingly implicitly HarvestGreen's & FM Arena's too, is that pace/acc is just as critical on DC as it is on ST, many attributes such as 'tackling' or 'passing' don't matter at all for any position, and if you try to test for positional differences, there doesn't appear to be any, or at least many. There might be still be some differences - I'm still wondering if pace & acc are of equal importance across all positions for instance, but overall we're talking small differences, not major ones.
Through brute force testing, I found that stamina of 13 is the essential requirement for the Premier League, while pace/acc is 15:
However, you can get away with as low as ~13 pace/acc (think Harry Kane, or Lewandowski), so long as stamina is ~15 or higher. Conversely, a 20 pace/acc player has less need of stamina, and can get away with 12 and no doubt lower.
So if you put all that together, basically stamina should be weighted at ~33% of pace/acc, is what I found.
So you can think of it as either 33% or 86% of acc, perhaps somewhere inbetween. So ~50% roughly. This was a change from the ~15% I previously used, which was primarily based on HarvestGreen's 14.1% 10 > 20 finding.
With my latest weightings, which wipe the slate clean of assumptions and simply try to align weightings with actual real player rankings supported by test results as evidence, I found that only somewhere around ~90% weighting for stamina worked - or ~90-110% of acc. This is far more than even keithb's 60% weighting for sta on fullbacks.
But why is that necessary, and why does it work if so? It is necessary because someone such as Maeda or Furuhashi has very few green attributes, yet ranks above the likes of both all-rounders such as Lewandowski and fairly well rounded high pace/acc players such as Osimhen. If you analyze it carefully (you need to compare multiple players in conjunction to find out it can't be his high work rate, det, aggression, etc.), the only way to make the rankings align seems to be to boost stamina to such a high weight. Perhaps I've missed some other option, but this where I'm at. As to the explanation as to why it is so, it makes more sense once you realize that stamina is preserving both pace and acc, so its value is doubled, and it is preserving the two most valuable attributes.
No one that I'm aware of has previously come to these conclusions about stamina. Even the overlap between this and the old 'stamina for fullbacks' myth doesn't extend this far. And it puts a big dent I reckon in the general pervading view that pace/acc is everything, which was always sort of heard of, but never really solid fact until HarvestGreen's findings. The trouble with the 'pace/acc is everything' explanation is that it failed to account for players such as Kane and Lewandowski. It would be (and still is by some even on this forum!) explained away as compensated by high technical skill such as passing, etc. which is a red herring - these technical skills really are completely useless, and the differences lie in other areas such as stamina, work rate, anticipation, and a handful of select others. 2/3rds of attributes continue to have zero effect on final season position.
I wasn't going to respond but I guess it's good to take stock of where we're at from where we've come from. Instead of vaguely telling a narrative of progress, I'll be presenting precise data for a clear comparison.
Let's start with GKs:
HarvestGreen did 2 seperate tests on GKs.
December 2024:
April 2026:
The first thing I'll draw your attention to is that the results seemingly contradict each other. In the 1st test, acceleration does little to nothing. In the 2nd, it is the third most important attribute, above aerial reach which got a tick in the 1st test. Vision got 2 ticks in the first test, but does nothing in the 2nd test. There are more examples like this.
In skawkclsrn's case the issue is lack of samples, but here the issue is the methodology:
- The 1st test is using a team with 10 in every attribute, and increasing a single GK attribute to 20, then measuring the goals conceded. The first two aspects underlined are potential problematic factors, for instance what if a GK needs better reflexes and agility because the outfield is deficient in speed - perhaps with a speedy team, a GK only needs aerial reach to handle potshots he can see coming from afar. That is just a minor critique to show even solid tests can have pitfalls. The real problem is measuring the goals conceded, because as shown in my testing results, goals conceded is not at all a consistent measure of success, similar to how league points is all over the place compared to final league position. It will give some correlation, hence why agility, reflexes, and aerial reach all correctly have ticks, but that is just because teams who finish 1st obviously concede less goals than teams that finish 20th.
- The 2nd test is starting with the GK at '20' in all attributes and setting 1 attribute to '1'. This is different to how HarvestGreen does his outfield tests, which increase/decrease from '10'/'12' for all attributes, and it's obviously less realistic to begin with. But the main issue, only evidenced by my own testing later, is that having all attributes at '20' is likely compensating for and obscuring certain deficiencies. So we see that reflexes and agility are pretty much the only 2 attributes that matter, and even then, their individual impact is fairly minimal (12.8% at most).
HarvestGreen also stated that his observation is that GK is 20% of effect of an outfield player, if I remember correctly, which is basically saying the GK should be an afterthought and not really worth worrying about.
All of his tests also carry the same caveat that he is testing against opposition that has identical attributes. For some attributes, such as Jumping Reach, Aerial Reach, Strength, etc. I hope it's obvious how this could constitute a problem when it comes to translating to how things work in real leagues.
Now, coming to my tests. I brute force tested individual attributes for GK until I got a solid picture of what an optimal GK looks like. So what I would do is I would start with all attributes at 1, except a few obviously necessary ones like agil, aer, ref, etc. and measure the position. The initial result might be 13th. Then I would increase each attribute from '1' to '20' and measure the improvement. If no improvement in position, ignore it going forward. After learning what attributes matter and which don't, I would begin refinement, seeing how low I could get an attribute to find the 'sweet spot'. And I also start testing those attributes I took for granted at the start: agil, aer, ref. I would test a few potential combos (i.e. does aer only matter if jump is high?), but obviously it can't be an exhaustive search. Now what I would most often find is something like this:
Usually I would choose 14 to settle on. But if '13' is relegated, or '18' is 1st, then obviously I would likely change that to a '15' - while also considering realistic availability in the game.
Now these are the essential GK attributes I ended up with:
We see from my latest comparative testing of real players within the game, that this player would constitute something of a middling Premier League team GK. This GK finishes ~7th, the best finishes 4th, Barcelona's finishes 6.4, and a League 1 GK finishes 8.4.
One key attribute that is hidden from that screenshot is technique, which turns out to be quite critical for GK. You can also see this reflected in the database data for GKs - the better the GK, the higher technique he is generally given. I actually notice for the first time now that HarvestGreen actually gave a tick next to technique in his 1st test, though in the 2nd it is listed as 0.0%.
It's not clear yet exactly how much a GK accounts for as a % of a team's overall performance, but we can see the GK is more critical to success than HarvestGreen gave the GK credit for. My current estimation is somewhere around ~66-100% of an outfield player.
So we know my template works. Let's compare to what we 'knew 2 years ago', which is HarvestGreen's data and nothing else:
- HarvestGreen found vision and pace, and later reflexes, to be the most important attributes. I found vision '1', pace '13', and reflexes '11' are adequate, while aerial reach is the most essential attribute. - HarvestGreen summed up that GK is has 20% impact of outfield player. My evidence shows it's much more than that, even the difference between relegation and a top 5 finish at times. - Generally speaking, there was no solid data on GK at all previously. Now I've provided you with a precise template for GK that is backed by solid evidence in the form of testing both real & artificial players in a real and popular league (English Premier League).
If I've missed anyone else's findings on GKs, that was any good, let me know.
Alisson - 6th, 3rd, 4th, 4th, 3rd = 4 position Thibaut Courtois - 3rd, 7th, 2nd, 5th, 8th = 5 position Jan Oblak - 5th, 8th, 6th, 4th, 3rd = 5.2 position Marc-Andre ter Stegen - 10th, 11th, 4th, 4th, 3rd = 6.4 position Anatolii Trubin - 15th (sacked), 6th, 4th (sacked), 5th = 7.5 position Jack Stevens - 13th (sacked), 4th, 9th (sacked), 8th (sacked), 8th (sacked) = 8.4 position Sinan Bolat - 4th, 2nd, 19th (sacked), 5th, 14th (sacked), 9th (sacked), 12th (sacked), 5th = 8.75 position
I removed Courtois' starting injury to test him.
It shows that GK does in fact matter, though it does seem a lot more variable - in the downwards direction.
So although a GK can't really make your team win much more, they can in fact make or break your team.
Observations through player comparison:
- Jack Stevens is a mediocre all rounder who normally plays for Cambridge, and show you can't really get away with playing mediocre GKs - Bolat is interesting.. he is like Stevens, but a tier above, but has green reflexes and jumping reach. Shows good reflexes aren't sufficient. - Trubin, like Bolat, also significantly underperforms. He has green Aer, Jump and Handling. None of these are sufficient alone therefore, even with seemingly above average values in all other areas.
We know that the following GK finishes ~5th and with good consistency too:
This GK finished 7th:
It seems like only a select few of real GKs can do better than the above players.
If I filter for attributes of that second GK (excluding natural fitness), no players come up.
It seems that GK is about what is the least worst deficiency do they have. I suspect it is this way so that, unlike the outfield, you don't get great GKs who make teams come 1st by the mere addition of themselves. Perhaps this also explains why GKs can't get really high match ratings, because their best is merely adequate.
Trubin has a slightly deficiency in determination, pace, acceleration. Alisson is only lacking 1 pace. Oblak slightly lacks first touch, acceleration, and agility.
To verify the theory, I reduced the filter by 1 for the attributes and tested the worst one of the 4 players that turned up (1 was Ederson @ Man City, 2 were Alisson and his backup GK @ Liverpool, and the worst was Davy Roef from AA Gent).
High variation, but the average result aligns with extra 1 in key attributes finishing ~7th
BTW it's stuff like Alisson's backup GK just coincidentally having exactly the right set of meta attributes that shows me SI are aware of and intentionally making most attributes useless and misleading. If it was an oversight, players that look like duds but nonetheless perform well and are placed in the top teams wouldn't be an identifiable pattern.
Now I know quite a few people are also interested in what kind of GK gives the most clean sheets, or lowest goals conceded. So here are those stats:
Alisson - 13 cln, 48 con | 14 cln, 39 con | 15 cln, 40 con | 14 cln, 36 con | 11 cln, 44 con = 13.4 cln, 41.4 con Thibaut Courtois - 9 cln, 41 con | 12 cln, 48 con | 10 cln, 56 con | 10 cln, 45 con | 11 cln, 50 con = 10.4 cln, 48 con Jan Oblak - 8 cln, 49 con | 11 cln, 43 con | 10 cln, 59 con | 14 cln, 34 con | 15 cln, 40 con = 11.6 cln, 45 con Marc-Andre ter Stegen - 4 cln, 58 con | 10 cln, 53 con | 11 cln, 52 con | 13 cln, 44 con | 14 cln, 46 con = 10.4 cln, 50.6 con Anatolii Trubin - 10 cln, 46 con Davy Roef - 14 cln, 45 con | 13 cln, 44 con = 13.5 cln, 44.5 con Jack Stevens - 16 cln, 41 con Sinan Bolat - 14 cln, 50 con | 10 cln, 52 con | 9 cln, 53 con | 8 cln, 54 con = 10.25 cln, 52.25 con
I only took stats from samples where there was no interruption by sacking.
bf3metro said: Just one small thing I noticed: you said you included Loyalty as “1” at the last moment, but in the weights you posted, Loyalty is 0.2. Is 0.2 the intended final value, or was Loyalty supposed to be 1?
Also, given what you said about Strength, Vision, Finishing, Teamwork and Loyalty, would you say these are attributes we should actually pay attention to when evaluating a player, or are they basically not worth considering unless the value is particularly low/high?
I’m especially wondering about Teamwork since you still seem a bit uncertain about it.
And one last thing: have you also updated the goalkeeper weights, or are these still the current ones?
PS: What is the correct way to write it? Expand Was referring to the FSMT26 weights. For FMSS I divided by 5 because I think it had problems if you put the numbers too large from memory.
Those 5 attributes are not really worth paying attention to.. except loyalty of course, for non-performance reasons. Not only are they low value, they are non-essential. By contrast, agility is also low value, but moderately essential around ~10. Work rate is absolutely essential to have at least 6. You can have 1 strength/finishing/vision and finish 5th, that's proven.
Overall I don't think teamwork does anything. There's just a small element of uncertainty.
GK weights unchanged. I'm currently doing samples of GKs for rankings list.
I think either way you've typed it is fine.. just test each and see
Kriek said: Is it still needed to use the filters with the new ratings on Genie Scout or only the ratings? Expand Only ratings is fine. Filters still useful, but should be able to phase this out later by adding non-linear weights to FMST26 (not GS though).
bf3metro said: I also noticed that you added 5 attributes that weren’t in the previous set: Strength, Finishing, Ambition, Loyalty and Vision.
Was that intentional as part of the new weighting method? Expand Finishing, strength, and vision were technically in the previous weights. I dropped them from the file because the attributes that are adequate you are bound to get at the required levels most of the time anyway:
Strength - 8 Vision - 7 Finishing - 7
I played around a fair bit with strength in particular in these new weights. I thought maybe it's around 10-20, but turned not to be the case. Vision is 20 on Messi which was food for thought, but it didn't seem to affect ranking alignment for the better.
Actually teamwork is the one I most wonder about.. players like Maeda have high teamwork, and there's some mathematical study on FM24 on github that a guy did for university work that found teamwork was a key correlate of performance.. but I read through it, and overall I think its just full of mistaken findings. My own testing and HarvestGreen's found teamwork doesn't matter, so at the moment I don't see a valid reason to include it.
Loyalty I included as '1' at the last moment pretty much symbolically. It would have some very minor to minor effect via morale, so i felt justified giving it a little '1'. But obviously loyalty deserves some weight for gameplay reasons.
bf3metro said: The WorkRate interaction theory is probably the most interesting part to me. If something like Determination × WorkRate × Stamina is happening, it could explain why isolated attribute testing and the ranking weights seem to tell slightly different stories. Expand
The main thing would be the non-linearity of attributes, which the testing did show, but can't be reflected in linear weightings.
HarvestGreen did do some theorizing on attribute interactions, and I've dug up quotes from the Collyer brothers and in-game tips and whatnot which describe attribute interactions as well. I found that those latter things turn out to be generally untrue, or at least outdated.
I reckon there are attribute interactions though, and HarvestGreen's ideas on it are interesting.. just lack the solid data on it at the moment though.
I don't really have solid data to back it, but I'm personally convinced that aggression x dirtiness is an interaction, it just seems logical and seemed to solve something insoluble with the ratings at one point too.
This is partly why I decided to release it as is now with mostly linear weights, because one could go down the rabbit role of trying out all these different attribute combos and either getting nowhere or being mislead into seeing something that isn't there, while also spending a lot of time doing so. It remains difficult to test attribute combinations, but I'm sure there's some method to bear it out properly.
bf3metro said: Also, I noticed there are some strange extra spaces between a few attributes in the list. Is that intentional, or just a formatting issue? Expand Probably a formatting issue
bf3metro said: One other thing I'm wondering: is it possible that some of your original FMSS weights were simply set too low at the beginning, which would also partly explain why a few attributes have increased so dramatically now?
Do you think these huge increases are entirely a consequence of the different methodology, or could your original FMSS weights also have been underestimating some of these attributes from the start? Expand The FMSS weights I gave before simply reflect the minimum set of attribute values to attain 5th in the Premier League. They represent the best 'bang-for-buck' relative ratio of attributes.
I guess the question that the objective ranking raises in relation to that is, if work rate supposedly caps out at 13, why is it so critical for Maeda, and important in general? And if dribbling isn't so essential, is it really necessary to include it high - even if it does actually have significant benefits. So it does shine a new light on certain things. Personally I strongly favor what is true in realistic testing (the player rankings derived from the team positions), rather than what the isolated or constrained testing shows (including my own).
johnconnerson said: Where does the 118 number come from? I'm trying to split up the visible and hidden attributes into two separate metrics for my own use. Expand The 118 number is just to get it to display with the highest % at ~90%. It's the same as the overall weight option in GS.
bf3metro said: The biggest question for me is: why such massive changes?
From what I understand, the old weights were based on finding the cheapest/minimum attribute balance capable of achieving around 5th in the Premier League, whereas the new weights are being fitted to a much larger set of objective player rankings.
So it makes sense that the result could be very different: instead of asking “what minimum combination of attributes gets results?”, you're now asking “which attributes best explain the actual ranking of players?” Expand Yes, that is precisely the correct summary of it, and I daresay a better explanation than I did.
I'm glad you immediately understand, because I think it's easy to misinterpret this latest thing I'm doing with the weights.
bf3metro said: The new model seems to put an enormous premium on Pace, Acceleration, Stamina, Work Rate and Anticipation, while attributes such as Dribbling, Consistency, Important Matches or Natural Fitness become much less influential.
And then you have Finishing at only 0.4, which is pretty wild too. Expand I started with the presumption that finishing is almost completely unnecessary, but ChatGPT independently came to the conclusion that finishing should be removed too when I fed it all the player's attributes and asked it to come up a set of weights to fit the rankings.
It was surprising how strongly weighted anticipation and work rate seems to be. Maeda is still down a bit, so work rate could in fact do with being a bit higher perhaps. It's strange to me, because I know that (and you can see HarvestGreen's data to verify this) that work rate almost caps out at 13. But that is when work rate has been tested in isolation or partial isolation. Perhaps there is something like 'determination x work rate x stamina' going on. I couldn't find a way of getting around weighting work rate highly.
Anticipation in my observations did continue to scale, but had diminishing returns. But putting that aside for now, it makes sense that it would be a key attribute because ostensibly it's the movement ('off the ball' as speed?) to the area before possessing the ball. So it sounds like it cuts down on the pace/acc requirement to me, because once they have the ball, they have less distance to travel. Concentration would be a similar kind of effect, where low concentration leads to a delay in getting into position to receive or stop a ball (the latter offset by 'positioning' perhaps).
Consistency, Important Matches, and Natural fitness has not been incorporated with much accuracy. They are kind of afterthoughts, as I know they will matter somewhat, but not sure to what degree. I've used HarvestGreen's data to set them, and so long as they're not interfering with the ranking, I generally leave them alone. Same with pressure, professionalism, etc. - and the left foot & right foot figures I've just pulled out of my ass, to give a little extra weight to the modest footedness bonus which we know exists. This stuff can be refined further later. I think HarvestGreen found that foot should also match the side they're on, and I think FMST26 allows us to do this as well.
Dribbling I've been in two minds about. Most of the time I had it set pretty high at ~40, sometimes as high as 60. I ended up reducing it to 15. ChatGPT actually said early on to do this too. Both were fairly equally functional. From testing, we know dribbling is a very strong attribute that scales well.. I suspect it's not needed much to match the rankings, because hardly any players have high dribbling anyway. It's the hardest attribute to get high in the game I believe.
bf3metro said: PS: Quick question: why are attributes like WorkRate, ImportantMatches, and NaturalFitness written without spaces? Is that just the naming format, and do they need to be written exactly like that in app.js as well? Also, should I add a comma after the final value, Consistency: 2.82, or should the last entry be left without one? Expand Yes, that is how they have to be written for FMSuperScout, otherwise the program won't work. That is how labels the attributes internally.
bf3metro said: Legend! Expand Actually this is proving more difficult than I thought it would be
I'm not a programmer, so I don't know how to implement the injury proneness & dirtiness weights. I thought, maybe I'll just add them as simple negatives, like for the GS weights, and adjust it until it matches the rankings, but then I realized it doesn't do FM24 db, only FM26.
Replace the weights in the app.js file with those (search for 'FM-Arena' ).
For adding injury proneness & dirtiness, which are in fact crucial (but I guess you could also just check players manually to have low inj and dirt values), we're going to need someone like Panneton0 or perhaps the creator of FMSuperScout itself to help out. Could probably do it myself, but can't be bothered doing an hour of research honestly.
bf3metro said: any release for FMSS? Expand Yeah, I left it out because I have to work out how to best translate it from the FMST26 weights first. I'll do that now.
This new file is a lot like mine LOL! Expand You have stamina weighted at 0 for DC, 20 for ST/TS/AMC/AML/AMR, and 60 elsewhere. It's not my fault your own weights contradict your own words.
Meanwhile, the key attribute of dirtiness you leave on 0, as you have 0 clue of course about what attributes actually matter beyond what we knew from HarvestGreen's testing a year or two ago.
But as I've said, you can have a variety of combinations that work to get to match the final ranking result. And at least you're implicitly accepting now that my objective rankings are correct.
Let's do a comparison of our two files to find out just how similar they really are. First, yours:
Junior is 1st for FS, even with 16 ST proficiency. Mbappe is above Haaland. Messi is 5% above Lewandowski and 10% above Maeda. Even Malen is above Maeda. Dybala is above Maeda and Furuhashi. Mbappe (8 jump, 8 head, 11 str) is a better target man than Glatzel (16 jump/head/str). It goes on.
Mine:
There are some minor issues, such as that Maeda should be ~3% higher, and that Malen & Glatzel should have a wider percentage gap between them.
keithb said: Why have you still used negative values and scored the hidden attributes in Genie Scout? Expand The only fair point you've raised in your entire posting history I believe
It is true that hiddens remain invalid for GS. I decided in the end to include them for my final GS adjustment at the end to try and closely match what I got in FMST26, halved for everything except dirtiness & injury proneness & natural fitness. Dirtiness & injury proneness are essential to the weightings, there's no way to do without them.
Because GS calculates incorrectly, the ratings remain a bit off compared to what you'll get in FMST26, hence I highly recommend FMST26, but I know many people aren't going to be willing to switch over.
Testing players inside plug-and-play tactics proves the tactic is OP, not that attributes work like a magical spreadsheet.
It has been tested with default tactics.
Angelski said: High Pace looks great on paper, but role fit, mental attributes, and relative league quality matter far more in practice.
Roles do not exist. Even positions seem to have none or at most few attribute differences between each other, except for GK and position proficiency of course.
Edit: On reflection it might be an exaggeration to say that roles do not exist. But not by much. I guess you could point out that in meta tactics, roles are a part of it, and this not my domain. I guess what I really mean to say is that roles do not affect what attributes a player should get. So roles may have some tactical value, but you are clearly talking about getting players with attributes to fit roles, not tactical differences themselves.
Angelski said: Case in point: I used a striker with just 13–14 Pace/Acc in the Bulgarian league (where average CA starts around 95 and below). He scored 50+ goals a season, won 2 Ballon d'Ors, and helped me reach the Champions League every year—even winning it once. I eventually followed the popular advice to replace him with a 19–20 Pace/Acceleration player, and the speed replacement performed far worse overall simply because he didn't fit my tactic or have the same high Off the Ball and Composure.
The first part doesn't contradict my findings. In fact it is fully in line with my template which you can view on page 1, which shows 13-14 pace/acc is a viable minimum for the English Premier League, as exemplified by players such as Harry Kane and Robert Lewandowski.
The second part mistakes the causes. Unless you're simply playing a player completely out of position, tactics make use of all players the same way. Off the ball is a completely useless attribute that has been tested to have no effect, not just by me, but by both HarvestGreen and the FM Arena testing.
Angelski said: With FM26 introducing distinct in-possession and out-of-possession roles, the devs clearly made squad and tactical fit far more critical than just hoarding fast players to break the engine. It's about how a player fits your system across both phases, not how high their physical stats can grow.
I don't test FM26, but those who have tested it - HarvestGreen and FM Arena - have found that there are only a few minor differences in the effect of attributes, and that the new OOP thing is pretty much placebo. Again I haven't tested FM26 myself, but if the attributes effect remain largely the same as others have found, then 2/3rds of attributes remain completely useless and hoarding fast players to break the engine remains the way to go.. unless of course you prefer the more realistic and successful method, which is using my weights and following my templates. If anything, FM26 appears even more broken to me, as I notice all the top tactics are the same tactics that lack strikers and I've heard it said that they are, or at least were, more effective than Knap tactics in FM24. Maybe I'm wrong on that though, I don't play FM26 because it's hot garbage.
Angelski said: But I still don't believe that just stacking 5 attributes at 18–20 makes a player great. CA is capped at 200, so a player can't have half their stats maxed out and expect to be a complete monster without balance.
You should check out the thread where I have used a team of 1 CA players to consistently win the Premier League.
That was many months ago. Since then I've come up with a more realistic formula: No high pace/acc, no unrealistic attributes like say 16 dribbling, and a total of just ~90 CA on average. The template is on page 1, and yes, you can find similar players in the actual starting database to emulate the template - it would cost only a few mil to find and buy the right players straight off the bat to win the Premier League.
I don't know where you're getting this idea that I'm using attributes at 18-20. The template has no attribute above 15, and it's what I've been using in all my testing for ages now:
Angelski said: Even when precision is minimized down to 1–1.5%, you can never fully eliminate RNG. Focusing strictly on raw physical players just isn't the right way to look at the game.
Can't speak for you or anyone else, but 1% margin of error sounds good enough to me.
Please update yourself on the 'raw physical players' claim.
Osimhen would easily outscore and get a better rating than Maeda.
Darmstardt players have ~120 CA. The players I have added to Man City are ~90 CA. Most people use a top tactic I assume. But otherwise the 4 teams test (80%, 78%, 76%, 74%) suffices as proof of validity of the weightings whatever you take exception to, as no custom tactic, set pieces, nor custom players were used (completely default tactics, with squads of real players).
You are not even familiar with what you are arguing against.
I tested teams in one league together consisting of 80%, 78%, 76%, and 74% rated players primarily because I wanted to speed the verification process up, but I also had some other reasons in mind. One was to test the 'its too random' theory.
It's one thing to say that Haaland finishes 1st to 4th reliably as a sole player in a largely controlled team. What about when you have a whole squad of real players that's both more variable and realistic - what happens to these small % distinctions then?
You can go back a few pages and see the results, but basically it turned out that even under these conditions, the results were reliably predictable. More than even I thought they would be to be honest. I thought it's probably going to be impossible to distinguish below ~5% difference, but there were clear distinctions even with just 2% Genie Scout difference.
The randomness was seemingly too overwhelming for you to distinguish factors, because you were assessing by the wrong factors. You were looking at league points and passing ability, rather than league position and anticipation, so overall it appeared too random to you to pin down.
And you have the weights at 18 and 12 respectively? This is why I always ask questions because you come out with statements like this where it's one way for a few days then it completely flips, but you are 100% certain each time. You had one set of weights one day, I found them strange and enquire about the reasoning behind them, then you completely switch the system with massive changes and somehow it's my fault for asking. The "truth" shouldn't switch around so much, sure there is some inaccuracy and with more data we can hone in better.
I generally like that you are pursuing these questions, but the attitude "I am right and everyone else is wrong and stupid" is kind of offputting, especially when you do 180 as often as you do. Like do the tests and lets discuss if it makes sense, what are the limitations, can you approach it differently etc. You've found some interesting stuff that is unfortunately the victim of the level of discussion in here.
I could be like those guys who put out one file, call it a day, and reap the patreon/'buy me a coffee' money for the next 2-3 years off various 'Ultimate FM26 Weightings Megapack' releases while vigorously defending the unalterable truth of my findings.
The god honest truth is that my releases are haphazard and change drastically because I actually put time and effort into finding new things and don't let myself fall into the trap of getting my knickers in a knot over 'b-but what will people think of me now'.
I don't know where you're getting this work rate 18 and stamina 12 thing from. The current weights have 63 work rate, 94 stamina.
Yarema said: Like do the tests and lets discuss if it makes sense, what are the limitations, can you approach it differently etc.
That is literally what I've been doing. It's people such as yourself and keithb who have been trying to interrupt and overwhelm things by constantly posting the same tired complaints of 'you're wrong.. you're just wrong, because I know you're wrong, okay?' over and over and over again. From my perspective, more posts in the thread draw more attention to it, and allows for a kind of dialetical process where I can address normal user's critiques that aren't voiced and also realize certain things I may have done mistakenly or glossed over on occasion. And it's a damn good feeling on those days when I can suddenly present solid data to blow the BS claims out of the water.
Here's a suggestion: Perhaps stop acting like you're the last thing standing between me and human decency, and instead start doing some testing of your own to either support your claims or otherwise complement my findings. Or if you're lazy or uninterested, which is perfectly fine, simply stop insisting on tired outdated BS like 'technicals still matter somewhat' or 'attributes are mostly linear'.
Yes, including one or two of my own you'll note.
It's literally just measuring the % disparity between two players where there shouldn't be one, and saying which has the largest or smallest error.
Testing shows that on average, Maeda finishes 3 position, Osimhen finishes 3.833 position. Yet your weights show Osimhen 88.50% FS 88.37% TS, Maeda 79.80% FS 69.06% TS.
That's embarrassing for someone who insists nothing has changed since they learned 'everything' 2 or 3 years ago. Or at least I'm embarrassed for you.
it's just randomness, that not prove nothing nothing nothing about atributes...c'mon einstein scientist (never wrong) get your scientific method together and don't do this bs
You are Just misguidind people for fun
I didn't mention it, because I thought the data self-evidently refuted the ol' 🤪 iTs AlL rAnDoM aNyWaY 🤪 argument.
As an aside, I'm enjoying these text color options.
We have heard it claimed for a long time that you cannot even measure, let alone employ, various factors in the game, because the inherent volatility is supposedly too great.
Yet observe the results I recorded: When I test Haaland, I don't have him suddenly finishing 9th or 16th. His lowest result in 6 samples taken is 4th. Doesn't seem very random to me. And it's precise enough to even distinguish clearly between him and Mbappe and various other contenders, who finish 2.5 or 3 instead of 1.666.
Now there is the valid point of seasonal variation throwing a spanner into the works, because realistically we're playing for each season, not waiting around for it to eventually get it right in season 2 or 3, even if season 2, 3, 4, 5, 6, 7 are all correct.
But I have calculated that variation and taken into account in assessing what level of certainty we can reasonably hold. I've stated before that I asked ChatGPT to combine this seasonal volatility, and the the current degree of statistical error, and it said we can be reasonably confident that to the point of +/- 7.8%. In other words, when I'm looking at GS for the best ST, I can have reasonable confidence that the best ST is going to between 83% and 90%, rather than 67% and 90% for GS default weights, or 60% and 90% for keithb's weights. I don't know if this sounds useless to you, but it doesn't to me!
More ground breaking results everyone; speed is important for centre backs as well as strikers!!
And there's more;
" 2/3rds of attributes continue to have zero effect on final season position" Who bloody knew!!
What are you even talking about when you say "The trouble with the 'pace/acc is everything' explanation is that it failed to account for players such as Kane and Lewandowski" What does that even mean?!?! They are accounted for lol.
More than half the stuff you say is just from your fantasy world. Its a complete fiction that you've created and then talk about. Its not real.
In your last testing you said jumping reach was super important for every position and now you've set it quite low. Before you said Mbappe was shit hahah. Now he's your 2nd best player lol. And lets not forget you said I was pissing in my own face for saying stamina was important haha. You said you didn't need it at all and now you say its 80%!! What level of reality are we supposed to be operating on? Because I cant operate on yours.
Keepers dont need technique
Forwards dont need concentration
Wingers and 10's can have jumping reach 1.
Centre backs dont need high stamina.
Rattling off a bunch of falsehoods doesn't make them true.
Unfortunately for you, EBFM's uploads on Youtube still exist there and can be referred to.
We see that work rate reduces in-match condition % identically to stamina. So stamina cannot be more important than work rate, or at least not more than ~20% more important say.
But the real showstopper to your argument is that in the video Max also states 'a player will use 15-20% OPC per half' while showing a match in progress that shows players that started with identical condition % and attributes, and the DL/DR condition % is indistinguishable from other positions.
That said, anecdotally I know that in standard play, fullbacks tend to use a bit more condition % than the other players. I suspect this is due to the 'best' fullbacks also tending to have high work rate. This is a kind of self-fulfilling prophecy: People use GS weightings that select for high work rate on fullbacks, so they end up using more condition %, requiring higher stamina, and therefore favoring stamina too. But very high work rate on fullbacks isn't necessary to begin with. Testing shows that work rate above ~11-13 is largely superfluous.
We also see that 1 > 20 sta is equivalent to 100% vs 97% condition difference when starting match. We know that bad natural fitness will result in an inevitable condition difference of at least that, yet you have natural fitness weighted at 0.
In regards to jumping reach, it remains important for every player. I don't know how you consider '46' weighting low. It was in fact viable at lower values of ~20-30, but I thought 46 was more appropriate and it aligned fine. I haven't yet examined if extra jump reach is necessary on DC, which is an important remainder matter to clarify.
In your weightings, Vinicius Junior is no.1 FS, Mbappe is 2nd beating Haaland, while you have to scroll down to find Maeda below a 32-year-old colombian at Orlando City. Comparable analysis showed that your weightings have worse variance than the Genie Scout default values from a decade ago. The question I have is not how did you come up with those ratings, but how did you manage to get them to do worse than even the Genie Scout default??
Why you keep using league position as the end point still baffles me. Yes it has less variance because there are only 20 possible spots compared to lets say 100 possible points. You have to go really out of your way not to be in a +-3 places range. Additionally there is a lot more volatility in the middle of the table than at very top or bottom.
There was decent data for GKs from FM Arena, Harvestgreen, Orion and probably others and they mostly match with what you found so it's not exactly groundbreaking. Still valuable additional data from a different approach.
Yeah.. look, I've heard so much rubbish from you that you're going in the keithb category for me now.
I feel like I've given you a fair go.
I responded to your claims several times with solid evidence to the contrary, whether it was about technical attributes (aside from drib) mattering at least somewhat as you insist they do when they don't, or your claim that attributes aside from work rate are 'pretty much linear', but you just ignored my responses - even when I reminded you that I had already responded to your claims.
It's understandable to not be in the know, or double down on something you've always believed in. I did that with personality distribution in FM. But I conceded the point after looking into it. You just seem intent to repeat the same spiel of 'you're wrong' over and over, and making claims that are easily refuted with a simple lookup of what I'm referring to:
harvestgreen22 said: 2.The goalkeeper is the least important person among the 11 players.
The verification method is to select any one of the player on the field, change the given attribute, and observe the extent of the impact on the winning rate.
The impact of the goalkeeper on the winning rate is roughly only one quarter or one fifth of that of the any one players on the field.
'There was decent data for GKs from FM Arena, HarvestGreen, Orion, and probably others' - Where is it? I've addressed the HarvestGreen data in the post you're directly responding to. C'mon man.
I will address your other point for the benefit of those who are being fair-minded and just want to know the truth of how FM works.
As you state, I use league position to assess because it has much less variance than league points. It is also true that there is will be more position volatility in the mid-table than the top 6, because the quality gaps between the top 6 are likely wider than the mid-table. However, the implicit idea here that league position only seems more reliable because it compresses the volatility is misleading - what actually happens is that league position is the first outcome, and league points seems simply tacked on as a narrative for whatever reason (it is at least consistent with this hypothesis). So for instance, a team will finish 2nd most times, but one season it might get 64 points, the next it will be 84 points. One season it will be 2nd, 14 points behind, the next it's 2nd only on goal difference. And this is not just some miracle oddity, I've observed this myself literally hundreds of times. Hence, position is a much more accurate/less volatile measure of performance - and how can you argue the league position you finish in (i.e. 1st), isn't the primary goal of fiddling around with attribute distribution? Who wants to reach 80 league points and finish 7th, rather than finish 1st regardless of the league points tally?
For whatever reason it seems to me that people take FM Arena's testing rather than HarvestGreen's as a starting point, so I'll begin with that:
So we see here that stamina is moderately, not highly, important. Specifically it is 29.7% of either acceleration or pace.
Two important things to note are that points is used a measurement rather than league position, and that we are measuring from 8 > 20 stamina. It's not clear what the makeup of the team or opposition is, but at least it's giving us a fair idea of what to expect in any case.
We see here in HarvestGreen's latest result for stamina that is it 27.1% of acceleration or pace going from 1 > 18.
For 6 > 18 it is 14.1% of acceleration or pace. In other words, stamina is never critical, but becomes even less important beyond '6'. He also has 1 > 6 results showing this.
In his initial attribute test shown here, stamina 10 > 20 is 30.9% of acceleration, which in my view is the most important result in regards to application to the Premier League.
In keithb's GS weightings, stamina is weighted at 60 for DLR/WBLR/DM/MC, 20 for AMLR/ST, and 0 for DC.
What this no doubt reflects is the now outdated myth that different positions require different sets of attributes. This is not an article of faith for me; I myself used to believe in, and initially employed, weightings that varied per position. Partly this was because each position has different CA weightings for attributes, and so high stamina for low CA cost on fullbacks seemed like a boon. I suspect that the myth has been generally perpetuated throughout the years with the CA weightings as a primary impetus, and that keithb isn't even aware of this - he seems to just think that stamina is what's necessary to win, simple as that. My mind was changed by the test results.
What my test results found, and seemingly implicitly HarvestGreen's & FM Arena's too, is that pace/acc is just as critical on DC as it is on ST, many attributes such as 'tackling' or 'passing' don't matter at all for any position, and if you try to test for positional differences, there doesn't appear to be any, or at least many. There might be still be some differences - I'm still wondering if pace & acc are of equal importance across all positions for instance, but overall we're talking small differences, not major ones.
Through brute force testing, I found that stamina of 13 is the essential requirement for the Premier League, while pace/acc is 15:
However, you can get away with as low as ~13 pace/acc (think Harry Kane, or Lewandowski), so long as stamina is ~15 or higher. Conversely, a 20 pace/acc player has less need of stamina, and can get away with 12 and no doubt lower.
So if you put all that together, basically stamina should be weighted at ~33% of pace/acc, is what I found.
So you can think of it as either 33% or 86% of acc, perhaps somewhere inbetween. So ~50% roughly. This was a change from the ~15% I previously used, which was primarily based on HarvestGreen's 14.1% 10 > 20 finding.
With my latest weightings, which wipe the slate clean of assumptions and simply try to align weightings with actual real player rankings supported by test results as evidence, I found that only somewhere around ~90% weighting for stamina worked - or ~90-110% of acc. This is far more than even keithb's 60% weighting for sta on fullbacks.
But why is that necessary, and why does it work if so? It is necessary because someone such as Maeda or Furuhashi has very few green attributes, yet ranks above the likes of both all-rounders such as Lewandowski and fairly well rounded high pace/acc players such as Osimhen. If you analyze it carefully (you need to compare multiple players in conjunction to find out it can't be his high work rate, det, aggression, etc.), the only way to make the rankings align seems to be to boost stamina to such a high weight. Perhaps I've missed some other option, but this where I'm at. As to the explanation as to why it is so, it makes more sense once you realize that stamina is preserving both pace and acc, so its value is doubled, and it is preserving the two most valuable attributes.
No one that I'm aware of has previously come to these conclusions about stamina. Even the overlap between this and the old 'stamina for fullbacks' myth doesn't extend this far. And it puts a big dent I reckon in the general pervading view that pace/acc is everything, which was always sort of heard of, but never really solid fact until HarvestGreen's findings. The trouble with the 'pace/acc is everything' explanation is that it failed to account for players such as Kane and Lewandowski. It would be (and still is by some even on this forum!) explained away as compensated by high technical skill such as passing, etc. which is a red herring - these technical skills really are completely useless, and the differences lie in other areas such as stamina, work rate, anticipation, and a handful of select others. 2/3rds of attributes continue to have zero effect on final season position.
Let's start with GKs:
HarvestGreen did 2 seperate tests on GKs.
December 2024:
April 2026:
The first thing I'll draw your attention to is that the results seemingly contradict each other. In the 1st test, acceleration does little to nothing. In the 2nd, it is the third most important attribute, above aerial reach which got a tick in the 1st test. Vision got 2 ticks in the first test, but does nothing in the 2nd test. There are more examples like this.
In skawkclsrn's case the issue is lack of samples, but here the issue is the methodology:
- The 1st test is using a team with 10 in every attribute, and increasing a single GK attribute to 20, then measuring the goals conceded. The first two aspects underlined are potential problematic factors, for instance what if a GK needs better reflexes and agility because the outfield is deficient in speed - perhaps with a speedy team, a GK only needs aerial reach to handle potshots he can see coming from afar. That is just a minor critique to show even solid tests can have pitfalls. The real problem is measuring the goals conceded, because as shown in my testing results, goals conceded is not at all a consistent measure of success, similar to how league points is all over the place compared to final league position. It will give some correlation, hence why agility, reflexes, and aerial reach all correctly have ticks, but that is just because teams who finish 1st obviously concede less goals than teams that finish 20th.
- The 2nd test is starting with the GK at '20' in all attributes and setting 1 attribute to '1'. This is different to how HarvestGreen does his outfield tests, which increase/decrease from '10'/'12' for all attributes, and it's obviously less realistic to begin with. But the main issue, only evidenced by my own testing later, is that having all attributes at '20' is likely compensating for and obscuring certain deficiencies. So we see that reflexes and agility are pretty much the only 2 attributes that matter, and even then, their individual impact is fairly minimal (12.8% at most).
HarvestGreen also stated that his observation is that GK is 20% of effect of an outfield player, if I remember correctly, which is basically saying the GK should be an afterthought and not really worth worrying about.
All of his tests also carry the same caveat that he is testing against opposition that has identical attributes. For some attributes, such as Jumping Reach, Aerial Reach, Strength, etc. I hope it's obvious how this could constitute a problem when it comes to translating to how things work in real leagues.
Now, coming to my tests. I brute force tested individual attributes for GK until I got a solid picture of what an optimal GK looks like. So what I would do is I would start with all attributes at 1, except a few obviously necessary ones like agil, aer, ref, etc. and measure the position. The initial result might be 13th. Then I would increase each attribute from '1' to '20' and measure the improvement. If no improvement in position, ignore it going forward. After learning what attributes matter and which don't, I would begin refinement, seeing how low I could get an attribute to find the 'sweet spot'. And I also start testing those attributes I took for granted at the start: agil, aer, ref. I would test a few potential combos (i.e. does aer only matter if jump is high?), but obviously it can't be an exhaustive search. Now what I would most often find is something like this:
'10' = relegated
'12' = relegated
'13' = 8th
'14' = 5th
'15' = 4th
'16' = 4th
'18' = 3rd
'20' = 3rd
Usually I would choose 14 to settle on. But if '13' is relegated, or '18' is 1st, then obviously I would likely change that to a '15' - while also considering realistic availability in the game.
Now these are the essential GK attributes I ended up with:
We see from my latest comparative testing of real players within the game, that this player would constitute something of a middling Premier League team GK. This GK finishes ~7th, the best finishes 4th, Barcelona's finishes 6.4, and a League 1 GK finishes 8.4.
One key attribute that is hidden from that screenshot is technique, which turns out to be quite critical for GK. You can also see this reflected in the database data for GKs - the better the GK, the higher technique he is generally given. I actually notice for the first time now that HarvestGreen actually gave a tick next to technique in his 1st test, though in the 2nd it is listed as 0.0%.
It's not clear yet exactly how much a GK accounts for as a % of a team's overall performance, but we can see the GK is more critical to success than HarvestGreen gave the GK credit for. My current estimation is somewhere around ~66-100% of an outfield player.
So we know my template works. Let's compare to what we 'knew 2 years ago', which is HarvestGreen's data and nothing else:
- HarvestGreen found vision and pace, and later reflexes, to be the most important attributes. I found vision '1', pace '13', and reflexes '11' are adequate, while aerial reach is the most essential attribute.
- HarvestGreen summed up that GK is has 20% impact of outfield player. My evidence shows it's much more than that, even the difference between relegation and a top 5 finish at times.
- Generally speaking, there was no solid data on GK at all previously. Now I've provided you with a precise template for GK that is backed by solid evidence in the form of testing both real & artificial players in a real and popular league (English Premier League).
If I've missed anyone else's findings on GKs, that was any good, let me know.
Alisson - 6th, 3rd, 4th, 4th, 3rd = 4 position
Thibaut Courtois - 3rd, 7th, 2nd, 5th, 8th = 5 position
Jan Oblak - 5th, 8th, 6th, 4th, 3rd = 5.2 position
Marc-Andre ter Stegen - 10th, 11th, 4th, 4th, 3rd = 6.4 position
Anatolii Trubin - 15th (sacked), 6th, 4th (sacked), 5th = 7.5 position
Jack Stevens - 13th (sacked), 4th, 9th (sacked), 8th (sacked), 8th (sacked) = 8.4 position
Sinan Bolat - 4th, 2nd, 19th (sacked), 5th, 14th (sacked), 9th (sacked), 12th (sacked), 5th = 8.75 position
I removed Courtois' starting injury to test him.
It shows that GK does in fact matter, though it does seem a lot more variable - in the downwards direction.
So although a GK can't really make your team win much more, they can in fact make or break your team.
Observations through player comparison:
- Jack Stevens is a mediocre all rounder who normally plays for Cambridge, and show you can't really get away with playing mediocre GKs
- Bolat is interesting.. he is like Stevens, but a tier above, but has green reflexes and jumping reach. Shows good reflexes aren't sufficient.
- Trubin, like Bolat, also significantly underperforms. He has green Aer, Jump and Handling. None of these are sufficient alone therefore, even with seemingly above average values in all other areas.
We know that the following GK finishes ~5th and with good consistency too:
This GK finished 7th:
It seems like only a select few of real GKs can do better than the above players.
If I filter for attributes of that second GK (excluding natural fitness), no players come up.
It seems that GK is about what is the least worst deficiency do they have. I suspect it is this way so that, unlike the outfield, you don't get great GKs who make teams come 1st by the mere addition of themselves. Perhaps this also explains why GKs can't get really high match ratings, because their best is merely adequate.
Trubin has a slightly deficiency in determination, pace, acceleration. Alisson is only lacking 1 pace. Oblak slightly lacks first touch, acceleration, and agility.
To verify the theory, I reduced the filter by 1 for the attributes and tested the worst one of the 4 players that turned up (1 was Ederson @ Man City, 2 were Alisson and his backup GK @ Liverpool, and the worst was Davy Roef from AA Gent).
Davy Roef - 13th (sacked), 4th, 14th (sacked), 6th (sacked), 1st = 7.6 position
High variation, but the average result aligns with extra 1 in key attributes finishing ~7th
BTW it's stuff like Alisson's backup GK just coincidentally having exactly the right set of meta attributes that shows me SI are aware of and intentionally making most attributes useless and misleading. If it was an oversight, players that look like duds but nonetheless perform well and are placed in the top teams wouldn't be an identifiable pattern.
Now I know quite a few people are also interested in what kind of GK gives the most clean sheets, or lowest goals conceded. So here are those stats:
Alisson - 13 cln, 48 con | 14 cln, 39 con | 15 cln, 40 con | 14 cln, 36 con | 11 cln, 44 con = 13.4 cln, 41.4 con
Thibaut Courtois - 9 cln, 41 con | 12 cln, 48 con | 10 cln, 56 con | 10 cln, 45 con | 11 cln, 50 con = 10.4 cln, 48 con
Jan Oblak - 8 cln, 49 con | 11 cln, 43 con | 10 cln, 59 con | 14 cln, 34 con | 15 cln, 40 con = 11.6 cln, 45 con
Marc-Andre ter Stegen - 4 cln, 58 con | 10 cln, 53 con | 11 cln, 52 con | 13 cln, 44 con | 14 cln, 46 con = 10.4 cln, 50.6 con
Anatolii Trubin - 10 cln, 46 con
Davy Roef - 14 cln, 45 con | 13 cln, 44 con = 13.5 cln, 44.5 con
Jack Stevens - 16 cln, 41 con
Sinan Bolat - 14 cln, 50 con | 10 cln, 52 con | 9 cln, 53 con | 8 cln, 54 con = 10.25 cln, 52.25 con
I only took stats from samples where there was no interruption by sacking.
Also, given what you said about Strength, Vision, Finishing, Teamwork and Loyalty, would you say these are attributes we should actually pay attention to when evaluating a player, or are they basically not worth considering unless the value is particularly low/high?
I’m especially wondering about Teamwork since you still seem a bit uncertain about it.
And one last thing: have you also updated the goalkeeper weights, or are these still the current ones?
Reflexes: 12.8, Agility: 8.0, Acceleration: 4.7, Pressure: 4.1, Pace: 3.5, AerialReach: 3.4
PS: What is the correct way to write it?
Was referring to the FSMT26 weights. For FMSS I divided by 5 because I think it had problems if you put the numbers too large from memory.
Those 5 attributes are not really worth paying attention to.. except loyalty of course, for non-performance reasons. Not only are they low value, they are non-essential. By contrast, agility is also low value, but moderately essential around ~10. Work rate is absolutely essential to have at least 6. You can have 1 strength/finishing/vision and finish 5th, that's proven.
Overall I don't think teamwork does anything. There's just a small element of uncertainty.
GK weights unchanged. I'm currently doing samples of GKs for rankings list.
I think either way you've typed it is fine.. just test each and see
Kriek said: Is it still needed to use the filters with the new ratings on Genie Scout or only the ratings?
Only ratings is fine. Filters still useful, but should be able to phase this out later by adding non-linear weights to FMST26 (not GS though).
Was that intentional as part of the new weighting method?
Finishing, strength, and vision were technically in the previous weights. I dropped them from the file because the attributes that are adequate you are bound to get at the required levels most of the time anyway:
Strength - 8
Vision - 7
Finishing - 7
I played around a fair bit with strength in particular in these new weights. I thought maybe it's around 10-20, but turned not to be the case. Vision is 20 on Messi which was food for thought, but it didn't seem to affect ranking alignment for the better.
Actually teamwork is the one I most wonder about.. players like Maeda have high teamwork, and there's some mathematical study on FM24 on github that a guy did for university work that found teamwork was a key correlate of performance.. but I read through it, and overall I think its just full of mistaken findings. My own testing and HarvestGreen's found teamwork doesn't matter, so at the moment I don't see a valid reason to include it.
Loyalty I included as '1' at the last moment pretty much symbolically. It would have some very minor to minor effect via morale, so i felt justified giving it a little '1'. But obviously loyalty deserves some weight for gameplay reasons.
The main thing would be the non-linearity of attributes, which the testing did show, but can't be reflected in linear weightings.
HarvestGreen did do some theorizing on attribute interactions, and I've dug up quotes from the Collyer brothers and in-game tips and whatnot which describe attribute interactions as well. I found that those latter things turn out to be generally untrue, or at least outdated.
I reckon there are attribute interactions though, and HarvestGreen's ideas on it are interesting.. just lack the solid data on it at the moment though.
I don't really have solid data to back it, but I'm personally convinced that aggression x dirtiness is an interaction, it just seems logical and seemed to solve something insoluble with the ratings at one point too.
This is partly why I decided to release it as is now with mostly linear weights, because one could go down the rabbit role of trying out all these different attribute combos and either getting nowhere or being mislead into seeing something that isn't there, while also spending a lot of time doing so. It remains difficult to test attribute combinations, but I'm sure there's some method to bear it out properly.
bf3metro said: Also, I noticed there are some strange extra spaces between a few attributes in the list. Is that intentional, or just a formatting issue?
Probably a formatting issue
bf3metro said: One other thing I'm wondering: is it possible that some of your original FMSS weights were simply set too low at the beginning, which would also partly explain why a few attributes have increased so dramatically now?
Do you think these huge increases are entirely a consequence of the different methodology, or could your original FMSS weights also have been underestimating some of these attributes from the start?
The FMSS weights I gave before simply reflect the minimum set of attribute values to attain 5th in the Premier League. They represent the best 'bang-for-buck' relative ratio of attributes.
I guess the question that the objective ranking raises in relation to that is, if work rate supposedly caps out at 13, why is it so critical for Maeda, and important in general? And if dribbling isn't so essential, is it really necessary to include it high - even if it does actually have significant benefits. So it does shine a new light on certain things. Personally I strongly favor what is true in realistic testing (the player rankings derived from the team positions), rather than what the isolated or constrained testing shows (including my own).
The 118 number is just to get it to display with the highest % at ~90%. It's the same as the overall weight option in GS.
From what I understand, the old weights were based on finding the cheapest/minimum attribute balance capable of achieving around 5th in the Premier League, whereas the new weights are being fitted to a much larger set of objective player rankings.
So it makes sense that the result could be very different: instead of asking “what minimum combination of attributes gets results?”, you're now asking “which attributes best explain the actual ranking of players?”
Yes, that is precisely the correct summary of it, and I daresay a better explanation than I did.
I'm glad you immediately understand, because I think it's easy to misinterpret this latest thing I'm doing with the weights.
bf3metro said: The new model seems to put an enormous premium on Pace, Acceleration, Stamina, Work Rate and Anticipation, while attributes such as Dribbling, Consistency, Important Matches or Natural Fitness become much less influential.
And then you have Finishing at only 0.4, which is pretty wild too.
I started with the presumption that finishing is almost completely unnecessary, but ChatGPT independently came to the conclusion that finishing should be removed too when I fed it all the player's attributes and asked it to come up a set of weights to fit the rankings.
It was surprising how strongly weighted anticipation and work rate seems to be. Maeda is still down a bit, so work rate could in fact do with being a bit higher perhaps. It's strange to me, because I know that (and you can see HarvestGreen's data to verify this) that work rate almost caps out at 13. But that is when work rate has been tested in isolation or partial isolation. Perhaps there is something like 'determination x work rate x stamina' going on. I couldn't find a way of getting around weighting work rate highly.
Anticipation in my observations did continue to scale, but had diminishing returns. But putting that aside for now, it makes sense that it would be a key attribute because ostensibly it's the movement ('off the ball' as speed?) to the area before possessing the ball. So it sounds like it cuts down on the pace/acc requirement to me, because once they have the ball, they have less distance to travel. Concentration would be a similar kind of effect, where low concentration leads to a delay in getting into position to receive or stop a ball (the latter offset by 'positioning' perhaps).
Consistency, Important Matches, and Natural fitness has not been incorporated with much accuracy. They are kind of afterthoughts, as I know they will matter somewhat, but not sure to what degree. I've used HarvestGreen's data to set them, and so long as they're not interfering with the ranking, I generally leave them alone. Same with pressure, professionalism, etc. - and the left foot & right foot figures I've just pulled out of my ass, to give a little extra weight to the modest footedness bonus which we know exists. This stuff can be refined further later. I think HarvestGreen found that foot should also match the side they're on, and I think FMST26 allows us to do this as well.
Dribbling I've been in two minds about. Most of the time I had it set pretty high at ~40, sometimes as high as 60. I ended up reducing it to 15. ChatGPT actually said early on to do this too. Both were fairly equally functional. From testing, we know dribbling is a very strong attribute that scales well.. I suspect it's not needed much to match the rankings, because hardly any players have high dribbling anyway. It's the hardest attribute to get high in the game I believe.
bf3metro said: PS: Quick question: why are attributes like WorkRate, ImportantMatches, and NaturalFitness written without spaces? Is that just the naming format, and do they need to be written exactly like that in app.js as well? Also, should I add a comma after the final value, Consistency: 2.82, or should the last entry be left without one?
Yes, that is how they have to be written for FMSuperScout, otherwise the program won't work. That is how labels the attributes internally.
I would add the comma.
Actually this is proving more difficult than I thought it would be
I'm not a programmer, so I don't know how to implement the injury proneness & dirtiness weights. I thought, maybe I'll just add them as simple negatives, like for the GS weights, and adjust it until it matches the rankings, but then I realized it doesn't do FM24 db, only FM26.
Here's the weights at least that will work:
Pace: 17.36, Acceleration: 17.24, JumpingReach: 9.24, Dribbling: 2.92, Pressure: 3.34, Balance: 2.7,
Concentration: 9.46, Anticipation: 11.64, Determination: 6, Agility: 3.4, Stamina: 18.78,
Strength: 0.2, Composure: 5.12, WorkRate: 12.6, Finishing: 0.4,
Aggression: 1.26, Professionalism: 3, Ambition: 0.4, Loyalty: 0.2, Vision: 0.1, ImportantMatches: 2.04, NaturalFitness: 1.58, Consistency: 2.82
Replace the weights in the app.js file with those (search for 'FM-Arena' ).
For adding injury proneness & dirtiness, which are in fact crucial (but I guess you could also just check players manually to have low inj and dirt values), we're going to need someone like Panneton0 or perhaps the creator of FMSuperScout itself to help out. Could probably do it myself, but can't be bothered doing an hour of research honestly.
Yeah, I left it out because I have to work out how to best translate it from the FMST26 weights first. I'll do that now.
Any insights about how newgen generation works under the hood?
Not stuff we already know generally about what effects newgens, but is your detection method say revealing anything that's going on?
This new file is a lot like mine LOL!
You have stamina weighted at 0 for DC, 20 for ST/TS/AMC/AML/AMR, and 60 elsewhere. It's not my fault your own weights contradict your own words.
Meanwhile, the key attribute of dirtiness you leave on 0, as you have 0 clue of course about what attributes actually matter beyond what we knew from HarvestGreen's testing a year or two ago.
But as I've said, you can have a variety of combinations that work to get to match the final ranking result. And at least you're implicitly accepting now that my objective rankings are correct.
Let's do a comparison of our two files to find out just how similar they really are. First, yours:
Junior is 1st for FS, even with 16 ST proficiency. Mbappe is above Haaland. Messi is 5% above Lewandowski and 10% above Maeda. Even Malen is above Maeda. Dybala is above Maeda and Furuhashi. Mbappe (8 jump, 8 head, 11 str) is a better target man than Glatzel (16 jump/head/str). It goes on.
Mine:
There are some minor issues, such as that Maeda should be ~3% higher, and that Malen & Glatzel should have a wider percentage gap between them.
keithb said: Why have you still used negative values and scored the hidden attributes in Genie Scout?
The only fair point you've raised in your entire posting history I believe
It is true that hiddens remain invalid for GS. I decided in the end to include them for my final GS adjustment at the end to try and closely match what I got in FMST26, halved for everything except dirtiness & injury proneness & natural fitness. Dirtiness & injury proneness are essential to the weightings, there's no way to do without them.
Because GS calculates incorrectly, the ratings remain a bit off compared to what you'll get in FMST26, hence I highly recommend FMST26, but I know many people aren't going to be willing to switch over.