Came across this glitch, don't think anyone has found it before?
How to get more than 2 match practice:
Make a schedule of 2x match practice, then apply it to a week with a match in it. Then copy & paste that week to an empty week, the match will be replaced with match practice (total 3x match practice). Then you can just fill it up with as many as you want.
I tested 21x match practice with my realistic Luton test for most of a season. Tremendous amounts of injuries, but in the few who escaped largely unscathed, their attribute increases were mediocre or even declined slightly.
I then tested it with the EBFM test file. Surprisingly hardly any injuries at all over an entire season; '1' injury proneness has a greater protective effect than I realized. Looking at attribute gains of one player that had no injuries, gains were good but not unusually great.
I'm still thinking of the implications here, but one thing is that these players have 20 professionalism. Could it be that professionalism alone accounts for why players in the EBFM league develop significantly while those in my Luton save do not..
tam1236 said: Worth saying that the ceiling is defined by country and league too. For example If You played as chinese club You could buy a high-PA chinese newgen generated somewhere else - or even could be lucky and get one, don't know if it's possible - but he will not progress in Your chinese club over a preset limit (about 70?). And same: if You played as Liechtensteiner club, You would get decent (but not great) newgens because You play in a swiss league but they dont cross 80 CA in Liechtenstein until they have second nationality (not sure here). You can loan players to Swiss club to get better CA it will not lower after return. Or wait until Liechtenstein/China/Andorra/sth else get better country rating after many succesful seasons. Expand How did you arrive at this conclusion
An example to disprove this hypothesis would be EBFM's test league which has North Korean players in an Algerian league, where North Korea is rank 111 and Algeria is rank 40, yet I've seen players at ~170 CA after 4 years of training
Using the default database, I've cleared out Luton's players (20th in Premier League) and replaced them with 88 players. That's overloading the squad a bit too much (29.33 per 1st/u21/u18 team), but coach workload was light to average, and I don't know about others but I tend to buy too many players anyway - in any case, it shouldn't change the results too much.
The players I choose were young players, age 16-23 (average ~19), who were pretty cheap to buy and met a lenient search filter criteria (i.e. for ST, ~11 pace/acc + 6 work rate). I would try to pick players with high CA-PA gap. So it's meant to be represent what you'd realistically be able to sign in game. Luton staff were kept the same, knap tactic used.
4 years meta (Quickness + Match Practice + 2 x Attack + Quickness/Agility focus):
GK - 4.75 agil, 1.25 aer, (-0.25) ref DL/DR - 2.14 acc, 3.69 pace, 0.07 drib DC - 2.85 acc, 3.69 pace, 0.5 drib DM - 2 acc, 3 pace AML/AMR - 1.71 acc, 2.76 pace ST - 1.58 acc, 3.25 pace
2 years full rest then 2 years meta:
DL/DR - (-1.81) drib AML/AMR - 3.2 acc, 4.4 pace ST - 3.8 acc, 5 pace
4 years meta with 5 weeks full rest at start of each season:
ST - 2.47 acc, 3.67 pace
I'll just cut to the chase about this. It's pretty clear to me that even full rest for just 5 weeks each season sucked too much due to the loss in mentals/technicals. But with meta, only 36 of 83 outfield players got to 16/16 and none got 19/19 or higher. I think at least 17/17 is necessary to dominate in premier league, and given there's a trade off in going for full rest, 18/18 I reckon is the sweet spot. With 5 weeks rest, 1 player reached 20 pace, 6 reached 18/18. I think ~3 weeks rest per season is probably best, and done in the pre-season so it doesn't result in poor performing players due to low match fitness during the competitive matches.
Now after I did these three tests, I used the EBFM test league to test different training schedules to try and find something superior. Eventually I did find a combo that works noticeably better than the meta, but then when I tested it in my Luton realistic test, it failed horribly with 1.87 acc, 2.2 pace and even significant losses in technicals/mentals. I noticed the EBFM file is using FM23 database, maybe this is the cause, but I doubt it, as it should still use the FM24 training values. Another reason I figured is that it could be that the number of matches effects it, so I tried a reduced training schedule that also worked about as good as meta - again I got 1.93 acc, 2.07 pace. Now neither of these employed match practice notably, so that could be the key factor here. It could also tie into the fact that the Luton players have less CA-PA headroom, worse personalities, and so on.. maybe the CA gain limit is too low to absorb all those gains, or maybe the low CA gain limit or poor professionalism changes the actual distribution of attribute gains (i.e. maybe poor professionalism player can only gain physicals, very little mentals/technicals). It could even be the coaches. Who knows.
So the main takeaway from that is that the EBFM test league, and probably all other artificial test leagues, will give different results to realistic gameplay when it comes to assessing training schedules.
Another reason why not to use full rest is because it's clear, albeit in EBFM test league tests, that those lost technicals/mentals are too difficult to recover even if you use 2 years rest then 2 years meta.
So I'm still working on trying to find a better schedule than the meta.
Seb Wassell, who designed the new training system, repeatedly emphasized that training changes attributes which changes CA (not CA first, then attributes), and I think he said that PA is then taken account which affects CA and then attributes.
I think how this works is:
+0.8 pace happens for that week first, this increases CA to 171 second, but PA is 170, so you get slight down arrows on almost all attributes (third) as they are reduced by the minimum change (0.2 points) to compensate for the PA cap. So 171 CA +0.8 pace becomes say 170 CA +0.6 pace, -0.2 tackling, -0.2 marking, -0.2 crossing, etc.
Matches generating CA/attributes themselves is an intriguing idea, but thinking about it, I think it's not true, because young players improve with zero matches. If you haven't seen them, I recommend having a look through this video and EBFM's video, as they have all the measured data on what affects player development.
Seb Wassell has said that both low and high match ratings reduce attribute gains. I just learned this myself going through his posts recently.
Though I would caution that Seb Wassell has been mistaken or has lied about several game mechanics, and that could be the case here too. But I would say that most of what he says is accurate, even when it comes to specific under-the-hood things like this.
I've been doing some training testing and my impression is that what is going on with match practice is that it does well because it has high CA growth per module. Other high CA modules also seem to do well, although not quite as good due to the specific attribute distribution match practice has.
My hypothesis was that high CA modules work best because rest sessions have their own CA value (i.e. pace/acc growth) so you want to cram as much technical/mental CA into as few modules as possible while not being too few. But it appears to not be as simple as this, as I tried to perfect the meta training using this idea, and yet Quick + 2xAttack + Match Pract + Quick focus still won out. But it largely holds true as a general principle.
I'm currently testing full rest vs the more balanced meta stated above, and combinations thereof over 4 years. I'll have these results soon, but my impression so far is that the more balanced meta is definitely superior. Not only does it give better match fitness (and therefore match results), but it actually produces better players overall, even if you do say 2 years full rest, then 2 years balanced. What I've noticed here is that it's easier to lose the attributes than to regain them.
You see, you are correct in your assertion that CA-PA gap matters a lot. And I think you, HarvestGreen, and I, are sharing the same general hypothesis here that since full rest boosts pace/acc greatly with minimal CA gain, and large CA-PA gap increases growth, it would be best to train pace/acc through full rest first, then do a balanced regime to boost CA greatly with technicals/mentals only at the very end. But my testing so far seems to be showing that those technicals/mentals you lose through full rest, you can't regain later through balanced training, and overall balanced training produces significantly better players. I would say that maybe full rest is preferable in cases where your players have low CA-PA cap to begin with, then it makes more sense perhaps.
One thing I've realized from all this, is that maybe CA-PA gap should matter more than anything else when you buy players. Because if you think about it, we're trying to use full rest to squeeze out extra CA-PA gap. You could actually calculate this precisely, I think it's somewhere around ~20 CA (~15 CA balanced vs ~5 CA rest, if rest for 2 years then balanced 2 years). Since balanced is simply better, you would be better off finding a player with 20 extra CA-PA gap, even if their PA is lower.
I realize that few would have the time and energy to read all of my posts, and that certain things in my initial post are now contradicted by things I've said more recently.
I'll fix and clean up everything in an update of my initial post, but that'll probably have to wait until after Christmas.
I feel like I've come to pretty solid conclusions on everything that matters now except the matter of practical training.
So my plan for the next few days is to create a proper test of training, where I take Luton and/or Man City in Premier League and replace their players with realistic value buys (cheap, high CA-PA difference, etc.) using my GS ratings file, and apply the meta training as well as other training methods. My aims are:
1) See if meta & rest training actually works as intended or not (I actually kind of suspect it doesn't work as well as HarvestGreen22 assumes it does, based on some tests I already did) 2) Conclude what actual boosts will be had with meta & rest training 3) Try and find the ideal training regime for realistic use
Yarema said: No idea where you are getting this from because majority of players are under 12, especially younger players Expand I thought so too, but tam1236 pointed out to me that this is not the case
..Although halfway through writing up this reply, I realize we are talking about two different things here - you have in mind absolute abundance, I intend to mean relative abundance. Tam1236 was saying that I shouldn't favor pro over det, because low pro is significantly rarer than low det.
Nonetheless I think just laying out the matter will bring everything together.
Opening up a save I have where I have 140k players loaded, 1.5 years in (I thought I was a few years in before, but its enough for 1 year of newgens at least), I find the following using Genie Scout:
Total players = 139,514 Total players newgen only = 14,033
Det 1-6 = 27,235 players Pro 1-6 = 1,870 players
Det 1-6 newgen only = 4,185 players Pro 1-6 newgen only = 524 players
Det 1-8 = 41,228 players Pro 1-8 = 7,998 players
Det 1-8 newgen only = 5035 players Pro 1-8 newgen only = 1045 players
Det 1-12 = 88,614 players Pro 1-12 = 104,391 players
Det 1-12 newgen only = 9721 players Pro 1-12 newgen only = 9328 players
So you are right that majority of players are going to be under 12 pro, however we don't need 100k, or even 1k of players, when we are filtering.
For an absolute filter I suggest 8 pro. This leaves 94% of players while getting rid of the bulk of the poor performance.
For a relative filter (Genie Scout attribute weightings) I suggest weighting pro at ~2/3rds of det, because while their performance and training impact is roughly similar overall (beyond ~7), your chance of having low det players in your search is much higher. I guess you could kind of think of this as the inverse corollary to the 'absolute' filter; it is designed to cull as many players as possible from the list while minimizing performance loss.
tam1236 said: I think it was HarvestGreen22, who proposed, after testing , not to accept field players under 7 work rate because of fatal negative impact. This attributes in italics for these positions (presume the rest are ok too) will be actually nearly impossible to find outside the editor or top stars, I'm afraid. And with such unreal demandings You still didn't get jumping for ST and only 13 pac/acc and 12 for DC? Expand This time I believe I am not mistaken.
If you look at my 1 CA templates, you'll see I use at least 7 work rate. Not because 7 is required, but because mentals/technicals can decrease as training redistributes attributes. I am, as you are, taking the 6 minimum from HarvestGreen's data. In this case I am still using the extra 1 point buffer, as in my ideal test of the meta training work rate was +1-2 after 4 years, and as mentioned I'm assuming you will get ~2/3rds of ideal training results, so 6 filter denotes 7 after training.
The attributes in italics are in italics for the reason that they are difficult to find, but they are not impossible to find. I have taken them from my ~115 CA ideal templates, which in turn are taken from a search of the database in Genie Scout to find attribute values for which there are at least ~50 players available. If I filter for attributes for DC, including the drib 13, there are 9 results, 1 of which is age 28 and valued at $150k pounds. If you remove the drib 13, as I recommended, I find 245 results.
As shown in my 1 CA results, jump reach of 1 scored best for ST. Not because 20 isn't better of course, but because it isn't strictly necessary. The reason jump 15 is included for AML/AMR is because I believe it is strictly necessary if want to get the bulk of performance. Yes I tried 20 jump ST with 1 jump AML/AMR. Yes I tried 20 jump with 20 head, 20 jump + head + strength, and so on. All failures. The poster asked for the minimums for a filter - 12-13 pace/acc results in ~16-17 pace/acc after training, which is the bare minimum I found to be able to do very good in the Premier League.
tom8 said: Thanks for the update and the file, will test it out!
Just have to ask on the topic of certain attributes, do you have like a summary of what you would consider the minimum viable value for certain attributes. For example, you have the example of 12 for professionalism being what you would consider a minimum value for 'good' player.
I guess what I am asking is to combine your ratings file with a filter for certain attributes as well, although file will do a great job by itself and this may be unnecessary. Apologies if I've missed it in another post. Expand
GK - agil 11, pace 4, work 6, aer 15, ref 9, comp 4 DL/DR - acc 13, pace 13, jump 15, work 6 DC - acc 12, pace 12, jump 16, drib 13 DM - acc 13, pace 13, work 6 AML/AMR - acc 13, pace 13, jump 15, pos 13, work 6 ST - acc 13, pace 13, work 6
Minimum 0 CA attributes for all positions (these I'm more guesstimating than the above): 6 det, 6 nat, 8 pressure, 8 pro, 14< dirt, 14< injury, 6 loyal, 8 consist, 6 imp match.
Attributes in italics will be difficult to find and you should probably leave out of your filter, even though they are important.
The idea here is that with training, you will reach what I consider the minimums for what accounts for the bulk of max performance. I actually wanted to make 'pre-training' templates, but I found the effects of training to be too variable when comparing idealized isolated tests vs realistic premier league use, and I don't know yet what is actually realistic or typical. But with this filter here, I'm assuming you will get ~2/3rds of ideal training over 4 years, resulting in pace 12 > 16, which is the bare minimum I believe to get tangible victories (for premier league). It may fall short, but it will still serve it's function of separating most of the chaff from the wheat.
In regards to pro 12 being necessary, what I meant was that primarily pro 12< is rare, and we also know the following (source: Orion's FM22 testing):
So actually it's pro ~10 that is necessary, but I would guess that ~8 is not too bad either, based on how work rate matters a great deal up to just 6 then it tails off.
YildizAli said: Where do I paste this? Expand .grf file goes into (FM Genie Scout folder) > Ratings. Then in Genie Scout, press top left button > Rating and load the file.
For the FM26 version I've simply made some changes to a few of the attributes based on HarvestGreen's FM26 testing, so it may be less accurate.
I've also updated WB/MC/AMC so they're reasonably usable now, but they're just copies of FB & DM.
The main changes have been to the hidden attributes. Some very important changes here.
For those interested, here is also some rationale behind the weightings I've chosen:
Let's take long shots as first example, because I've given long shots very low weighting (1 or 2) in spite of both HarvestGreen22's findings (low moderate positive effect, similar to finishing or composure, which I give around ~8) and the fact that long shots have low CA cost in pretty much every position. Well in my 1 CA testing, I found that switching between 20 and 1 long shots made no observable difference in performance. And when I looked where goals were scored from in the in-game analytics, virtually none from long shots, and my DC with 20 long shots scored 1 or 2 at most over the season. And then there is also my own purely speculative theory that the knap tactic isn't designed for throwing the ball away on distant pot shots, and even if it is, it probably requires high levels of 'vision' and whatnot to pull off. Contrast all this info to pace, which it's clear as day even 1 point extra in makes a difference. So I figure my starting point is, I wouldn't trade 1 point of pace on a DC for 20 points of long shots - that makes it's weighting a maximum of '4' when pace is '100'. Now if I look at DL, long shots has a CA weight of just '1' and I gave it weighting of '1' in genie scout, whereas for DC I gave it '2' - and of course, neither of these are '4'. Consider that for DL, dribbling is also '1' CA weight, and it does actually matter. But I also have to weigh up what other attributes take precedence before long shots. Take strength - it's a fairly costly attribute ('4' weight), but it demonstrably has at least has some performance effect. Would I rather 5 strength on my DL, or 20 long shots? I think the 5 strength is going to perform a little better, but I wouldn't even trade 20 strength for 1 pace partly because of performance, partly because of the high CA cost (we can't have everything). So I give pace 100 weight, strength 3, and long shots 1.
Now if you read all that, you will realize that what I'm doing is quite subjective and prone to imprecision. I prefer precision, and this is largely possible with attribute testing, but unfortunately filtering methods don't work that way - the more precise you are, the less results you get, or conversely you start trawling through a few hundred laxly bounded results but only get as far as the 7th player you pick out and unwittingly choose a 17 long 16 pace 15 acc player over a 8 long 18 pace 18 acc player that was further down the list because it was too tedious to go through them all. Filter weightings alleviate this somewhat by creating a kind of hierarchy of attributes, so those 18 pace/acc players will be near top of your list no matter what. But if you assign too much high (or even moderate) weights to too many attributes, you muddy the hierarchy and end up with at the top of your list a complete dog's breakfast of a player that is 18 long 16 tack 17 str 12 pac 14 acc 20 ant and apparently a 78.41% rating.
And sometimes the attributes are almost purely subjective or circumstantial in ways you cannot pin down to a precise number. For instance, natural fitness technically doesn't affect performance much, so long as you manage it right - but managing a squad's match fitness I find requires you to have say ~30% of your players with high natural fitness so they can skip games to allow for a sizable squad without ill consequence. Loyalty.. I'm not aware it has any impact on performance at all, yet we all know the value of a loyal player - personally I figure I'd be willing to trade 1 each of pace & acc for it. For someone else it might be more or less, but I have to put some figure there.
One more I will make an example of is professionalism. I've reduced it to '16' in this update, which is relatively low now. Professionalism is actually very important for both performance and training, and it's 0 CA cost. But it turns out professionalism under 12 is quite rare, and the benefits of 12 -> 20 are much less than 1 -> 12. I do calculations based on all of this info to get a true relative figure. Although again a bit wishy washy, this type is actually quite solidly grounded in the performance stats.
Didico said: What is the average rating of the players that you bought? Expand Taking one from the middle and applying latest genie scout ratings I haven't uploaded yet (position is AMR):
mohamed salah 77.28% messi 71.13% jack grealish 66.83% my player 66.30%
rank 102nd
Highest valued player (ST) was:
mbappe 77.59% haaland 77.25% raheem sterling 69.09% my player 68.08%
tam1236 said: Check how many do you have newgens/players with det<5 and then how many with prof<5 - then we can talk - btw 9/10 are medium values - of course if tests are not only editor generated theory. And not only PA>140 because this filter information is nothing. Especially if applied to newgens. Expand Interesting.. I had assumed that after they got rid of national personality templates, they replaced it with a purely random assignment of them.
SI said at the time:
Rather than using a template based on the averages of existing players of a certain nationality, all personality attributes of newgen players are now generated randomly for every nationality
The new system Sports Interactive moved to abandoned the nation-based averages and replaced them with a system that gives newgens a random set of attributes based on their role and position
If I search for pro 5< there seems to be an equal representation of positions, but I do notice that 80% of them also have det 5<. I also notice that for both 5< pro and >18 pro, there are often repeating clubs (i.e. 3-5 players from the same club in the list).
One club I looked at (San Francisco City FC) that had 2 x 19 pro newgens, they had no staff, not even a manager. They have terrible facilities. I couldn't work out any possible correlates.
You are quite right in the points you make and there are indeed a lot more low 5< det newgens than 5< pro newgens. They seem to converge at ~12, and going by Orion's FM22 test of hidden attributes, pro actually matters less than det at this point (20 vs 10 pro = +5.6%, 20 vs 10 det = +12.7%, whereas 10 vs 1 pro = +46.3%, 10 vs 1 det = +10.2%). It holds true for existing players too, not just newgens. Then again, from memory pro does better for training than det.
Putting it all together, I guess det should be weighted in genie scout say 1.5x pro.
Bogdan said: yeah, been scheduling some friendlies here and there in international breaks. i manage a second league team and all the players are available in these international breaks. what i did find is that i need to choose carefully the opposing team in a friendly to not be such an easy match. i found that in matches against team that i demolish match sharpness doesn't improve. thank you and keep up the good work! P.S. i'm hoping to soon find a guide you create based on your findings to essentially assemble and manage a great team Expand I may be wrong, but I believe that opposing team quality/play doesn't affect match sharpness gain. I suspect what you are observing is the fact that friendlies only increase match sharpness by 60% of competitive matches (if I remember correctly). As match sharpness gain slows the higher it is, friendlies become less useful but still usable at match sharpness >90%. I've actually noticed that playing against minnows is best because it boosts morale, which is beneficial to future performance.
That's a 32% drop in performance. It is important to note though that Orion found the difference between 10 pressure and 20 pressure to only be 10%, but even that is significant.
This doesn't effect my templates, as they have 20 pressure, but I will have to update my Genie Scout ratings file.
The other one I'm particularly interested in is important matches. Orion said important matches makes ~6% difference. My results:
115 CA templates important matches 1 - 107, +168 | 108, +175
I forgot to mention that I did a quick test of my Genie Scout ratings before posting, as I'm really not sure if it works as intended.
I took Luton (relegation candidate in premier league), removed their players, and added 15 players who cost 3mil pounds or less who were high in the Genie Scout ratings. Average value *after* transferring to Luton was 3.1mil pounds. Value of top 15 default players was ~10mil pounds average. Using the knap tactic and blue set piece routines, results were:
Unlike the attribute template results, this isn't clear cut enough to be certain it's working as intended. Seems like it's doing alright though.
Samet2772 said: **Hello, if I use this file in FM 26 Genie Scout, will it work?** Expand I assume the ratings files are compatible between versions
My data is for FM24, but I don't think much changed for FM26. I heard long shots got a boost in particular, so maybe make long shots '3' instead of '1' say.
Bogdan said: I've been using your training schedule and got some nice results for half a season but I'm losing a lot of match sharpness between matches. is there any way around this? Expand I've managed it manually myself in testing successfully. It can be a bit tricky, but it's definitely doable, although I played with Man City which have a lot of fixtures. I had to schedule some friendlies during the times when there were 2 week gaps, and I suggest you do the same to maintain match sharpness (you could do it as reserve matches as well).
I would caution against simply adding more training sessions in, as I found it worsens training results in most cases, I would guess because you're replacing a rest session (high distribution to pace/acc) with a training session (distributes more to technicals/mentals) rather than simply 'adding' the training session on top.
Tested the old templates to compare, and also tested the new templates without attribute availability limits (i.e. 20 agil for GK instead of 17; 1 decisions instead of 6):
119.6 CA average (new templates) no limits - 114, +223 | 114, +204 Old templates 139.2 CA average - 110, +221 | 112, +172 | 114, +177 | 114, +167 | 114, +206
To make it fair I made all 0 CA attributes for the old templates the same, except flair and aggression. The results were a bit all over the place, not sure why.
+182.2 for new templates, +188.6 for old templates, but +17% CA cost for old templates, so overall the new templates are about ~14% better.
My 1 CA testing was run with the same conditions, so you can also add this to compare:
Didico said: Im using George's ratings and filtering the best U22 CB's that are south american, under value 15M. I have to go to best DC Rating, am i doing right? Expand Correct
A few extra things I left out of my post:
Genie Scout ratings will be ~70% max. This is necessary, don't try and change this, I put the weight as high as I can but Genie Scout limits it to 120 so I can't make it nicer.
The height values I give on the positions mean something. They are the minimum height you need to have a chance of getting 20 jumping reach off the bat, but more importantly what you need if you intend to train to 20, as the meta training only increases jump reach by ~2 over 4 years. Kudos to ClaudeJ for pointing out this interesting find that comes from an Italian official FM researcher guide page. They tell the low-level researchers to assign jump reach according to this table:
And Aerial Reach for GK:
I put pace/acc as 20 in my templates, but I'll re-emphasize that the rest of the attributes represent high/low attribute values in existing players. I think it's kinda useful, because while we know now that 1 decisions and 20 dribbling is ideal, what we really need to know is whether or not a player with 6 decisions or 14 dribbling is the best we're going to be able to realistically find. With these templates, you can look at a DC with 13 dribbling and realize, this is actually a valuable find.
Kma said: Please, can you explain that altering knap tactic (i.e. 'very attacking'/'defensive', 'work into box', etc.) was worse than default ? Expand I would change the Knap tactic slightly in those ways, i.e. just put 'work ball into box' on, to see if it would boost my results. None of my changes did, the tactic was best as-is.
max 737 said: How do you get to this screen? I am getting this option of IP and OOP roles? Expand I wouldn't know, I'm not using FM26
Sweeper is GK with captain emphasis. Target striker is ST with penalties & free kicks emphasis. WB, MC and AMC I've left unchanged from my first genie scout ratings version, so they're outdated, but they're not bad if you want to search those positions.
Two quotes spring to mind:
You go to war with the army you have, not the army you might want or wish to have at a later time.
As we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns — the ones we don’t know we don’t know.
There are several known unknowns in my templates. For instance I know from my 1 CA testing that balance of 17 on my ST is probably unnecessary for performance. But I don't have the time and energy to test and verify every attribute, so I employed the following method:
1. Use and verify the key attributes identified by my 1 CA testing 2. Test and verify a bunch of remaining high CA weight attributes 3. Leave remaining attributes as-is that got best results just before my 1 CA testing
So I know for certain that these ideal templates could be better, but they are nonetheless say ~90% peak performance, where balance 17 may be too excessive and flair 12 may be too low.. or high.. and corners 12 may be just plain stupid - but these attributes will account for just a small handful of CA. Perhaps sometime later I will fix these lesser attributes up.
I also realized that my 1 CA templates distill the key attributes better. So I thought I would instead use these templates to illustrate not just the key attributes, but also what's actually realistic. So with my ~115 CA templates, attribute values are bounded by what players are actually available - there had to be at least ~50 players available with each particular value in the specified position.
That doesn't make my templates worse performing, I haven't compared them to the old ones specifically yet, but they should be significantly better.
I've isolated and optimized things better, so it's not comparable to the old results.
Here's the performance data (English Premier League, as Man City):
Outfield 20 acc/pace/jump/drib 12 other visible (133 CA outfield average) - 111, +138 115 CA template (118 CA outfield average) - 112, +198 | 109, +178 | 114, +169 | 114, +183 | 110, +183 Outfield 20 acc/pace/jump/drib 14 other visible (169 CA outfield average) - 114, +251 200 CA template - 114, +409
For those interested, here are the results of some attribute changes:
DC pos mark agil tack 1 - 114, +167 DL/DR DC DM mark 1 - 112, +175 | 114, +175 GK hand acc bra con communic command pos 1 - 112, +182 | 112, +195 outfield bal 1 - 110, +163 | 112, +173 outfield ant 20 - 107, +169 | 110, +170 no preferred moves - 114, +170 | 107, +170 outfield ant 1 - 110, +127 | 110, +135 GK str one pass bal 1 - 109, +160 | 111, +173 outfield 1 agil - 109, +144 flair 20 - 112, +173 teamwork 20 - 114, +158 first str 20 - 114, +242 | 114, +223 first 20 - 114, +198 | 110, +187 AML/AMR ST first 20 - 111, +176 DL/DR DC DM first 20 - 110, +182 str 20 - 111, +216 DL/DR DC DM str 20 - 110, +190 AML/AMR ST str 20 - 114, +207 DL/DR DC 7 pos (reduced from high pos) - 110, +198 | 114, +177 | 112, +187
I've highlighted the most surprising results. Note: These should really be compared against [112, +187] as I made some adjustments (i.e. GK; attribute decreases) after these results.
So preferred moves don't seem to matter; not much at least. Anticipation matters quite significantly. Agility matters moderately, but would be outweighed by its expensive CA cost. High flair doesn't seem to be bad, but perhaps not beneficial either. GK acceleration, and a bunch of other GK specific attributes, seem to hardly matter at all. Strength matters moderately, and only on forwards, but probably outweighed by high CA cost. First touch has minor to moderate effect, but it costs too much like strength. In the end I decided not to boost first touch + strength, the CA cost isn't worth it, especially when you consider that CA-PA difference is crucial to training growth (if your ideal player is 140 CA, that means you actually need say ~170+ PA if you want him to realistically grow into it). As you can see I tested balance even though its only weight is only 2, and lowering it resulted in worse performance, so I decided to just leave it high, even though the real optimum I would guesstimate is ~12.
These are from me testing attributes I wasn't sure on. A lot of attributes I already have a solid idea about from my 1 CA testing.
In regards to the Genie Scout Ratings, it's quite radically changed from my previous version. The way I decided to approach it is this: An attribute is worth as much pace/acc I would intuitively be willing to swap for it. I've been working with the data and reckon I have a good feel for it, so I trust my intuition here. So for instance, if I think about dribbling for AML/AMR, I know dribbling is important and also hard to train, but I also know that under 17 pace/acc is a no-go no matter what the dribbling is. So really what I'm saying is, 20 drib = 6 pace/acc. So drib = 30% (6/20). Then I might apply a very minor adjustment, in this case to 29%, for other factors I feel are relevant.
So in my refreshed genie scout ratings, most attributes are reduced a lot compared to pace/acc. The one that makes me a bit uncomfortable is jumping reach. It can be jumping reach is around about as valuable as pace/acc perhaps, but my thinking is that pace/acc does very good without jump, whereas jump without pace/acc sucks. Also jump only matters a lot if its very high, I would guess. I think the limits are 15 pace/acc + 20 jump for DC (20 jump = 10 pace/acc), 18 pace/acc + 20 jump for AML/AMR (20 jump = 4 pace/acc), and so I weighted it accordingly. There are a few other attributes I want to reassess, such as 'pressure' and 'important matches', but this is where I'm at right now.
flob said: I am a bit late to the party, but I have decided to play FM24 again and wait with FM26. Anyway, I was wondering if I also should team train my U21 and U19 teams + if I should individual train those players. If so, do I use the same team train schedule or a different one? Expand Same training should be applied to all teams and players
However I see now the reason why HarvestGreen suggested full rest for young players before switching to a more balanced regime once they reach near 20 pace/acc. They need the pace/acc first to play good, but also the balanced regimes can be simply too slow to build pace/acc.
Even in a perfect test environment with 5 star coaching, pace/acc with the more balanced training is +6 after 4 years. In a more realistic test I did, it was only +1 acc and +2 pace after 3 years.
Part of what is going on here is that growth is highly influenced by CA-PA difference, so you really don't want to pump all that technical/mental CA into your player until the low CA pace/acc growth is largely done.
I haven't tried it myself yet, but I imagine the best way to go about it would be this:
First team - Quick + Match Prac + 2xAttack; full rest for congested weeks U21s - Quick + Match Prac + 2xAttack U18s - Full rest
Age 15 -> 17 (U18s) = +5 pace/acc Age 18 -> 20 (U21s) = +2 pace/acc
The thing is, certain key technicals such as dribbling are very hard to gain, so I think you're shooting yourself in the foot if you choose to start by losing drib points with full rest. But pace/acc is more important than drib. So it's debatable.
How to get more than 2 match practice:
Make a schedule of 2x match practice, then apply it to a week with a match in it. Then copy & paste that week to an empty week, the match will be replaced with match practice (total 3x match practice). Then you can just fill it up with as many as you want.
I tested 21x match practice with my realistic Luton test for most of a season. Tremendous amounts of injuries, but in the few who escaped largely unscathed, their attribute increases were mediocre or even declined slightly.
I then tested it with the EBFM test file. Surprisingly hardly any injuries at all over an entire season; '1' injury proneness has a greater protective effect than I realized. Looking at attribute gains of one player that had no injuries, gains were good but not unusually great.
I'm still thinking of the implications here, but one thing is that these players have 20 professionalism. Could it be that professionalism alone accounts for why players in the EBFM league develop significantly while those in my Luton save do not..
How did you arrive at this conclusion
An example to disprove this hypothesis would be EBFM's test league which has North Korean players in an Algerian league, where North Korea is rank 111 and Algeria is rank 40, yet I've seen players at ~170 CA after 4 years of training
Using the default database, I've cleared out Luton's players (20th in Premier League) and replaced them with 88 players. That's overloading the squad a bit too much (29.33 per 1st/u21/u18 team), but coach workload was light to average, and I don't know about others but I tend to buy too many players anyway - in any case, it shouldn't change the results too much.
The players I choose were young players, age 16-23 (average ~19), who were pretty cheap to buy and met a lenient search filter criteria (i.e. for ST, ~11 pace/acc + 6 work rate). I would try to pick players with high CA-PA gap. So it's meant to be represent what you'd realistically be able to sign in game. Luton staff were kept the same, knap tactic used.
4 years meta (Quickness + Match Practice + 2 x Attack + Quickness/Agility focus):
GK - 4.75 agil, 1.25 aer, (-0.25) ref
DL/DR - 2.14 acc, 3.69 pace, 0.07 drib
DC - 2.85 acc, 3.69 pace, 0.5 drib
DM - 2 acc, 3 pace
AML/AMR - 1.71 acc, 2.76 pace
ST - 1.58 acc, 3.25 pace
2 years full rest then 2 years meta:
DL/DR - (-1.81) drib
AML/AMR - 3.2 acc, 4.4 pace
ST - 3.8 acc, 5 pace
4 years meta with 5 weeks full rest at start of each season:
ST - 2.47 acc, 3.67 pace
I'll just cut to the chase about this. It's pretty clear to me that even full rest for just 5 weeks each season sucked too much due to the loss in mentals/technicals. But with meta, only 36 of 83 outfield players got to 16/16 and none got 19/19 or higher. I think at least 17/17 is necessary to dominate in premier league, and given there's a trade off in going for full rest, 18/18 I reckon is the sweet spot. With 5 weeks rest, 1 player reached 20 pace, 6 reached 18/18. I think ~3 weeks rest per season is probably best, and done in the pre-season so it doesn't result in poor performing players due to low match fitness during the competitive matches.
Now after I did these three tests, I used the EBFM test league to test different training schedules to try and find something superior. Eventually I did find a combo that works noticeably better than the meta, but then when I tested it in my Luton realistic test, it failed horribly with 1.87 acc, 2.2 pace and even significant losses in technicals/mentals. I noticed the EBFM file is using FM23 database, maybe this is the cause, but I doubt it, as it should still use the FM24 training values. Another reason I figured is that it could be that the number of matches effects it, so I tried a reduced training schedule that also worked about as good as meta - again I got 1.93 acc, 2.07 pace. Now neither of these employed match practice notably, so that could be the key factor here. It could also tie into the fact that the Luton players have less CA-PA headroom, worse personalities, and so on.. maybe the CA gain limit is too low to absorb all those gains, or maybe the low CA gain limit or poor professionalism changes the actual distribution of attribute gains (i.e. maybe poor professionalism player can only gain physicals, very little mentals/technicals). It could even be the coaches. Who knows.
So the main takeaway from that is that the EBFM test league, and probably all other artificial test leagues, will give different results to realistic gameplay when it comes to assessing training schedules.
Another reason why not to use full rest is because it's clear, albeit in EBFM test league tests, that those lost technicals/mentals are too difficult to recover even if you use 2 years rest then 2 years meta.
So I'm still working on trying to find a better schedule than the meta.
I think how this works is:
+0.8 pace happens for that week first, this increases CA to 171 second, but PA is 170, so you get slight down arrows on almost all attributes (third) as they are reduced by the minimum change (0.2 points) to compensate for the PA cap. So 171 CA +0.8 pace becomes say 170 CA +0.6 pace, -0.2 tackling, -0.2 marking, -0.2 crossing, etc.
Matches generating CA/attributes themselves is an intriguing idea, but thinking about it, I think it's not true, because young players improve with zero matches. If you haven't seen them, I recommend having a look through this video and EBFM's video, as they have all the measured data on what affects player development.
Seb Wassell has said that both low and high match ratings reduce attribute gains. I just learned this myself going through his posts recently.
Though I would caution that Seb Wassell has been mistaken or has lied about several game mechanics, and that could be the case here too. But I would say that most of what he says is accurate, even when it comes to specific under-the-hood things like this.
I've been doing some training testing and my impression is that what is going on with match practice is that it does well because it has high CA growth per module. Other high CA modules also seem to do well, although not quite as good due to the specific attribute distribution match practice has.
My hypothesis was that high CA modules work best because rest sessions have their own CA value (i.e. pace/acc growth) so you want to cram as much technical/mental CA into as few modules as possible while not being too few. But it appears to not be as simple as this, as I tried to perfect the meta training using this idea, and yet Quick + 2xAttack + Match Pract + Quick focus still won out. But it largely holds true as a general principle.
I'm currently testing full rest vs the more balanced meta stated above, and combinations thereof over 4 years. I'll have these results soon, but my impression so far is that the more balanced meta is definitely superior. Not only does it give better match fitness (and therefore match results), but it actually produces better players overall, even if you do say 2 years full rest, then 2 years balanced. What I've noticed here is that it's easier to lose the attributes than to regain them.
You see, you are correct in your assertion that CA-PA gap matters a lot. And I think you, HarvestGreen, and I, are sharing the same general hypothesis here that since full rest boosts pace/acc greatly with minimal CA gain, and large CA-PA gap increases growth, it would be best to train pace/acc through full rest first, then do a balanced regime to boost CA greatly with technicals/mentals only at the very end. But my testing so far seems to be showing that those technicals/mentals you lose through full rest, you can't regain later through balanced training, and overall balanced training produces significantly better players. I would say that maybe full rest is preferable in cases where your players have low CA-PA cap to begin with, then it makes more sense perhaps.
One thing I've realized from all this, is that maybe CA-PA gap should matter more than anything else when you buy players. Because if you think about it, we're trying to use full rest to squeeze out extra CA-PA gap. You could actually calculate this precisely, I think it's somewhere around ~20 CA (~15 CA balanced vs ~5 CA rest, if rest for 2 years then balanced 2 years). Since balanced is simply better, you would be better off finding a player with 20 extra CA-PA gap, even if their PA is lower.
I'll fix and clean up everything in an update of my initial post, but that'll probably have to wait until after Christmas.
I feel like I've come to pretty solid conclusions on everything that matters now except the matter of practical training.
So my plan for the next few days is to create a proper test of training, where I take Luton and/or Man City in Premier League and replace their players with realistic value buys (cheap, high CA-PA difference, etc.) using my GS ratings file, and apply the meta training as well as other training methods. My aims are:
1) See if meta & rest training actually works as intended or not (I actually kind of suspect it doesn't work as well as HarvestGreen22 assumes it does, based on some tests I already did)
2) Conclude what actual boosts will be had with meta & rest training
3) Try and find the ideal training regime for realistic use
I thought so too, but tam1236 pointed out to me that this is not the case
..Although halfway through writing up this reply, I realize we are talking about two different things here - you have in mind absolute abundance, I intend to mean relative abundance. Tam1236 was saying that I shouldn't favor pro over det, because low pro is significantly rarer than low det.
Nonetheless I think just laying out the matter will bring everything together.
Opening up a save I have where I have 140k players loaded, 1.5 years in (I thought I was a few years in before, but its enough for 1 year of newgens at least), I find the following using Genie Scout:
Total players = 139,514
Total players newgen only = 14,033
Det 1-6 = 27,235 players
Pro 1-6 = 1,870 players
Det 1-6 newgen only = 4,185 players
Pro 1-6 newgen only = 524 players
Det 1-8 = 41,228 players
Pro 1-8 = 7,998 players
Det 1-8 newgen only = 5035 players
Pro 1-8 newgen only = 1045 players
Det 1-12 = 88,614 players
Pro 1-12 = 104,391 players
Det 1-12 newgen only = 9721 players
Pro 1-12 newgen only = 9328 players
So you are right that majority of players are going to be under 12 pro, however we don't need 100k, or even 1k of players, when we are filtering.
For an absolute filter I suggest 8 pro. This leaves 94% of players while getting rid of the bulk of the poor performance.
For a relative filter (Genie Scout attribute weightings) I suggest weighting pro at ~2/3rds of det, because while their performance and training impact is roughly similar overall (beyond ~7), your chance of having low det players in your search is much higher. I guess you could kind of think of this as the inverse corollary to the 'absolute' filter; it is designed to cull as many players as possible from the list while minimizing performance loss.
This time I believe I am not mistaken.
If you look at my 1 CA templates, you'll see I use at least 7 work rate. Not because 7 is required, but because mentals/technicals can decrease as training redistributes attributes. I am, as you are, taking the 6 minimum from HarvestGreen's data. In this case I am still using the extra 1 point buffer, as in my ideal test of the meta training work rate was +1-2 after 4 years, and as mentioned I'm assuming you will get ~2/3rds of ideal training results, so 6 filter denotes 7 after training.
The attributes in italics are in italics for the reason that they are difficult to find, but they are not impossible to find. I have taken them from my ~115 CA ideal templates, which in turn are taken from a search of the database in Genie Scout to find attribute values for which there are at least ~50 players available. If I filter for attributes for DC, including the drib 13, there are 9 results, 1 of which is age 28 and valued at $150k pounds. If you remove the drib 13, as I recommended, I find 245 results.
As shown in my 1 CA results, jump reach of 1 scored best for ST. Not because 20 isn't better of course, but because it isn't strictly necessary. The reason jump 15 is included for AML/AMR is because I believe it is strictly necessary if want to get the bulk of performance. Yes I tried 20 jump ST with 1 jump AML/AMR. Yes I tried 20 jump with 20 head, 20 jump + head + strength, and so on. All failures. The poster asked for the minimums for a filter - 12-13 pace/acc results in ~16-17 pace/acc after training, which is the bare minimum I found to be able to do very good in the Premier League.
Just have to ask on the topic of certain attributes, do you have like a summary of what you would consider the minimum viable value for certain attributes. For example, you have the example of 12 for professionalism being what you would consider a minimum value for 'good' player.
I guess what I am asking is to combine your ratings file with a filter for certain attributes as well, although file will do a great job by itself and this may be unnecessary. Apologies if I've missed it in another post.
GK - agil 11, pace 4, work 6, aer 15, ref 9, comp 4
DL/DR - acc 13, pace 13, jump 15, work 6
DC - acc 12, pace 12, jump 16, drib 13
DM - acc 13, pace 13, work 6
AML/AMR - acc 13, pace 13, jump 15, pos 13, work 6
ST - acc 13, pace 13, work 6
Minimum 0 CA attributes for all positions (these I'm more guesstimating than the above): 6 det, 6 nat, 8 pressure, 8 pro, 14< dirt, 14< injury, 6 loyal, 8 consist, 6 imp match.
Attributes in italics will be difficult to find and you should probably leave out of your filter, even though they are important.
The idea here is that with training, you will reach what I consider the minimums for what accounts for the bulk of max performance. I actually wanted to make 'pre-training' templates, but I found the effects of training to be too variable when comparing idealized isolated tests vs realistic premier league use, and I don't know yet what is actually realistic or typical. But with this filter here, I'm assuming you will get ~2/3rds of ideal training over 4 years, resulting in pace 12 > 16, which is the bare minimum I believe to get tangible victories (for premier league). It may fall short, but it will still serve it's function of separating most of the chaff from the wheat.
In regards to pro 12 being necessary, what I meant was that primarily pro 12< is rare, and we also know the following (source: Orion's FM22 testing):
So actually it's pro ~10 that is necessary, but I would guess that ~8 is not too bad either, based on how work rate matters a great deal up to just 6 then it tails off.
YildizAli said: Where do I paste this?
.grf file goes into (FM Genie Scout folder) > Ratings. Then in Genie Scout, press top left button > Rating and load the file.
https://files.catbox.moe/hrvdl8.grf
New file for FM26 users:
https://files.catbox.moe/r5xm3t.grf
For the FM26 version I've simply made some changes to a few of the attributes based on HarvestGreen's FM26 testing, so it may be less accurate.
I've also updated WB/MC/AMC so they're reasonably usable now, but they're just copies of FB & DM.
The main changes have been to the hidden attributes. Some very important changes here.
For those interested, here is also some rationale behind the weightings I've chosen:
Let's take long shots as first example, because I've given long shots very low weighting (1 or 2) in spite of both HarvestGreen22's findings (low moderate positive effect, similar to finishing or composure, which I give around ~8) and the fact that long shots have low CA cost in pretty much every position. Well in my 1 CA testing, I found that switching between 20 and 1 long shots made no observable difference in performance. And when I looked where goals were scored from in the in-game analytics, virtually none from long shots, and my DC with 20 long shots scored 1 or 2 at most over the season. And then there is also my own purely speculative theory that the knap tactic isn't designed for throwing the ball away on distant pot shots, and even if it is, it probably requires high levels of 'vision' and whatnot to pull off. Contrast all this info to pace, which it's clear as day even 1 point extra in makes a difference. So I figure my starting point is, I wouldn't trade 1 point of pace on a DC for 20 points of long shots - that makes it's weighting a maximum of '4' when pace is '100'. Now if I look at DL, long shots has a CA weight of just '1' and I gave it weighting of '1' in genie scout, whereas for DC I gave it '2' - and of course, neither of these are '4'. Consider that for DL, dribbling is also '1' CA weight, and it does actually matter. But I also have to weigh up what other attributes take precedence before long shots. Take strength - it's a fairly costly attribute ('4' weight), but it demonstrably has at least has some performance effect. Would I rather 5 strength on my DL, or 20 long shots? I think the 5 strength is going to perform a little better, but I wouldn't even trade 20 strength for 1 pace partly because of performance, partly because of the high CA cost (we can't have everything). So I give pace 100 weight, strength 3, and long shots 1.
Now if you read all that, you will realize that what I'm doing is quite subjective and prone to imprecision. I prefer precision, and this is largely possible with attribute testing, but unfortunately filtering methods don't work that way - the more precise you are, the less results you get, or conversely you start trawling through a few hundred laxly bounded results but only get as far as the 7th player you pick out and unwittingly choose a 17 long 16 pace 15 acc player over a 8 long 18 pace 18 acc player that was further down the list because it was too tedious to go through them all. Filter weightings alleviate this somewhat by creating a kind of hierarchy of attributes, so those 18 pace/acc players will be near top of your list no matter what. But if you assign too much high (or even moderate) weights to too many attributes, you muddy the hierarchy and end up with at the top of your list a complete dog's breakfast of a player that is 18 long 16 tack 17 str 12 pac 14 acc 20 ant and apparently a 78.41% rating.
And sometimes the attributes are almost purely subjective or circumstantial in ways you cannot pin down to a precise number. For instance, natural fitness technically doesn't affect performance much, so long as you manage it right - but managing a squad's match fitness I find requires you to have say ~30% of your players with high natural fitness so they can skip games to allow for a sizable squad without ill consequence. Loyalty.. I'm not aware it has any impact on performance at all, yet we all know the value of a loyal player - personally I figure I'd be willing to trade 1 each of pace & acc for it. For someone else it might be more or less, but I have to put some figure there.
One more I will make an example of is professionalism. I've reduced it to '16' in this update, which is relatively low now. Professionalism is actually very important for both performance and training, and it's 0 CA cost. But it turns out professionalism under 12 is quite rare, and the benefits of 12 -> 20 are much less than 1 -> 12. I do calculations based on all of this info to get a true relative figure. Although again a bit wishy washy, this type is actually quite solidly grounded in the performance stats.
Taking one from the middle and applying latest genie scout ratings I haven't uploaded yet (position is AMR):
mohamed salah 77.28%
messi 71.13%
jack grealish 66.83%
my player 66.30%
rank 102nd
Highest valued player (ST) was:
mbappe 77.59%
haaland 77.25%
raheem sterling 69.09%
my player 68.08%
rank 60th
Interesting.. I had assumed that after they got rid of national personality templates, they replaced it with a purely random assignment of them.
SI said at the time:
Rather than using a template based on the averages of existing players of a certain nationality, all personality attributes of newgen players are now generated randomly for every nationality
The new system Sports Interactive moved to abandoned the nation-based averages and replaced them with a system that gives newgens a random set of attributes based on their role and position
If I search for pro 5< there seems to be an equal representation of positions, but I do notice that 80% of them also have det 5<. I also notice that for both 5< pro and >18 pro, there are often repeating clubs (i.e. 3-5 players from the same club in the list).
One club I looked at (San Francisco City FC) that had 2 x 19 pro newgens, they had no staff, not even a manager. They have terrible facilities. I couldn't work out any possible correlates.
You are quite right in the points you make and there are indeed a lot more low 5< det newgens than 5< pro newgens. They seem to converge at ~12, and going by Orion's FM22 test of hidden attributes, pro actually matters less than det at this point (20 vs 10 pro = +5.6%, 20 vs 10 det = +12.7%, whereas 10 vs 1 pro = +46.3%, 10 vs 1 det = +10.2%). It holds true for existing players too, not just newgens. Then again, from memory pro does better for training than det.
Putting it all together, I guess det should be weighted in genie scout say 1.5x pro.
what i did find is that i need to choose carefully the opposing team in a friendly to not be such an easy match. i found that in matches against team that i demolish match sharpness doesn't improve.
thank you and keep up the good work!
P.S. i'm hoping to soon find a guide you create based on your findings to essentially assemble and manage a great team
I may be wrong, but I believe that opposing team quality/play doesn't affect match sharpness gain. I suspect what you are observing is the fact that friendlies only increase match sharpness by 60% of competitive matches (if I remember correctly). As match sharpness gain slows the higher it is, friendlies become less useful but still usable at match sharpness >90%. I've actually noticed that playing against minnows is best because it boosts morale, which is beneficial to future performance.
So I've tested it, and it really is important:
115 CA templates pressure 1 - 102, +127 | 105, +119
That's a 32% drop in performance. It is important to note though that Orion found the difference between 10 pressure and 20 pressure to only be 10%, but even that is significant.
This doesn't effect my templates, as they have 20 pressure, but I will have to update my Genie Scout ratings file.
The other one I'm particularly interested in is important matches. Orion said important matches makes ~6% difference. My results:
115 CA templates important matches 1 - 107, +168 | 108, +175
So ~6% decrease, which matches Orion's finding.
I took Luton (relegation candidate in premier league), removed their players, and added 15 players who cost 3mil pounds or less who were high in the Genie Scout ratings. Average value *after* transferring to Luton was 3.1mil pounds. Value of top 15 default players was ~10mil pounds average. Using the knap tactic and blue set piece routines, results were:
Luton (3mil< pound players) 7th, 70, +33
Luton default players 5th, 77, +33
Unlike the attribute template results, this isn't clear cut enough to be certain it's working as intended. Seems like it's doing alright though.
Samet2772 said: **Hello, if I use this file in FM 26 Genie Scout, will it work?**
I assume the ratings files are compatible between versions
My data is for FM24, but I don't think much changed for FM26. I heard long shots got a boost in particular, so maybe make long shots '3' instead of '1' say.
is there any way around this?
I've managed it manually myself in testing successfully. It can be a bit tricky, but it's definitely doable, although I played with Man City which have a lot of fixtures. I had to schedule some friendlies during the times when there were 2 week gaps, and I suggest you do the same to maintain match sharpness (you could do it as reserve matches as well).
I would caution against simply adding more training sessions in, as I found it worsens training results in most cases, I would guess because you're replacing a rest session (high distribution to pace/acc) with a training session (distributes more to technicals/mentals) rather than simply 'adding' the training session on top.
119.6 CA average (new templates) no limits - 114, +223 | 114, +204
Old templates 139.2 CA average - 110, +221 | 112, +172 | 114, +177 | 114, +167 | 114, +206
To make it fair I made all 0 CA attributes for the old templates the same, except flair and aggression. The results were a bit all over the place, not sure why.
+182.2 for new templates, +188.6 for old templates, but +17% CA cost for old templates, so overall the new templates are about ~14% better.
My 1 CA testing was run with the same conditions, so you can also add this to compare:
1 CA templates (best single result) - 91, +78
I have to go to best DC Rating, am i doing right?
Correct
A few extra things I left out of my post:
Genie Scout ratings will be ~70% max. This is necessary, don't try and change this, I put the weight as high as I can but Genie Scout limits it to 120 so I can't make it nicer.
The height values I give on the positions mean something. They are the minimum height you need to have a chance of getting 20 jumping reach off the bat, but more importantly what you need if you intend to train to 20, as the meta training only increases jump reach by ~2 over 4 years. Kudos to ClaudeJ for pointing out this interesting find that comes from an Italian official FM researcher guide page. They tell the low-level researchers to assign jump reach according to this table:
And Aerial Reach for GK:
I put pace/acc as 20 in my templates, but I'll re-emphasize that the rest of the attributes represent high/low attribute values in existing players. I think it's kinda useful, because while we know now that 1 decisions and 20 dribbling is ideal, what we really need to know is whether or not a player with 6 decisions or 14 dribbling is the best we're going to be able to realistically find. With these templates, you can look at a DC with 13 dribbling and realize, this is actually a valuable find.
I would change the Knap tactic slightly in those ways, i.e. just put 'work ball into box' on, to see if it would boost my results. None of my changes did, the tactic was best as-is.
max 737 said: How do you get to this screen? I am getting this option of IP and OOP roles?
I wouldn't know, I'm not using FM26
115 CA average:
200 CA:
Genie Scout ratings file:
https://files.catbox.moe/f2052w.grf
GK, FB, DC, DM, Winger, Fast ST are valid.
Sweeper is GK with captain emphasis. Target striker is ST with penalties & free kicks emphasis. WB, MC and AMC I've left unchanged from my first genie scout ratings version, so they're outdated, but they're not bad if you want to search those positions.
Two quotes spring to mind:
You go to war with the army you have, not the army you might want or wish to have at a later time.
As we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns — the ones we don’t know we don’t know.
There are several known unknowns in my templates. For instance I know from my 1 CA testing that balance of 17 on my ST is probably unnecessary for performance. But I don't have the time and energy to test and verify every attribute, so I employed the following method:
1. Use and verify the key attributes identified by my 1 CA testing
2. Test and verify a bunch of remaining high CA weight attributes
3. Leave remaining attributes as-is that got best results just before my 1 CA testing
So I know for certain that these ideal templates could be better, but they are nonetheless say ~90% peak performance, where balance 17 may be too excessive and flair 12 may be too low.. or high.. and corners 12 may be just plain stupid - but these attributes will account for just a small handful of CA. Perhaps sometime later I will fix these lesser attributes up.
I also realized that my 1 CA templates distill the key attributes better. So I thought I would instead use these templates to illustrate not just the key attributes, but also what's actually realistic. So with my ~115 CA templates, attribute values are bounded by what players are actually available - there had to be at least ~50 players available with each particular value in the specified position.
That doesn't make my templates worse performing, I haven't compared them to the old ones specifically yet, but they should be significantly better.
I've isolated and optimized things better, so it's not comparable to the old results.
Here's the performance data (English Premier League, as Man City):
Outfield 20 acc/pace/jump/drib 12 other visible (133 CA outfield average) - 111, +138
115 CA template (118 CA outfield average) - 112, +198 | 109, +178 | 114, +169 | 114, +183 | 110, +183
Outfield 20 acc/pace/jump/drib 14 other visible (169 CA outfield average) - 114, +251
200 CA template - 114, +409
For those interested, here are the results of some attribute changes:
DC pos mark agil tack 1 - 114, +167
DL/DR DC DM mark 1 - 112, +175 | 114, +175
GK hand acc bra con communic command pos 1 - 112, +182 | 112, +195
outfield bal 1 - 110, +163 | 112, +173
outfield ant 20 - 107, +169 | 110, +170
no preferred moves - 114, +170 | 107, +170
outfield ant 1 - 110, +127 | 110, +135
GK str one pass bal 1 - 109, +160 | 111, +173
outfield 1 agil - 109, +144
flair 20 - 112, +173
teamwork 20 - 114, +158
first str 20 - 114, +242 | 114, +223
first 20 - 114, +198 | 110, +187
AML/AMR ST first 20 - 111, +176
DL/DR DC DM first 20 - 110, +182
str 20 - 111, +216
DL/DR DC DM str 20 - 110, +190
AML/AMR ST str 20 - 114, +207
DL/DR DC 7 pos (reduced from high pos) - 110, +198 | 114, +177 | 112, +187
I've highlighted the most surprising results.
Note: These should really be compared against [112, +187] as I made some adjustments (i.e. GK; attribute decreases) after these results.
So preferred moves don't seem to matter; not much at least. Anticipation matters quite significantly. Agility matters moderately, but would be outweighed by its expensive CA cost. High flair doesn't seem to be bad, but perhaps not beneficial either. GK acceleration, and a bunch of other GK specific attributes, seem to hardly matter at all. Strength matters moderately, and only on forwards, but probably outweighed by high CA cost. First touch has minor to moderate effect, but it costs too much like strength. In the end I decided not to boost first touch + strength, the CA cost isn't worth it, especially when you consider that CA-PA difference is crucial to training growth (if your ideal player is 140 CA, that means you actually need say ~170+ PA if you want him to realistically grow into it). As you can see I tested balance even though its only weight is only 2, and lowering it resulted in worse performance, so I decided to just leave it high, even though the real optimum I would guesstimate is ~12.
These are from me testing attributes I wasn't sure on. A lot of attributes I already have a solid idea about from my 1 CA testing.
In regards to the Genie Scout Ratings, it's quite radically changed from my previous version. The way I decided to approach it is this: An attribute is worth as much pace/acc I would intuitively be willing to swap for it. I've been working with the data and reckon I have a good feel for it, so I trust my intuition here. So for instance, if I think about dribbling for AML/AMR, I know dribbling is important and also hard to train, but I also know that under 17 pace/acc is a no-go no matter what the dribbling is. So really what I'm saying is, 20 drib = 6 pace/acc. So drib = 30% (6/20). Then I might apply a very minor adjustment, in this case to 29%, for other factors I feel are relevant.
So in my refreshed genie scout ratings, most attributes are reduced a lot compared to pace/acc. The one that makes me a bit uncomfortable is jumping reach. It can be jumping reach is around about as valuable as pace/acc perhaps, but my thinking is that pace/acc does very good without jump, whereas jump without pace/acc sucks. Also jump only matters a lot if its very high, I would guess. I think the limits are 15 pace/acc + 20 jump for DC (20 jump = 10 pace/acc), 18 pace/acc + 20 jump for AML/AMR (20 jump = 4 pace/acc), and so I weighted it accordingly. There are a few other attributes I want to reassess, such as 'pressure' and 'important matches', but this is where I'm at right now.
Same training should be applied to all teams and players
However I see now the reason why HarvestGreen suggested full rest for young players before switching to a more balanced regime once they reach near 20 pace/acc. They need the pace/acc first to play good, but also the balanced regimes can be simply too slow to build pace/acc.
Even in a perfect test environment with 5 star coaching, pace/acc with the more balanced training is +6 after 4 years. In a more realistic test I did, it was only +1 acc and +2 pace after 3 years.
Part of what is going on here is that growth is highly influenced by CA-PA difference, so you really don't want to pump all that technical/mental CA into your player until the low CA pace/acc growth is largely done.
I haven't tried it myself yet, but I imagine the best way to go about it would be this:
First team - Quick + Match Prac + 2xAttack; full rest for congested weeks
U21s - Quick + Match Prac + 2xAttack
U18s - Full rest
Age 15 -> 17 (U18s) = +5 pace/acc
Age 18 -> 20 (U21s) = +2 pace/acc
The thing is, certain key technicals such as dribbling are very hard to gain, so I think you're shooting yourself in the foot if you choose to start by losing drib points with full rest. But pace/acc is more important than drib. So it's debatable.