Em said: This is the best tactic in FM right now, right? So it can be used with any team, right?" Expand
Any tactic from 84+ is equally good, because they fall under the margin of error from FM-Arena. This one is just the current top, and it will perform well with any team that has a squad to support it (two or three players proficient in each position).
GeorgeFloydOverdosed said: I was going to add, but decided in the end not to, that you seem to to believe that GK only contributes to goals prevented rather than goals scored. The reason I decided not to is because I had a brief look at the past results to find evidence to disprove this, but realized it would just take too much time to collate all the necessary information to present it neatly. But my impression just from having running the tests and seeing the results, is that GK doesn't just prevent goals, better GK results in more team goals scored. This is more clearly demonstrated/known with better defenders resulting in more goals/wins.
So you're saying I should make your confirmation bias a prerequisite of my testing?
My testing can be taken seriously because I'm running them in the real league on full detail with players that have realistically attainable attributes, with cumulatively thousands of samples not on every attribute precisely but the basic picture. So I can say with certainty that a team of 1 CA players can win the Premier League, and that a team of 15 pace/acc players can also win the Premier League (4th/5th most typically though), but when I found in my 1 CA testing that concentration only mattered for DC/DL/DR that turned out to be wrong - or perhaps it only applied to that 1 CA limitation. Stamina also did not matter for those 1 CA players, and according to HarvestGreen stamina is one of the least important physical attributes - less important than agility he finds.
It is telling that you think changing your mind is a sign of weakness. What the data is showing is actually that there perhaps aren't any outfield positional differences. I think what is going on with stamina is that it affects the decay rate of pace/acc over a match. So in my 1 CA tests where players have 17-20 pace/acc, low stamina is insignificant because pace/acc is from a high starting point to begin with. This would also therefore be true once you have your 18+ pace/acc players in game. For this 15 pace/acc where we are hitting the lower threshold of the pace/acc requirement, stamina is necessary to make up for its shortfall towards the end of the match. I have done another test earlier where 14 pace/acc/sta comes 8th, so it does appear to work this way. What this means is that in any case, stamina is less important than pace/acc in any position including fullbacks, and one would guess that 20 stamina = 4 pace or acc, so 20% of pace/acc. This happens to align neatly with HarvestGreen's finding of 21.7%.
As far as I'm aware, no one else is doing the kind of testing I'm doing, which is using the real league in situ. Everybody else seems to be doing carefully controlled custom leagues where every player has 10 attributes or whatnot, and even making morale and such identical. I've been influenced to take my approach from previous experience, where I couldn't work out the newgen factors based on isolation tests, so I ended up simply testing the median PA for every single playable club in Football Manager. And from those results it turned out that there is a hidden factor I couldn't have found in my isolation tests and why I could never get my formula to work properly. The great thing about this method is that there's no question of 'yes, but will this work in the Premier League?' - it simply does.
My observations are:
1) Certain attributes seem to have more result variability than others. For instance, important matches 13 > 8 results in say 4th, 12th, 4th, 12th instead of the typical 4th, 5th, 5th, 8th, 5th, 3rd, 4th, 10th.
2) Say my team comes 2nd almost every test. Liverpool gets 68 points, I get 66 points. I get 88 points, Liverpool gets 91 points. Liverpool gets 76 points, I get 71 points. I'm not sure how this gels with your idea of how it works, but if it was just 2 sets of randomness, why would two team's points track each other like this? In any case, I don't know on this one, all I want to assert is that table position is an adequately reliable measure. Even if you're correct in that it's just as random, but the scale is smaller so it doesn't show, then for all intents & purposes it still works - the RNG does not overwhelm the predictability of table position. Not that I mean to be dismissive of your theory, I actually think it's somewhat important to get to why it works like this and I'm open to ideas on it.
2) The typical position variability for me is less than you have observed. Maybe it's the tactic, maybe it's the unevenness of players, and so on. Now unfortunately I don't save my dud results, so I can't really present data on this. But based on my saved successes, and recollections of duds, I reckon it will be in the range ~3rd-8th ~80-90% of the time. But I've done a test of this below to determine the truth.
So I did 10 one-season tests of 14 pace/acc/stamina + new GK template since I want more samples on both of those anyway:
8th 4th 3rd 9th (sacked after 14 games) 8th 2nd 13th (sacked after 15 games) 6th (sacked after 36 games) 10th (sacked after 20 games) 6th
So 7/10 times (I'm going to count 6th sacked after 36 games) the team did 8th or better. The mode was 6th or 8th. The median was 7th. The average was 7th (6.9).
And that's assuming those early sackings wouldn't have led to higher results later if not sacked. If I were summing up these results, I'd say 14 pace/acc/sta does ~6th-8th. Expand
There are many ways to avoid the variance in tests. You could perform statistical tests, use stuff like medians instead of averages, or try simple solutions like "Do ten tests, discard the three best and three worst results, and average the middle four". Those techniques are made to avoid outliers.
keithb said: Obviously he's a physical monster. But what else do you think makes him so superior? In FM 23/24 I had him score 100 a season. Never had anything like it before or after him in a save. Expand
I have no clue what he has better that is much better than other elite games, unless "gets the crowd going" makes him score more goals.
keithb said: It's a single player game and anyone can play it in any way they like. Sometimes I wanted to see how badly I could "cheat the game" and others I was happier to play slowly with my own tactics and only buying players recommended from my scouts. I also used python which I thought was far superior to genie scout. Setting a level 7 for speed.
If I feel someone is misleading others then I will say something. Or like a few years ago when someone was clearly bullying others on the forum I spoke up. Lockjaw probably loves free speech and he loves people sticking it to people and organisations. So I'm just joining in.
Anyone not finding Haaland isn't by far the best player in the game and declaring "Tactic is actually somewhat more important than attributes." Needs to really ask themselves how they got it so wrong. It shows unreliable testing methods and a poor understanding of methodology, or just testing in general. If that person also has a fragile personality they will just crack on regardless and take no notice, as they couldn't possibly be wrong. As we endlessly see.
Take it easy Expand
I did a test with Haaland edited to play as AMC in Burnley, using Autumn. He was scoring over two goals per match.
First of all, thank you for the detailed training guide. I have one question regarding your recommended schedules.
In the “Most Effective Schedules” list, the schedule [Quickness] [Match Practice] [Attacking]x2 appears to have the highest CA gain, around 37.93 CA. However, in your suggested Growth schedules, you mainly use [Physical]x5 [Attacking]x5 for one-match weeks and [Physical]x2 [Attacking]x2 for two-match weeks.
Could you please explain why you recommend the Physical + Attacking Growth schedules instead of using the highest-performing schedule with Quickness + Match Practice + Attacking x2?
Is it because of match sharpness, fatigue, injury risk, long-term attribute distribution, or because Quickness is already covered through individual training?
Also, in my save I usually use two completely different starting elevens when I have two matches in the same week. For example, one eleven plays the league match and another eleven plays the European match. I also set very tired players to no pitch or gym work.
In that case, what would you recommend as the most effective weekly training schedule for maximum player development over a full season?
Should I still use your suggested [Physical]x2 [Attacking]x2 schedule for two-match weeks, or would it make more sense to use the higher-performing [Quickness] [Match Practice] [Attacking]x2 schedule because my players are rotated heavily?
Thank you! Expand
The goal of my training suggestion is not to "reach PA", but to "reach PA with maximum performance". In short, the more you gain CA, the slower it increases, and when you reach maximum, you cannot gain Pace and Acceleration anymore. Therefore, my training is aimed at reaching Pace and Acceleration 18+ with as much CA as possible, instead of reaching maximum CA with Pace and Acceleration 14 or 15. That means you want to increase only the attributes that really matter, so you do not waste CA with attributes that barely increase performance.
If you want to speed up development, then my recommendation is to skip [Physical]x5 [Attacking]x5 and use only [Physical]x2 [Attacking]x2 (because for some reason that one increases more CA. Keep in mind that is just my personal opinion, and there might be other people much more qualified to talk about this subject.
If I run the same tactic five times with Burnley, sometimes it will end 3rd place, sometimes 12th, and sometimes it will have awful results that make me even check if I loaded the right tactic before the test. What I mean is that a single league test is definitely not enough to prove anything, and even two or three would still not be reliable enough to draw any conclusions.
Other than that, the fact that table position have less variance than points in the table do not mean much. The table goes from 1st to 20th, while points go from 0 to 114. Realistically, the same experiment ran multiple times could variate position between 6 different positions, while points in the table would naturally variate 30 or so. They are simply different ranges, but the position in table depends much more from performance of other teams against each other, while points in the table depends mostly of your performance against each of them.
drake87 said: I think central midfielders are shooting too much just wasting good opportunities, can you test them with the shot less often instruction? Im pretty sure it will be a improve. Expand
Thank you for the suggestion. I will test tomorrow when my submissions reset.
Jessexripley said: Will this also get a light version now it it your best result scoring tactic ? Expand
It already has in the description, but I will update and test again the defensive and light versions, as well as the one with striker. Those might get better or worse, so testing is needed.
Yeah, it got the highest points per match in the tests.
Any tactic from 84+ is equally good, because they fall under the margin of error from FM-Arena. This one is just the current top, and it will perform well with any team that has a squad to support it (two or three players proficient in each position).
So you're saying I should make your confirmation bias a prerequisite of my testing?
My testing can be taken seriously because I'm running them in the real league on full detail with players that have realistically attainable attributes, with cumulatively thousands of samples not on every attribute precisely but the basic picture. So I can say with certainty that a team of 1 CA players can win the Premier League, and that a team of 15 pace/acc players can also win the Premier League (4th/5th most typically though), but when I found in my 1 CA testing that concentration only mattered for DC/DL/DR that turned out to be wrong - or perhaps it only applied to that 1 CA limitation. Stamina also did not matter for those 1 CA players, and according to HarvestGreen stamina is one of the least important physical attributes - less important than agility he finds.
It is telling that you think changing your mind is a sign of weakness. What the data is showing is actually that there perhaps aren't any outfield positional differences. I think what is going on with stamina is that it affects the decay rate of pace/acc over a match. So in my 1 CA tests where players have 17-20 pace/acc, low stamina is insignificant because pace/acc is from a high starting point to begin with. This would also therefore be true once you have your 18+ pace/acc players in game. For this 15 pace/acc where we are hitting the lower threshold of the pace/acc requirement, stamina is necessary to make up for its shortfall towards the end of the match. I have done another test earlier where 14 pace/acc/sta comes 8th, so it does appear to work this way. What this means is that in any case, stamina is less important than pace/acc in any position including fullbacks, and one would guess that 20 stamina = 4 pace or acc, so 20% of pace/acc. This happens to align neatly with HarvestGreen's finding of 21.7%.
As far as I'm aware, no one else is doing the kind of testing I'm doing, which is using the real league in situ. Everybody else seems to be doing carefully controlled custom leagues where every player has 10 attributes or whatnot, and even making morale and such identical. I've been influenced to take my approach from previous experience, where I couldn't work out the newgen factors based on isolation tests, so I ended up simply testing the median PA for every single playable club in Football Manager. And from those results it turned out that there is a hidden factor I couldn't have found in my isolation tests and why I could never get my formula to work properly. The great thing about this method is that there's no question of 'yes, but will this work in the Premier League?' - it simply does.
My observations are:
1) Certain attributes seem to have more result variability than others. For instance, important matches 13 > 8 results in say 4th, 12th, 4th, 12th instead of the typical 4th, 5th, 5th, 8th, 5th, 3rd, 4th, 10th.
2) Say my team comes 2nd almost every test. Liverpool gets 68 points, I get 66 points. I get 88 points, Liverpool gets 91 points. Liverpool gets 76 points, I get 71 points. I'm not sure how this gels with your idea of how it works, but if it was just 2 sets of randomness, why would two team's points track each other like this? In any case, I don't know on this one, all I want to assert is that table position is an adequately reliable measure. Even if you're correct in that it's just as random, but the scale is smaller so it doesn't show, then for all intents & purposes it still works - the RNG does not overwhelm the predictability of table position. Not that I mean to be dismissive of your theory, I actually think it's somewhat important to get to why it works like this and I'm open to ideas on it.
2) The typical position variability for me is less than you have observed. Maybe it's the tactic, maybe it's the unevenness of players, and so on. Now unfortunately I don't save my dud results, so I can't really present data on this. But based on my saved successes, and recollections of duds, I reckon it will be in the range ~3rd-8th ~80-90% of the time. But I've done a test of this below to determine the truth.
So I did 10 one-season tests of 14 pace/acc/stamina + new GK template since I want more samples on both of those anyway:
8th
4th
3rd
9th (sacked after 14 games)
8th
2nd
13th (sacked after 15 games)
6th (sacked after 36 games)
10th (sacked after 20 games)
6th
So 7/10 times (I'm going to count 6th sacked after 36 games) the team did 8th or better.
The mode was 6th or 8th.
The median was 7th.
The average was 7th (6.9).
And that's assuming those early sackings wouldn't have led to higher results later if not sacked. If I were summing up these results, I'd say 14 pace/acc/sta does ~6th-8th.
There are many ways to avoid the variance in tests. You could perform statistical tests, use stuff like medians instead of averages, or try simple solutions like "Do ten tests, discard the three best and three worst results, and average the middle four". Those techniques are made to avoid outliers.
I have no clue what he has better that is much better than other elite games, unless "gets the crowd going" makes him score more goals.
SK-D, 3CD(2BD and and a Libero-S on center) 2FB-A, 2CM-S, 2W-S on both central wings, and a AF
or, 3BDP-D(Libero Support on center), 2FB-S or 2WB-S on both WB, 2VOL-A, 2W-S and an AF
Regroup and hold shape maybe help.
That would be a completely different tactic.
3CD or 2CD one HB maybe
SK-D, 2BDP-D, 2FB-A or 2WB-S, a HB on DM, 2MEZ-S or 2CM-S(with stay wider or move into channel p.i), 2W-S or 2IF-S and an AF or a PF-S
Balanced or Positive mentality
Mid block and Much Lower or Standard defensive line
It is tricky because some positions are hard to replicate, and there was no Out of Possession system.
If I feel someone is misleading others then I will say something. Or like a few years ago when someone was clearly bullying others on the forum I spoke up. Lockjaw probably loves free speech and he loves people sticking it to people and organisations. So I'm just joining in.
Anyone not finding Haaland isn't by far the best player in the game and declaring "Tactic is actually somewhat more important than attributes." Needs to really ask themselves how they got it so wrong. It shows unreliable testing methods and a poor understanding of methodology, or just testing in general. If that person also has a fragile personality they will just crack on regardless and take no notice, as they couldn't possibly be wrong. As we endlessly see.
Take it easy
I did a test with Haaland edited to play as AMC in Burnley, using Autumn. He was scoring over two goals per match.
How can I replicate it for FM24? Or do you have any suggest a strikerless tactic for FM24?
Strikerless was pretty weak in the previous version. You can see all strikerless tactics tested in FM-Arena here:
https://fm-arena.com/table/30-patch-24-4-0-db3-0/?pos_data=%7B_22_-_3_._23_-_3_._24_-_3_%7D&mirror=2&asym=2#options
Nikos said: Hi ZaZ,
First of all, thank you for the detailed training guide. I have one question regarding your recommended schedules.
In the “Most Effective Schedules” list, the schedule [Quickness] [Match Practice] [Attacking]x2 appears to have the highest CA gain, around 37.93 CA. However, in your suggested Growth schedules, you mainly use [Physical]x5 [Attacking]x5 for one-match weeks and [Physical]x2 [Attacking]x2 for two-match weeks.
Could you please explain why you recommend the Physical + Attacking Growth schedules instead of using the highest-performing schedule with Quickness + Match Practice + Attacking x2?
Is it because of match sharpness, fatigue, injury risk, long-term attribute distribution, or because Quickness is already covered through individual training?
Also, in my save I usually use two completely different starting elevens when I have two matches in the same week. For example, one eleven plays the league match and another eleven plays the European match. I also set very tired players to no pitch or gym work.
In that case, what would you recommend as the most effective weekly training schedule for maximum player development over a full season?
Should I still use your suggested [Physical]x2 [Attacking]x2 schedule for two-match weeks, or would it make more sense to use the higher-performing [Quickness] [Match Practice] [Attacking]x2 schedule because my players are rotated heavily?
Thank you!
The goal of my training suggestion is not to "reach PA", but to "reach PA with maximum performance". In short, the more you gain CA, the slower it increases, and when you reach maximum, you cannot gain Pace and Acceleration anymore. Therefore, my training is aimed at reaching Pace and Acceleration 18+ with as much CA as possible, instead of reaching maximum CA with Pace and Acceleration 14 or 15. That means you want to increase only the attributes that really matter, so you do not waste CA with attributes that barely increase performance.
If you want to speed up development, then my recommendation is to skip [Physical]x5 [Attacking]x5 and use only [Physical]x2 [Attacking]x2 (because for some reason that one increases more CA. Keep in mind that is just my personal opinion, and there might be other people much more qualified to talk about this subject.
Other than that, the fact that table position have less variance than points in the table do not mean much. The table goes from 1st to 20th, while points go from 0 to 114. Realistically, the same experiment ran multiple times could variate position between 6 different positions, while points in the table would naturally variate 30 or so. They are simply different ranges, but the position in table depends much more from performance of other teams against each other, while points in the table depends mostly of your performance against each of them.
Not necessarily.
Thank you for the suggestion. I will test tomorrow when my submissions reset.
Yeah, you can simply load 3.46 and change that instruction in the two CHMs.
Exactly.
My what? If you mean player instructions, they are in the first post.
It already has in the description, but I will update and test again the defensive and light versions, as well as the one with striker. Those might get better or worse, so testing is needed.