No doubts, "Overlaps" and "Underlaps" are among the most "mysterious" Team Instructions in FM.
Unfortunately, it's quite hard to measure the impact of these TIs through out our regular tactic testing because even if you take 4,000 matches tests then the RNG still can be as high as 2.5 points which only allows "catching" changes that make at least 2.5 points difference.
So we decided to shed light on "Overlaps" and "Underlaps" TIs and test them for 32,000 matches, such amount of matches gives us a very small RNG which is no more than 0.5 points.
P.S. I predict, some people will rush to remove "Overlaps" TIs from the top tactics in hope to see their scores improves but I can assure your that highly likely it won't happen because almost all tactics at the top already hit a quite high positive RNG which is about +2 points so imagine you remove the Overlaps TI from the top tactics which would give +1 points but you struck -2 points negative RNG after 4,000 matches and as result you end up with a worse score. Expand
Your general point is correct that making this change is unlikely to lead to a higher score in the test league, but the logic is wrong since what you refer to is actually ranking bias which is not the same as the random error in estimating the score. The size of the ranking bias is not possible to calculate given the nature of the submissions to the ranking table, but it would only be +2 by an unlikely coincidence.
I think more relevant is that for use in the game, removing overlaps would be expected to be favorable (at least for the role combinations used). Anyway thanks for the useful experiment.
Gianaa9 said: Could you provide IF and FB PIs? I think that using Hold Up on IFs could improve Overlaps’ rating respect to others one, did you used it? Thank You! Expand
Hindering your tactic just to make a specific thing function better is not the way to success.
And the main issue I'd guess from data is that the instructions put players more out of position, thus leaving the team more vulnerable in defense.
Gianaa9 said: I think that using Hold Up on IFs could improve Overlaps’ rating respect to others one, did you used it? Thank You! Expand
Hi,
Yes, I agree that some PIs such as "Hold Up Ball" or even "Attacking" Mentality might make "Overlaps" TIs shine in such tactics as Katana or COSMOS.
Probably, by tomorrow evening or earlier I'll try to run Katana tactic without "Overlaps" for 32,000 matches and we'll find out whether the Mentality or other ITs and PIs make a difference for the performance of "Overlaps" TIs.
Here's data for Katana 4231 104p v3.1 tactic with "Overlaps" TIs and without "Overlaps" TIs after 32,000 matches tested.
Judge for yourself.
There's no obvious negative effect from "Overlaps" TIs as it happened for the two previous "generic" tactics.
It seems the "Attacking" mentality or other TIs and PIs might influence on how "Overlaps" TIs perform and in "Katana 4231 104p v3.1" tactic the "Overlaps" TI even give a small positive effect which is about "+0.1" points but don't forget that even 32,0000 matches test also have at least 0.2 RNG.
@alex this tactic - https://fm-arena.com/thread/9578-katana-4231-104p-v3-1-27/ might scored that low due to the fact that the "Overlaps" TIs weren't the only instructions that were changed, I noticed that "Moves Into Channels" PI was also removed from AM, so there's no point in testing it for 32,000 matches because anyway, it isn't possible to "isolate" the absence of "Overlaps" TIs in that case.
Zippo said: Here's data for Katana 4231 104p v3.1 tactic with "Overlaps" TIs and without "Overlaps" TIs after 32,000 matches tested.
Judge for yourself.
There's no obvious negative effect from "Overlaps" TIs as it happened for the two previous "generic" tactics.
It seems the "Attacking" mentality or other TIs and PIs might influence on how "Overlaps" TIs perform and in Katana 4231 104p v3.1 the "Overlaps" TI even give a small positive effect which is about "0.1" points but don't forget even 32,0000 matches test also have at least 0.2 RNG.
@alex this tactic - https://fm-arena.com/thread/9578-katana-4231-104p-v3-1-27/ might scored that low due to the fact that the "Overlaps" TIs weren't the only instructions that were changed, I noticed that "Moves Into Channels" PI was also removed from AM, so there's no point in testing it for 32,000 matches because anyway, it isn't possible to isolate the absence of "Overlaps" TIs in that case. Expand
Zippo said: @alex this tactic - https://fm-arena.com/thread/9578-katana-4231-104p-v3-1-27/ might scored that low due to the fact that the "Overlaps" TIs weren't the only instructions that were changed, I noticed that "Moves Into Channels" PI was also removed from AM, so there's no point in testing it for 32,000 matches because anyway, it isn't possible to isolate the absence of "Overlaps" TIs in that case. Expand
I did initially forget to add that I have removed the move into channels on the AM in the opening post, I have since updated the OP.
I think all I can say is that this result is accurate at the two fixed combinations.none, overlaps, and underlaps are different for position duty selection and optimal selection for PI. I tested a variety of different 4231 Settings early in the release, both single-player and online, and couldn't come to a conclusion about which was stronger. Moreover, only with an ST formation can the attacking mentality compete with the balanced mentality, but the attacking mentality is stronger than the attacking mentality in the online game. 4231 is indeed a top formation, but it can occupy the top or rely on more test samples and luck, because 424 and 4231 are the first two tactics to occupy the top for a long time, so we fine-tune the test too much. There are actually some formations that don't lose to either.
It doesn't mean they don't do anything, only that the total effect of various options is quite close. Which is good. People complain non stop how gegenpress is the only way to play the game and that you cannot do other styles. And now when you have 3 more or less equal options it's again not right...
Zippo said: Here's data for Katana 4231 104p v3.1 tactic with "Overlaps" TIs and without "Overlaps" TIs after 32,000 matches tested.
Judge for yourself.
There's no obvious negative effect from "Overlaps" TIs as it happened for the two previous "generic" tactics.
It seems the "Attacking" mentality or other TIs and PIs might influence on how "Overlaps" TIs perform and in "Katana 4231 104p v3.1" tactic the "Overlaps" TI even give a small positive effect which is about "+0.1" points but don't forget that even 32,0000 matches test also have at least 0.2 RNG.
@alex this tactic - https://fm-arena.com/thread/9578-katana-4231-104p-v3-1-27/ might scored that low due to the fact that the "Overlaps" TIs weren't the only instructions that were changed, I noticed that "Moves Into Channels" PI was also removed from AM, so there's no point in testing it for 32,000 matches because anyway, it isn't possible to "isolate" the absence of "Overlaps" TIs in that case. Expand
Amazing test. Do you think of doing the same for focus play down the left/right?
Hey,
No doubts, "Overlaps" and "Underlaps" are among the most "mysterious" Team Instructions in FM.
Unfortunately, it's quite hard to measure the impact of these TIs through out our regular tactic testing because even if you take 4,000 matches tests then the RNG still can be as high as 2.5 points which only allows "catching" changes that make at least 2.5 points difference.
So we decided to shed light on "Overlaps" and "Underlaps" TIs and test them for 32,000 matches, such amount of matches gives us a very small RNG which is no more than 0.5 points.
Overlaps & Underlaps TIs
Katana 4231 104p v3.1
GENERIC TACTIC:
Inside Forwards(Support) + Full Backs(Attack)
PIs: "Dribble More", "Tackle Harder"
GENERIC TACTIC:
Inside Forwards(Support) + Full Backs(Support)
PIs: "Dribble More", "Tackle Harder"
@Zippo did you use any PIs in any position?
dzek said: @Zippo did you use any PIs in any position?
Hi,
Yes, "Dribble More" and "Tackle Harder" PIs when it was possible to apply them.
Zippo said: ,
P.S. I predict, some people will rush to remove "Overlaps" TIs from the top tactics in hope to see their scores improves but I can assure your that highly likely it won't happen because almost all tactics at the top already hit a quite high positive RNG which is about +2 points so imagine you remove the Overlaps TI from the top tactics which would give +1 points but you struck -2 points negative RNG after 4,000 matches and as result you end up with a worse score.
Your general point is correct that making this change is unlikely to lead to a higher score in the test league, but the logic is wrong since what you refer to is actually ranking bias which is not the same as the random error in estimating the score. The size of the ranking bias is not possible to calculate given the nature of the submissions to the ranking table, but it would only be +2 by an unlikely coincidence.
I think more relevant is that for use in the game, removing overlaps would be expected to be favorable (at least for the role combinations used). Anyway thanks for the useful experiment.
I've updated the 1st post and added data for Inside Forwards(Support) + Full Backs(Attack) combination.
Could you provide IF and FB PIs? I think that using Hold Up on IFs could improve Overlaps’ rating respect to others one, did you used it?
Thank You!
Gianaa9 said: Could you provide IF and FB PIs?
He already said that the positions have only 2 PIs: Dribble More and Tackle Harder.
Gianaa9 said: Could you provide IF and FB PIs? I think that using Hold Up on IFs could improve Overlaps’ rating respect to others one, did you used it?
Thank You!
Hindering your tactic just to make a specific thing function better is not the way to success.
And the main issue I'd guess from data is that the instructions put players more out of position, thus leaving the team more vulnerable in defense.
Lapidus said: He already said that the positions have only 2 PIs: Dribble More and Tackle Harder.
oh sorry i just thought they were regarding only FBs, don’t know why, thank you!
Gianaa9 said: I think that using Hold Up on IFs could improve Overlaps’ rating respect to others one, did you used it?
Thank You!
Hi,
Yes, I agree that some PIs such as "Hold Up Ball" or even "Attacking" Mentality might make "Overlaps" TIs shine in such tactics as Katana or COSMOS.
Probably, by tomorrow evening or earlier I'll try to run Katana tactic without "Overlaps" for 32,000 matches and we'll find out whether the Mentality or other ITs and PIs make a difference for the performance of "Overlaps" TIs.
Here's data for Katana 4231 104p v3.1 tactic with "Overlaps" TIs and without "Overlaps" TIs after 32,000 matches tested.
Judge for yourself.
There's no obvious negative effect from "Overlaps" TIs as it happened for the two previous "generic" tactics.
It seems the "Attacking" mentality or other TIs and PIs might influence on how "Overlaps" TIs perform and in "Katana 4231 104p v3.1" tactic the "Overlaps" TI even give a small positive effect which is about "+0.1" points but don't forget that even 32,0000 matches test also have at least 0.2 RNG.
@alex this tactic - https://fm-arena.com/thread/9578-katana-4231-104p-v3-1-27/ might scored that low due to the fact that the "Overlaps" TIs weren't the only instructions that were changed, I noticed that "Moves Into Channels" PI was also removed from AM, so there's no point in testing it for 32,000 matches because anyway, it isn't possible to "isolate" the absence of "Overlaps" TIs in that case.
Zippo said: Here's data for Katana 4231 104p v3.1 tactic with "Overlaps" TIs and without "Overlaps" TIs after 32,000 matches tested.
Judge for yourself.
There's no obvious negative effect from "Overlaps" TIs as it happened for the two previous "generic" tactics.
It seems the "Attacking" mentality or other TIs and PIs might influence on how "Overlaps" TIs perform and in Katana 4231 104p v3.1 the "Overlaps" TI even give a small positive effect which is about "0.1" points but don't forget even 32,0000 matches test also have at least 0.2 RNG.
@alex this tactic - https://fm-arena.com/thread/9578-katana-4231-104p-v3-1-27/ might scored that low due to the fact that the "Overlaps" TIs weren't the only instructions that were changed, I noticed that "Moves Into Channels" PI was also removed from AM, so there's no point in testing it for 32,000 matches because anyway, it isn't possible to isolate the absence of "Overlaps" TIs in that case.
Zippo said: @alex this tactic - https://fm-arena.com/thread/9578-katana-4231-104p-v3-1-27/ might scored that low due to the fact that the "Overlaps" TIs weren't the only instructions that were changed, I noticed that "Moves Into Channels" PI was also removed from AM, so there's no point in testing it for 32,000 matches because anyway, it isn't possible to isolate the absence of "Overlaps" TIs in that case.
I did initially forget to add that I have removed the move into channels on the AM in the opening post, I have since updated the OP.
V3.1.26 has that instruction on there however
@Zippo thanks for the test. And yeah, I noticed later that move into channels was also removed from the am, besides the overlaps. Ok, good to know
I think all I can say is that this result is accurate at the two fixed combinations.none, overlaps, and underlaps are different for position duty selection and optimal selection for PI. I tested a variety of different 4231 Settings early in the release, both single-player and online, and couldn't come to a conclusion about which was stronger. Moreover, only with an ST formation can the attacking mentality compete with the balanced mentality, but the attacking mentality is stronger than the attacking mentality in the online game. 4231 is indeed a top formation, but it can occupy the top or rely on more test samples and luck, because 424 and 4231 are the first two tactics to occupy the top for a long time, so we fine-tune the test too much. There are actually some formations that don't lose to either.
Zippo said: a small positive effect which is about "+0.1" points
I don't know who need TIs that make such a small difference? What is the point in having them in the game?
It doesn't mean they don't do anything, only that the total effect of various options is quite close. Which is good. People complain non stop how gegenpress is the only way to play the game and that you cannot do other styles. And now when you have 3 more or less equal options it's again not right...
Zippo said: Here's data for Katana 4231 104p v3.1 tactic with "Overlaps" TIs and without "Overlaps" TIs after 32,000 matches tested.
Judge for yourself.
There's no obvious negative effect from "Overlaps" TIs as it happened for the two previous "generic" tactics.
It seems the "Attacking" mentality or other TIs and PIs might influence on how "Overlaps" TIs perform and in "Katana 4231 104p v3.1" tactic the "Overlaps" TI even give a small positive effect which is about "+0.1" points but don't forget that even 32,0000 matches test also have at least 0.2 RNG.
@alex this tactic - https://fm-arena.com/thread/9578-katana-4231-104p-v3-1-27/ might scored that low due to the fact that the "Overlaps" TIs weren't the only instructions that were changed, I noticed that "Moves Into Channels" PI was also removed from AM, so there's no point in testing it for 32,000 matches because anyway, it isn't possible to "isolate" the absence of "Overlaps" TIs in that case.
Amazing test. Do you think of doing the same for focus play down the left/right?