At the moment our testing algorithm looks like this:
Every tactic gets 2 testing rounds by default ( 1 round is 320 matches so 2 rounds would be 640 matches ) then our testing server calculates the result and if after 2 first testing rounds the result is below 48 points then the test is over and the testing server uploads the result but if after 2 first testing rounds the result is above 48 points then it gets an additional testing round ( 1 round = 320 matches ) and after that additional testing round the testing server calculates the result again and if it hits below 48 points then the test is over and the testing server uploads the result but if it hits above 48 points then it gets another additional testing round and that continues up to maximum 9 rounds ( 2,880 matches ).
The purpose of the algorithm above is to keep the results at the top as accurate as possible and the same time consume the resources of our testing server at reasonable rate.
Please note, that we can "tune" the numbers in our testing algorithm at any moment depending on the situation. For example, at this moment we have reduced the maximum testing rounds from 9 rounds ( 2,880 matches ) to 6 rounds ( 1,920 matches ) because there are lot of tactics in the queue await to be tested at the moment.
pixar said: why are some tactics being tested more often? Is there a standard for this? How do you decide for how many matches a tactic will be tested?
for example why was this tactic tested for 1920 matches?
At the moment our testing algorithm looks like this:
Every tactic gets 2 testing rounds by default ( 1 round is 320 matches so 2 rounds would be 640 matches ) then our testing server calculates the result and if after 2 first testing rounds the result is below 48 points then the test is over and the testing server uploads the result but if after 2 first testing rounds the result is above 48 points then it gets an additional testing round ( 1 round = 320 matches ) and after that additional testing round the testing server calculates the result again and if it hits below 48 points then the test is over and the testing server uploads the result but if it hits above 48 points then it gets another additional testing round and that continues up to maximum 9 rounds ( 2,880 matches ).
The purpose of the algorithm above is to keep the results at the top as accurate as possible and the same time consume the resources of our testing server at reasonable rate.
Please note, that we can "tune" the numbers in our testing algorithm at any moment depending on the situation. For example, at this moment we have reduced the maximum testing rounds from 9 rounds ( 2,880 matches ) to 6 rounds ( 1,920 matches ) because there are lot of tactics in the queue await to be tested at the moment.
I would like to note that even when we test a tactic for 2K-3K matches there still might be +/- 1 or 2 points RNG, which can't be eliminated even when we test that many matches.
Please, provide results from a normal game save. Any custom leagues result doesn't count.
Also, if this tactic is a tweaked version of a tactic that has been tested already then we need to know what tweaks were made to evaluate them if they're worth testing.
tom100000000000 said: So if 2.5CA will get you 1 point in acceleration but 8 points in teamwork, its still probably worth it. 1 acc > 8 teamwork probably, which is crazy Expand For example, 'Decisions' attribute is very expensive for every position and it costs almost the same as 'Acceleration' or 'Pace' attributes but 'Decisions' attribute almost doesn't have any significant impact on the result unlike the 'Acceleration' or 'Pace' attributes, which have a great impact on the result.
But I think there might be people who don't fully understand why we're doing that and how it can be used in their saves.
So I just want to reveal you the power of the player attributes testing.
Here's a striker from our tactic testing league. He is about 145CA:
If you look at the results of our attribute testing then you'll notice that some attributes such as 'Acceleration' or 'Pace' makes a huge difference and some attributes such as 'Teamwork' or 'Off The Ball' almost doesn't make any difference.
Of course, it's important to understand that 'Acceleration' and 'Pace' attributes are much more expensive in terms of CA(Current Ability) than 'Teamwork' or 'Off The Ball' attributes. For example, 1 Point of Acceleration costs about 2.5CA when 1 Point of Teamwork costs only 0.3CA.
But what if take our striker and redistribute his CA from less important attributes such as Composure, Teamwork, Off The Ball, Long Shots to very important important attributes such as Acceleration and Pace keeping the same CA:
Notice that the CA of our striker hasn't changed, it's still 145CA, we just redistributed it.
Now, let's do a similar redistribution for other positions keeping the same CA and then test it:
Default Attributes Result:
Redistributed Attributes Result:
As you can see the points have skyrocketed from 53 to 90 and the G.D. has skyrocketed from +3 to +62 but the CA of the players hasn't changed!
From the result above we can conclude if players have the same CA then it doesn't mean they are equal quality, some of them might be much better than others. Obviously, in FM some attributes such as Acceleration and Pace are game-changing attributes, they make a huge difference and you should prioritize them.
Would the performance difference be 15% worse for a player ranked in position 17 than a player ranked 20? Or 1.36% for one player and 15% if everyone has 17 instead of 20? Expand
In this test the playing position rating of all the players was decreased from Natural(20) to Accomplished(17).
Obviously, if only one player had the playing position rating decreased then the result would be much better.
crizeKOS said: For sure there is some stuff that are just cosmetic in this game. Check it out guys ^^ Expand
Indeed, there's a lot of instructions that make like about 1% difference and you really can call them 'cosmetic'.
At the moment, we've done many internal testing and tested many different instructions like different mentalities and D-lines, as soon as we process the collected data we'll share it.
Droid said: I guess it'll be like it was in FM22, at some point later they we'll give the top tested tactics additional testing rounds to reach 3,000 matches Expand
yes, that's what we're planning to do in the future
If you want your tactic gets tested then the following minimal requirements must be met.
You need to provide results with your tactic. The results must be in ANY of the leagues below:
England - English Premier league Spain - La Liga German - Bundesliga French - League 1 Italy - Serie A Portugal - Liga Portugal Netherlands - Eredivisie
ALL SCREENSHOTS MUST BE IN ENGLISH
You must provide a screenshot of the tactic:
You must provide a screenshot of the league table. At least one full season must be played ( FOR/AG/G.D. stats must be visible ):
You must provide a screenshot of the team selection ( Goals/Assists/Ratings stats must be visible )
You must provide a screenshot of the fixtures ( from the start to the end ):
- The result must be yours and not a copy/paste of someone else's result. - The result must be from a normal game and not from a custom testing league. - Using in-game editors to boost your players is restricted. You should play the game in the normal way. - If your tactic is based on someone else's tactic then you should mention it. - If you make tweaks then you should say exactly what tweaks you made. For example, "I've added "Pass Into Space" TI and removed "Runt at Defense" TI." - Mirror versions of tested tactics aren't not allowed for testing. - Give your tactic a readable and distinguishable name in English. Avoid using acronyms and abbreviations. For example, the following names aren't acceptable: "442", "442 IFs IWBs", "433 DTF", "4-2-3-1", "4-3-3 Balanced", "5-3-2 attacking", "4-4-2 Wide", "4-4-2 Narrow", "451 WBT TDV ZBF", "4-3-3 3xAFs" and similar undistinguishable and unreadable names that contain acronyms and abbreviations. - Use only English language.
Only if the minimal requirements are met then a tactic is considered as a candidate for the testing but meeting the minimal requirements still doesn't guarantee that a tactic get tested because there're other factors that we take into consideration such as:
- The better presentation of your tactic, the more likely it gets tested so don't be lazy and add any additional useful information about your tactic. The more reasons you give us for it gets tested, the better.
IMPORTANT: We can't promise that we test every uploaded tactic and you shouldn't expect that. It's unlikely that we test your tactic if we find it's very similar to what that has been tested already so if you want your tactic to get tested then make sure you offering really new ideas for the testing.
At the moment our testing algorithm looks like this:
Every tactic gets 2 testing rounds by default ( 1 round is 320 matches so 2 rounds would be 640 matches ) then our testing server calculates the result and if after 2 first testing rounds the result is below 48 points then the test is over and the testing server uploads the result but if after 2 first testing rounds the result is above 48 points then it gets an additional testing round ( 1 round = 320 matches ) and after that additional testing round the testing server calculates the result again and if it hits below 48 points then the test is over and the testing server uploads the result but if it hits above 48 points then it gets another additional testing round and that continues up to maximum 9 rounds ( 2,880 matches ).
The purpose of the algorithm above is to keep the results at the top as accurate as possible and the same time consume the resources of our testing server at reasonable rate.
Please note, that we can "tune" the numbers in our testing algorithm at any moment depending on the situation. For example, at this moment we have reduced the maximum testing rounds from 9 rounds ( 2,880 matches ) to 6 rounds ( 1,920 matches ) because there are lot of tactics in the queue await to be tested at the moment.
for example why was this tactic tested for 1920 matches?
or why is this super crazy tacticc tested for 2880 matches?
At the moment our testing algorithm looks like this:
Every tactic gets 2 testing rounds by default ( 1 round is 320 matches so 2 rounds would be 640 matches ) then our testing server calculates the result and if after 2 first testing rounds the result is below 48 points then the test is over and the testing server uploads the result but if after 2 first testing rounds the result is above 48 points then it gets an additional testing round ( 1 round = 320 matches ) and after that additional testing round the testing server calculates the result again and if it hits below 48 points then the test is over and the testing server uploads the result but if it hits above 48 points then it gets another additional testing round and that continues up to maximum 9 rounds ( 2,880 matches ).
The purpose of the algorithm above is to keep the results at the top as accurate as possible and the same time consume the resources of our testing server at reasonable rate.
Please note, that we can "tune" the numbers in our testing algorithm at any moment depending on the situation. For example, at this moment we have reduced the maximum testing rounds from 9 rounds ( 2,880 matches ) to 6 rounds ( 1,920 matches ) because there are lot of tactics in the queue await to be tested at the moment.
Hey,
I would like to note that even when we test a tactic for 2K-3K matches there still might be +/- 1 or 2 points RNG, which can't be eliminated even when we test that many matches.
More info - https://fm-arena.com/thread/2713-10-944-matches-tested-fm-rng-measured/
The first results should be available within an hour.
We'll try to find a balance between testing existing tactics and new tactics but of course, the top tactics will be tested on a priority basis.
Hi,
Please, provide results from a normal game save. Any custom leagues result doesn't count.
Also, if this tactic is a tweaked version of a tactic that has been tested already then we need to know what tweaks were made to evaluate them if they're worth testing.
Thank you.
For example, 'Decisions' attribute is very expensive for every position and it costs almost the same as 'Acceleration' or 'Pace' attributes but 'Decisions' attribute almost doesn't have any significant impact on the result unlike the 'Acceleration' or 'Pace' attributes, which have a great impact on the result.
I'm sure you guys are aware that we've been testing player attributes to find out which of them make the most difference - https://fm-arena.com/table/18-fm23-attribute-testing/
But I think there might be people who don't fully understand why we're doing that and how it can be used in their saves.
So I just want to reveal you the power of the player attributes testing.
Here's a striker from our tactic testing league. He is about 145CA:
If you look at the results of our attribute testing then you'll notice that some attributes such as 'Acceleration' or 'Pace' makes a huge difference and some attributes such as 'Teamwork' or 'Off The Ball' almost doesn't make any difference.
Of course, it's important to understand that 'Acceleration' and 'Pace' attributes are much more expensive in terms of CA(Current Ability) than 'Teamwork' or 'Off The Ball' attributes. For example, 1 Point of Acceleration costs about 2.5CA when 1 Point of Teamwork costs only 0.3CA.
But what if take our striker and redistribute his CA from less important attributes such as Composure, Teamwork, Off The Ball, Long Shots to very important important attributes such as Acceleration and Pace keeping the same CA:
Notice that the CA of our striker hasn't changed, it's still 145CA, we just redistributed it.
Now, let's do a similar redistribution for other positions keeping the same CA and then test it:
Default Attributes Result:
Redistributed Attributes Result:
As you can see the points have skyrocketed from 53 to 90 and the G.D. has skyrocketed from +3 to +62 but the CA of the players hasn't changed!
From the result above we can conclude if players have the same CA then it doesn't mean they are equal quality, some of them might be much better than others. Obviously, in FM some attributes such as Acceleration and Pace are game-changing attributes, they make a huge difference and you should prioritize them.
Would the performance difference be 15% worse for a player ranked in position 17 than a player ranked 20? Or 1.36% for one player and 15% if everyone has 17 instead of 20?
In this test the playing position rating of all the players was decreased from Natural(20) to Accomplished(17).
Obviously, if only one player had the playing position rating decreased then the result would be much better.
We're going to start testing the attributes soon, unfortunately, I can't give you any ETA.
Please, add a screenshot of the tactic.
Thank you.
Please, refrain from making off topic comments. This thread is about Vujevic's tactics.
Thank you.
Hi,
Our testing server's been busy with testing tactical instructions, when we're done with them then we'll continue testing tactics as usual.
Cheers.
https://fm-arena.com/board/10-tactical-instructions-testing/
Our testing server's been busy with testing tactical instructions, when we're done with them then we'll continue testing tactics as usual.
I'm 99% sure that nothing has changed regarding that area but as I sad we'll check at some point later.
Yes, it's in our plans but at some point latter.
Indeed, there's a lot of instructions that make like about 1% difference and you really can call them 'cosmetic'.
At the moment, we've done many internal testing and tested many different instructions like different mentalities and D-lines, as soon as we process the collected data we'll share it.
yes, that's what we're planning to do in the future
I've edited the thread and tactic names.
Please, use only English.
Thank you.
If you want your tactic gets tested then the following minimal requirements must be met.
You need to provide results with your tactic. The results must be in ANY of the leagues below:
England - English Premier league
Spain - La Liga
German - Bundesliga
French - League 1
Italy - Serie A
Portugal - Liga Portugal
Netherlands - Eredivisie
ALL SCREENSHOTS MUST BE IN ENGLISH
You must provide a screenshot of the tactic:
You must provide a screenshot of the league table. At least one full season must be played ( FOR/AG/G.D. stats must be visible ):
You must provide a screenshot of the team selection ( Goals/Assists/Ratings stats must be visible )
You must provide a screenshot of the fixtures ( from the start to the end ):
- The result must be yours and not a copy/paste of someone else's result.
- The result must be from a normal game and not from a custom testing league.
- Using in-game editors to boost your players is restricted. You should play the game in the normal way.
- If your tactic is based on someone else's tactic then you should mention it.
- If you make tweaks then you should say exactly what tweaks you made. For example, "I've added "Pass Into Space" TI and removed "Runt at Defense" TI."
- Mirror versions of tested tactics aren't not allowed for testing.
- Give your tactic a readable and distinguishable name in English. Avoid using acronyms and abbreviations. For example, the following names aren't acceptable: "442", "442 IFs IWBs", "433 DTF", "4-2-3-1", "4-3-3 Balanced", "5-3-2 attacking", "4-4-2 Wide", "4-4-2 Narrow", "451 WBT TDV ZBF", "4-3-3 3xAFs" and similar undistinguishable and unreadable names that contain acronyms and abbreviations.
- Use only English language.
Only if the minimal requirements are met then a tactic is considered as a candidate for the testing but meeting the minimal requirements still doesn't guarantee that a tactic get tested because there're other factors that we take into consideration such as:
- The better presentation of your tactic, the more likely it gets tested so don't be lazy and add any additional useful information about your tactic. The more reasons you give us for it gets tested, the better.
IMPORTANT: We can't promise that we test every uploaded tactic and you shouldn't expect that. It's unlikely that we test your tactic if we find it's very similar to what that has been tested already so if you want your tactic to get tested then make sure you offering really new ideas for the testing.