We've implemented a new algorithm for determining the number of matches to test tactics.
We find the new algorithm has an excellent balance between "testing speed" and "result accuracy".
I want to explain you in details how it works.
Every tactic should get 2 runs(800 matches) by default.
If after 2 runs a tactic hits 49 points or above, then it'll be given 1 additional run for total 3 runs(1,200 matches)
If after 3 runs(1,200 matches) a tactic hits 51 points or above and also manages to get in the Hall Of Fame or get a score that is no more than 1 point less than the score of its competitor in the Hall of Fame then it'll be given 7 additional runs for total 10 runs(10,000 matches). Please note, it might not happen instantly but don't worry, it'll happen at some point.
Here's a small example how it works. Let's imagine there's 4-2-3-1 with 53 pts in the Hall of Fame and after 3 runs(1,200 matches) your 4-2-3-1 got 52 pts, it means that your 4-2-3-1 is qualified to get 7 additional runs. As I already said, the 7 additional runs might not happen instantly but don't worry, it'll happen at some point.
Also, I want to add that some additional improvements that we made to our new tactic testing league greatly helped to reduce the RNG( almost by half) so the RNG of 3 runs(1,200 matches) on our new tactic testing league is no higher than -/+ 2 points from the avg. of 10 runs(10,000 matches), which means if after 3 runs (1,200 matches) a tactic hit 2 points less than its competitor in the Hall of Fame then it already won't be able to improve the score of its competitor even if it gets 10 runs, so not running it for additional runs in this case, help us not to waste the computing power of the testing server for noting.
Cheers. Expand
Sometimes i do watch at those changelogs like i look at a "hot" chica :
dzek said: The only thing is that the game, year after year, is going downhill. If you pay attention to the official SI community, many of them are talking about the downfall of the game and I too wonder how far this can go.
Every year we have a "new" game the same as the previous year. They just upgrade its ugly graphics, introduce new unimportant(and often non-functional) features and it seems the teams working on the game have no cohesion with each other.
If we all start posting errors or omissions from something that came to our attention in-game then we should call the community FM-Buggers. Expand
And how SI is even supposed to listen to before i would've said "period" people now they feel like on full hemorrhoids.
You are right, but to me the problem is that: Those so called "tester" or whoever SI is working to in order to test the game are not the right people to help the game to get development on the right direction.
It's no sense that we have like 36 stats (?) and how many of them are usefull? Ah, right we already know that...
They should work on the Attributes first, since this seem to be really no-sense over and over.
And jesus, implemenent a "regen-3D pack" or whatever we wanna call it.
More i test more i do find cosmetic TI/PI.
Did someone yet understand why did they implement "trap inside/Outside"/ Stop Crosses/Invite Crosses?
It feel like they used multiple code language packs over the years and now all the mess is coming out.
How do we apply Zippo to be the Chief Executor Tester. Maybe we can do that behind his back and we come out with a big "SURPRISE".
Or at least we can elaborate a "smart petition" with logical arguments ?_?
Removed Get Stuck in and Winger ispired about this result : https://fm-arena.com/thread/4920-zaz-major-tweak-hurry/
Removing some cosmetic (at least from my point of view)
Structura TI = MHD and Def combo (Cover+Defend)
Added Overlaps + focus flanks
Added Overlaps and Focus Flanks
Adapted Swanton Bomb V with Structura (Basically is One Defender on cover and one defender duty)
For TI and PI :
FM23 Spaghetti League - 433 Structura Bomb V
The results on Spaghetti league are :
8x 🍝 Leagues
6x 🍜 Cups
points on average = 84
IWs isteand of IFs
I did tested DM+RPM combo like tons of time let's see if now is avaible
Been a while i didn't made some video-editing and OBS seem a bit against it, i might have fixed the resolution now
2nd Season with Bournemouth
Yes, now that Arena is testing like 1000000 of tactics i am making some video-editing stuff
I suggest you to take a look here : https://fm-arena.com/thread/5351-should-you-follow-the-highlighted-attributes-of-the-roles/§
And here :
https://fm-arena.com/thread/5345-attributes-vs-tactics/
As you can see even AM Likes Lens! YELLOW/RED like the best team world o.o
9th!
Did just va bigger teams
We've implemented a new algorithm for determining the number of matches to test tactics.
We find the new algorithm has an excellent balance between "testing speed" and "result accuracy".
I want to explain you in details how it works.
Every tactic should get 2 runs(800 matches) by default.
If after 2 runs a tactic hits 49 points or above, then it'll be given 1 additional run for total 3 runs(1,200 matches)
If after 3 runs(1,200 matches) a tactic hits 51 points or above and also manages to get in the Hall Of Fame or get a score that is no more than 1 point less than the score of its competitor in the Hall of Fame then it'll be given 7 additional runs for total 10 runs(10,000 matches). Please note, it might not happen instantly but don't worry, it'll happen at some point.
Here's a small example how it works. Let's imagine there's 4-2-3-1 with 53 pts in the Hall of Fame and after 3 runs(1,200 matches) your 4-2-3-1 got 52 pts, it means that your 4-2-3-1 is qualified to get 7 additional runs. As I already said, the 7 additional runs might not happen instantly but don't worry, it'll happen at some point.
Also, I want to add that some additional improvements that we made to our new tactic testing league greatly helped to reduce the RNG( almost by half) so the RNG of 3 runs(1,200 matches) on our new tactic testing league is no higher than -/+ 2 points from the avg. of 10 runs(10,000 matches), which means if after 3 runs (1,200 matches) a tactic hit 2 points less than its competitor in the Hall of Fame then it already won't be able to improve the score of its competitor even if it gets 10 runs, so not running it for additional runs in this case, help us not to waste the computing power of the testing server for noting.
Cheers.
Sometimes i do watch at those changelogs like i look at a "hot" chica :
Just doing a journey save while FM-Arena is testing all the old tactics
Removed underlaps
For Console/Tablet i will start to make mini-videos review i will make "game-play" if i do really like the results or they did very well on Arena.
Every year we have a "new" game the same as the previous year. They just upgrade its ugly graphics, introduce new unimportant(and often non-functional) features and it seems the teams working on the game have no cohesion with each other.
If we all start posting errors or omissions from something that came to our attention in-game then we should call the community FM-Buggers.
Ok, i did a mistake again and went to read : https://community.sigames.com/forums/topic/569294-official-football-manager-2023-feedback-thread/page/89/
And how SI is even supposed to listen to before i would've said "period" people now they feel like on full hemorrhoids.
You are right, but to me the problem is that: Those so called "tester" or whoever SI is working to in order to test the game are not the right people to help the game to get development on the right direction.
It's no sense that we have like 36 stats (?) and how many of them are usefull? Ah, right we already know that...
They should work on the Attributes first, since this seem to be really no-sense over and over.
And jesus, implemenent a "regen-3D pack" or whatever we wanna call it.
More i test more i do find cosmetic TI/PI.
Did someone yet understand why did they implement "trap inside/Outside"/ Stop Crosses/Invite Crosses?
It feel like they used multiple code language packs over the years and now all the mess is coming out.
How do we apply Zippo to be the Chief Executor Tester. Maybe we can do that behind his back and we come out with a big "SURPRISE".
Or at least we can elaborate a "smart petition" with logical arguments ?_?
But i don't think they even read those.
Please tell me ...