dzek said: Read it again and pay attention to the big bold letters.
Let me give you an example. All AI managers without exception use IWB(s) on DR/WBR and WB(a) on DL/WBL and after my own testing and that of others here and elsewhere we conclude that IWBs inside FM24 are weak at defending. So with that in mind I can simply put in 'Focus Play Down The Left' & 'Overlap Left' and I will get better results with almost all formations and roles that encourage these moves. There are other things we can take as examples as well, but I think you get the point here.
Other than that I have my opinion (whether right or wrong) and anyone can refute me with facts so that I can find out where I am wrong.
The fact that this year WB/FB (and not just this year) are very strong and playing in this particular tactic against a W and an IWB, on the left side completely alone, but at the same time being efficient doesn't tell you anything? I consistently get better results from my DR than from my DL and I use a test league that is almost identical to FMArena's. And just to catch you up in case you happen to have read my other thread here, no it's not that test league I'm referring to. Expand
Mate, some interaction are better with overalp some others are not. It's pretty simple. What you should check if how much those "TI" literally impact on the scenario you did literally mentioned.
About this test, to be honest the only thing caught my attention is "GK on Attack".
Gianaa9 said: How can you have a so rapid test run? It will be so useful e.e Expand
As i said there are many ways to make a test faster and i am using the old model, i didn't yet finished my web site did some craps around, lost even 100+ tactics in the process and on overall i tested more then 800 tactics at the moment, 500+ on 4k games.
Now i am trying to understand how to make discord api interacting with the Site, so i won't even upload the tactics over my test machine.
Here for example is easier because the forum is doing that part. But i am not really interested yet to build a forum. Prefer to have my community on discord.
I made the threshold for "PERFECT/BRILLIANT/POTATO" material. So we know already they gonna be super bad.
I mean there is a way to make this schedule with "by creating cups", but i didn't yet understand how to make it into a league (didn't really checked it yet), but if people won't share their knowdledge i do pretty understand why.
This is my next step to automate everything about testing tactics. Do you think you'll see any difference with the runs until the 8-12k matches? Regarding the players I have a different approach that I have yet to apply and I am considering whether it is worth it or not after what I read and see on the official SI forum. Expand
Pretty sure it does and will at the moment some tactics are biased by some negative runs or positive one.
dzek said: Brother I really appreciate your concern and indeed many don't know how much effort it takes to build such a thing from scratch. This "User vs AI" matchmaking thing I prefer not to share (at least for now) and hope for your understanding.
Because of the opportunity I would like to say that I built this particular database almost 5 times to date from scratch. The first 4 times I was having problems with my RAM and was shutting down my computer halfway through. You can't imagine how angry I was but I immediately regained my appetite and started all over again with the same zeal to finish it. Expand
I do understand, it's just something that i wish to learn by watching a guide, doesn't mean you are forced to share it, literally none on the community does. There are literally many ways to gain processing speed.
About time believe me, to make very fuctional automatizzated test took me like 3-4 months of work. But at least i can literally push out 30-50 tactics per day tested 4k times if they do 45+ .
Now building a method to test top 50 or 100 (need to decide yet) to 8k or 12k. Will implement at a point even min and max graphs etc.
That's why i am looking to make it faster then 300 sec, because it means you can make push more and more.
For example about players building, i would something on average 160CA players stats vs 140CA average stats.
There are very good way to build them around and it's very funny.
As I have said above but you probably didn't even bother to read, the test league is still not quite ready. When I feel it's ready and I WANT it, I'll give it away publicly or to those who have shown interest here. We have a life outside of here you know. If you can't wait then you can create your own and I'm sure no one will stop you from doing it. If on the other hand you can't create your own and want something ready then please be patient!
Thank you so much for your interest!
Edit: Just so you know I've spent almost 1-2 months perfecting it and I wasn't going to upload it. I made the decision to upload it and all I ask is patience. Appreciate a little of the work others do for you and learn not to demand once you get it for free. Expand
People don't even know how complicated can be to interact with that thing called database.
Btw, i am really interested to learn how to build that calendar schedule.
If you like to explain those stuff, make a youtube video or idk. Since people nowdays are getting attention with 2 elementar python's skill, this is at least for me is way more interesting tool or something affascinating to learn.
What i am curious is just about if the speed is faster then "old-build".
dzek said: Haha that's very easy but the hard part is making the AI Managers unsackable. Luckily I found a way and I have both of them in my test leagues. Expand
It's more complicated more that schedule then Ai managers unsackable 😂
dzek said: It's based on the FM-Arena test league, so it has a lot of similar things in it and for me what I'm most interested in is the running time and I think it's pretty fast. I haven't measured it though.
The players are the same as they are in the FM-Arena Tactic Testing League. Nothing different in that regard. Expand
I mean if you can even make a guide about how to make Bot vs Test would be great. Literally there is no-one making guides about the editor.
I am interested as well. But do you achieved to make like FM-Arena 10 team 5 "bot" and 5 "test teams" as well? Trying to figuring out that. But at the end of the day for me what matter is "time". For now every test on my side take make 300 sec on my pc and 340 sec on test league machine.
And pretty curious how you built the players as well.
- Ask your self how much cost wakanda-forever and how much cost Odri.
Now we supposed that the Purity's "test" is attendibile (means not much RNG involved)
The different between them is how much 0.10+ on Odri's favor. (You can apply same method to Adama and Junior)
- Now after that you can see where the poor Wakanda or Adama falls down are the hidden attribute.
Since @NandaldiaN screen shows even CA and hidden attributes (you can use fmrte or just going guessing with icons)
We can see that wakanda forever has :
-14 consistency -9 Temperament -14 Pressure
(it's even mono leg basically)
vs Odri
-16 Consistency -14 Temperament -12 Pressure
one is 123 CA and the other 139
(keep in mind this my understanding that those 3 hidden stats are boosting or killing a player perfomance)
Now we should know how exatcly those hidden affecting on the m.e. and alot of infinite blabering.
Since both of players won't really develop any futher :
If i pay my wakanda forever around 10 and Odri around 20 it's worth to spend those 10 mil more for 0.10+ av.Rate? (you can even pay 6 and 13) same concept.
Another thing, when you are testing players, keep in mind to put the same hiddens if you want to compare which are more impactful. (That's how stats were tested on FM-Arena)
At the end of the day i would go Wakanda. Simple because Odri won't scale and i can make profit from Wakanda.
NandaldiaN said: Wemby: 76,08 DR / 73,31 WBR Odry: 83,25 DR / 81,57 WBR
Dont know if these results are more accurate for you, i would still choose the fastest player anyways following test here. But maybe im wrong and odry is far better in the match engine... i really dont know
I think every test/about the game is worth, especially here in this website, each of us is here for that exact reason 😀 Expand
What i meant isn't something easy to compile. Literally every stats should be calculate how affect the M.E. after a certain point. That's why i said it might not worth if they gonna change it all for FM25
Mark said: For what it is worth both ykykyky and MDW22 rate Odrizola as the better of the two assuming their DR positional ratings are both 20/20. Calculating them manually using ykykyky I have Odrizola 78.77% and Wembangomo 75.75%, and using MDW22 I make it Odrizola 78.60% and Wembangomo 73.73%. Expand
I literally have no clue what are those rating, but still at least on FM24 i would never go for Odrizola between the 2. Rating or not rating. I will soon release the Lincoln adventure that i am doing.
I am sure Stats are scaling in the wrong way possible. like after 10 some stats are literally doing nothing, i did noticed it by trying to build de zerbi possession/control tactic on FM24. Literally you can dominate possession easier with those gegenpress tactics. If you have the right players on the right positions.
Like with City, with "plugNplay" tactics you will reach very easy 55% possession. And if you literally go after possession you can get out 65%. But you will lose too much in this transition.
I mean we should analyze how much Physical/Mentality are scaling but i don't even know if that's is any worth "project" since they might change everything for FM25
NandaldiaN said: Using your rating system in my current save, there is one in particular that catches my attention and I would like to ask the people on the forum which of these 2 they would choose to play on their team.
Currently your rating puts Odriozola as a 68% rating, compared to Wembangomo with a 61% for the DR/WBR position.
It is obvious that in everything other than physical statistics, Odriozola is better, but if we stick to the tests here, wouldn't Wembangomo be much superior in the match engine? Expand
dzek said: The question for this case actually is if the “Use of Subs” attribute has an effect of when and/or if subs occur inside the game. Expand
Keep in mind that players will get tired during the game,even tho they are frozen on coditions/morale/etc. So if test team or IA team can't sub out it may affect the results.
dzek said: That's why I said above when I tested them in my own tests they were more reliable in many parts of the game. Definitely a slight change though, it only has to do with the RNG. The goal is to find the weakness of ME, i.e. how your team builds up and how your team attacks, and try to use the right roles and team/player instructions to break it down.
FBs alone gave good results regardless of what roles were around them but that again we can find the best combination to maximize the effect.
It could, but I think "Get Further Forward" does already a lot of work along with the team and player mentality. Expand
I can tell you, once i will start to run 8k runs or 12k. But literally at this point, to me 90% is a cosmetic scenario. And we just build a "puzzle" the tactic. PI/TI interaction is very strange.
At first i did believe it was like you said on FM23, that there was an hidden "multiplier" like pass into space + take more risk was rising it to 20 (example).
But literally they are the same thing on FM24. And not every role needs it. Even Hold up is doing something good. (Or maybe is another cosmetic feature, can't understand from the game, because i can't notice any difference).
But the CF(su) results are leading me to thinking the contrary.
Let me give you an example.
All AI managers without exception use IWB(s) on DR/WBR and WB(a) on DL/WBL and after my own testing and that of others here and elsewhere we conclude that IWBs inside FM24 are weak at defending. So with that in mind I can simply put in 'Focus Play Down The Left' & 'Overlap Left' and I will get better results with almost all formations and roles that encourage these moves. There are other things we can take as examples as well, but I think you get the point here.
Other than that I have my opinion (whether right or wrong) and anyone can refute me with facts so that I can find out where I am wrong.
The fact that this year WB/FB (and not just this year) are very strong and playing in this particular tactic against a W and an IWB, on the left side completely alone, but at the same time being efficient doesn't tell you anything? I consistently get better results from my DR than from my DL and I use a test league that is almost identical to FMArena's. And just to catch you up in case you happen to have read my other thread here, no it's not that test league I'm referring to.
Mate, some interaction are better with overalp some others are not. It's pretty simple. What you should check if how much those "TI" literally impact on the scenario you did literally mentioned.
About this test, to be honest the only thing caught my attention is "GK on Attack".
As i said there are many ways to make a test faster and i am using the old model, i didn't yet finished my web site did some craps around, lost even 100+ tactics in the process
Now i am trying to understand how to make discord api interacting with the Site, so i won't even upload the tactics over my test machine.
Here for example is easier because the forum is doing that part. But i am not really interested yet to build a forum. Prefer to have my community on discord.
I made the threshold for "PERFECT/BRILLIANT/POTATO" material. So we know already they gonna be super bad.
I mean there is a way to make this schedule with "by creating cups", but i didn't yet understand how to make it into a league (didn't really checked it yet), but if people won't share their knowdledge i do pretty understand why.
This is my next step to automate everything about testing tactics. Do you think you'll see any difference with the runs until the 8-12k matches? Regarding the players I have a different approach that I have yet to apply and I am considering whether it is worth it or not after what I read and see on the official SI forum.
Pretty sure it does and will at the moment some tactics are biased by some negative runs or positive one.
It's like literally flipping a coin.
Good luck on automatizzating it :P
Because of the opportunity I would like to say that I built this particular database almost 5 times to date from scratch. The first 4 times I was having problems with my RAM and was shutting down my computer halfway through. You can't imagine how angry I was but I immediately regained my appetite and started all over again with the same zeal to finish it.
I do understand, it's just something that i wish to learn by watching a guide, doesn't mean you are forced to share it, literally none on the community does. There are literally many ways to gain processing speed.
About time believe me, to make very fuctional automatizzated test took me like 3-4 months of work. But at least i can literally push out 30-50 tactics per day tested 4k times if they do 45+ .
Now building a method to test top 50 or 100 (need to decide yet) to 8k or 12k. Will implement at a point even min and max graphs etc.
That's why i am looking to make it faster then 300 sec, because it means you can make push more and more.
For example about players building, i would something on average 160CA players stats vs 140CA average stats.
There are very good way to build them around and it's very funny.
As I have said above but you probably didn't even bother to read, the test league is still not quite ready. When I feel it's ready and I WANT it, I'll give it away publicly or to those who have shown interest here. We have a life outside of here you know. If you can't wait then you can create your own and I'm sure no one will stop you from doing it. If on the other hand you can't create your own and want something ready then please be patient!
Thank you so much for your interest!
Edit: Just so you know I've spent almost 1-2 months perfecting it and I wasn't going to upload it. I made the decision to upload it and all I ask is patience. Appreciate a little of the work others do for you and learn not to demand once you get it for free.
People don't even know how complicated can be to interact with that thing called database.
Btw, i am really interested to learn how to build that calendar schedule.
If you like to explain those stuff, make a youtube video or idk. Since people nowdays are getting attention with 2 elementar python's skill, this is at least for me is way more interesting tool or something affascinating to learn.
What i am curious is just about if the speed is faster then "old-build".
It's more complicated more that schedule then Ai managers unsackable 😂
Yes!
The players are the same as they are in the FM-Arena Tactic Testing League. Nothing different in that regard.
I mean if you can even make a guide about how to make Bot vs Test would be great. Literally there is no-one making guides about the editor.
And pretty curious how you built the players as well.
So basically you have 2 players :
- Ask your self how much cost wakanda-forever and how much cost Odri.
Now we supposed that the Purity's "test" is attendibile (means not much RNG involved)
The different between them is how much 0.10+ on Odri's favor.
(You can apply same method to Adama and Junior)
- Now after that you can see where the poor Wakanda or Adama falls down are the hidden attribute.
Since @NandaldiaN screen shows even CA and hidden attributes (you can use fmrte or just going guessing with icons)
We can see that wakanda forever has :
-14 consistency
-9 Temperament
-14 Pressure
(it's even mono leg basically)
vs Odri
-16 Consistency
-14 Temperament
-12 Pressure
one is 123 CA and the other 139
(keep in mind this my understanding that those 3 hidden stats are boosting or killing a player perfomance)
Now we should know how exatcly those hidden affecting on the m.e. and alot of infinite blabering.
Since both of players won't really develop any futher :
If i pay my wakanda forever around 10 and Odri around 20 it's worth to spend those 10 mil more for 0.10+ av.Rate? (you can even pay 6 and 13) same concept.
Another thing, when you are testing players, keep in mind to put the same hiddens if you want to compare which are more impactful. (That's how stats were tested on FM-Arena)
At the end of the day i would go Wakanda. Simple because Odri won't scale and i can make profit from Wakanda.
My 2 cookies.
Poggers in the chat!
New grapich with unity, i don't believe they do something else without make things from 0..
Odry: 83,25 DR / 81,57 WBR
Dont know if these results are more accurate for you, i would still choose the fastest player anyways following test here. But maybe im wrong and odry is far better in the match engine... i really dont know
I think every test/about the game is worth, especially here in this website, each of us is here for that exact reason 😀
What i meant isn't something easy to compile. Literally every stats should be calculate how affect the M.E. after a certain point. That's why i said it might not worth if they gonna change it all for FM25
I literally have no clue what are those rating, but still at least on FM24 i would never go for Odrizola between the 2. Rating or not rating. I will soon release the Lincoln adventure that i am doing.
I am sure Stats are scaling in the wrong way possible. like after 10 some stats are literally doing nothing, i did noticed it by trying to build de zerbi possession/control tactic on FM24. Literally you can dominate possession easier with those gegenpress tactics. If you have the right players on the right positions.
Like with City, with "plugNplay" tactics you will reach very easy 55% possession. And if you literally go after possession you can get out 65%. But you will lose too much in this transition.
I mean we should analyze how much Physical/Mentality are scaling but i don't even know if that's is any worth "project" since they might change everything for FM25
Currently your rating puts Odriozola as a 68% rating, compared to Wembangomo with a 61% for the DR/WBR position.
It is obvious that in everything other than physical statistics, Odriozola is better, but if we stick to the tests here, wouldn't Wembangomo be much superior in the match engine?
Wakanda-forever hands down!
Keep in mind that players will get tired during the game,even tho they are frozen on coditions/morale/etc. So if test team or IA team can't sub out it may affect the results.
I don't know if is in a positive or negative way.
FBs alone gave good results regardless of what roles were around them but that again we can find the best combination to maximize the effect.
It could, but I think "Get Further Forward" does already a lot of work along with the team and player mentality.
I can tell you, once i will start to run 8k runs or 12k. But literally at this point, to me 90% is a cosmetic scenario. And we just build a "puzzle" the tactic. PI/TI interaction is very strange.
At first i did believe it was like you said on FM23, that there was an hidden "multiplier" like pass into space + take more risk was rising it to 20 (example).
But literally they are the same thing on FM24. And not every role needs it. Even Hold up is doing something good. (Or maybe is another cosmetic feature, can't understand from the game, because i can't notice any difference).
But the CF(su) results are leading me to thinking the contrary.
-^
And using a "clear" skin, because i am literally going blind sometimes to understand the hell some people using