We decided not to do 4k tests for the beta, there's really no point in that at this stage.
But 4k tests still might happens for some tactics during the beta, the purpose of that is our need to tune the configuration of our testing server and test some other things such the game RNG and other stuff.
dzek said: @Zippo, I have noticed that you remove 442 formation for AI Managers and you insert another 443. Did you find that in this ME 442 is not powerful? Expand
Yup, 4-4-2 under AI isn't impressive at all at this moment if the situation changes with patches we'll bring it back.
dzek said: And “Team C” text on testing league table is not highlighted with blue color. Its still under control of you or you increase AI teams? Expand
It looks like a UI bug in the beta, multiple human managers not returning from Holiday simultaneously at the supposed time but as I said it's more a UI bug that doesn't have any impact on the result.
@Zippo is there a possibility of testing attributes by position? I dont know if it has been done before? I think keepers could be a position this would be extremelyy useful for.
I would like to make an input here on a change to the Testing League that would probably help even more to minimize the "luck" factor.
There are 5 controlled teams and 5 AI teams. These teams all play each other and it is here that I would like to suggest since the league has this structure then it would be nice if each team played the same team in all fixtures until the end. I think that way the AI team would adapt even better to our tactics and I think that would reduce the "luck" factor.
*I'm leaving a picture of a reply given by Jack Joyce (SI dev) on how you can reduce statistical error(luck) within the game. What do you think?
It would be nice to have @Zippo's point of view here.
I would like to make an input here on a change to the Testing League that would probably help even more to minimize the "luck" factor.
There are 5 controlled teams and 5 AI teams. These teams all play each other and it is here that I would like to suggest since the league has this structure then it would be nice if each team played the same team in all fixtures until the end. I think that way the AI team would adapt even better to our tactics and I think that would reduce the "luck" factor.
*I'm leaving a picture of a reply given by Jack Joyce (SI dev) on how you can reduce statistical error(luck) within the game. What do you think?
It would be nice to have @Zippo's point of view here. Expand
Hi @dzek , In a nutshell, my understanding is that Teams A to E share identical attributes and current abilities (CA). Currently, lets say they are facing off against an AI team using a 4-2-3-1 formation. The question is: why might the statistical error vary if Team A faces the AI team 4-2-3-1 five times, as opposed to each of Teams A to E playing against the AI team 4-2-3-1 once?
sponsorkindest said: Hi @dzek , In a nutshell, my understanding is that Teams A to E share identical attributes and player abilities (PA). Currently, lets say they are facing off against an AI team using a 4-2-3-1 formation. The question is: why might the statistical error vary if Team A faces the AI team 4-2-3-1 five times, as opposed to each of Teams A to E playing against the AI team 4-2-3-1 once? Expand That's a good question to discuss as I'm not sure. The only thing that comes to my mind right now is that the AI managers have some tactical/non tactical attributes that are mostly the same across all 5 but there are differences in a few others that probably(again I'm not 100% sure) affect the way each team plays regardless if they use the same formations or even the same players attributes.
Apart from the tactical attributes, tendencies of AI Manager play a big role (during the game). That is, which instructions they will give in the game, when they will give them and if they will give them. Will they instruct their players to play out from defense? Will they work ball into the box? Will they push higher up at some point in the game? And much more..
dzek said: That's a good question to discuss as I'm not sure. The only thing that comes to my mind right now is that the AI managers have some tactical/non tactical attributes that are mostly the same across all 5 but there are differences in a few others that probably(again I'm not 100% sure) affect the way each team plays regardless if they use the same formations or even the same players attributes.
Apart from the tactical attributes, tendencies of AI Manager play a big role (during the game). That is, which instructions they will give in the game, when they will give them and if they will give them. Will they instruct their players to play out from defense? Will they work ball into the box? Will they push higher up at some point in the game? And much more.. Expand
I get your point but I don't have enuf knowledge about test league setup to say one way or the other. I am sure the moderators will know more and that's why I brought up that question.
I would like to make an input here on a change to the Testing League that would probably help even more to minimize the "luck" factor.
There are 5 controlled teams and 5 AI teams. These teams all play each other and it is here that I would like to suggest since the league has this structure then it would be nice if each team played the same team in all fixtures until the end. I think that way the AI team would adapt even better to our tactics and I think that would reduce the "luck" factor.
*I'm leaving a picture of a reply given by Jack Joyce (SI dev) on how you can reduce statistical error(luck) within the game. What do you think?
It would be nice to have @Zippo's point of view here. Expand @Zippo Would you like to apply this to the FM-Arena Tactic Testing League? I can create the custom fixtures for you. I also want to add a few more things that could make it even more "full" in terms of options with more matches per run. So each run would include 1,100 matches so 3,300 for every 3 runs and so on.
Details below: 1. Increase AI teams from 5 to 11 and each team will have 1 of the 11 tactics present in the game. The reason is to have our tactics play against all tactics in the game regardless if one tactic is better than another by nature in the game. Like 4-2-3-1 compared to 4-3-2-1 for example.
2. Increase human controlled teams to 11 from 5 so each team can play 1 of the 11 AI teams always until the end of the season.
3. No Staff for any team - this prevents Team Talks / Shouts being given during the match. (EBFM)
4. Each team will play 100 matches per season. • ((22 teams x 100 matches) = 2.200 matches per run for all the teams together) / 2 = 1.100 matches (for human controlled teams)
Hey all, is it planed to have Atributes testing ?
dzek said: @Zippo can we have a finished game save of the testing league?
Probably, at some point later when I got time.
Falbravv said: Hey all, is it planed to have Atributes testing ?
Yes, of course it's in plans but I can say it won't happen during the beta that's for sure.
@Zippo 4k tests won’t happen during beta or?
opq said: @Zippo 4k tests won’t happen during beta or?
Hey,
We decided not to do 4k tests for the beta, there's really no point in that at this stage.
But 4k tests still might happens for some tactics during the beta, the purpose of that is our need to tune the configuration of our testing server and test some other things such the game RNG and other stuff.
@Zippo, I have noticed that you remove 442 formation for AI Managers and you insert another 443. Did you find that in this ME 442 is not powerful?
And “Team C” text on testing league table is not highlighted with blue color. Its still under control of you or you increase AI teams?
dzek said: @Zippo, I have noticed that you remove 442 formation for AI Managers and you insert another 443. Did you find that in this ME 442 is not powerful?
Yup, 4-4-2 under AI isn't impressive at all at this moment if the situation changes with patches we'll bring it back.
dzek said: And “Team C” text on testing league table is not highlighted with blue color. Its still under control of you or you increase AI teams?
It looks like a UI bug in the beta, multiple human managers not returning from Holiday simultaneously at the supposed time but as I said it's more a UI bug that doesn't have any impact on the result.
Is atributes going to be tested for 24 or its all same as previously?
White Europe said: Is atributes going to be tested for 24 or its all same as previously?
Yes, but we're considering to start attribute testing after the next M.E. update, which should happen soon according to SI.
Ah, sad news, it's the thing i'm waiting the most too
@Zippo is there a possibility of testing attributes by position? I dont know if it has been done before? I think keepers could be a position this would be extremelyy useful for.
Does anybody know when 4000 matches testing is going to start? not sure if this was asked before on here.
JGdaCaesar said: Does anybody know when 4000 matches testing is going to start? not sure if this was asked before on here.
I believe there is a small match engine update soon? And so it would be after that as should be final patch before winter update
keithb said: I believe there is a small match engine update soon? And so it would be after that as should be final patch before winter update
Oh that makes sense.
Hi everyone,
I would like to make an input here on a change to the Testing League that would probably help even more to minimize the "luck" factor.
There are 5 controlled teams and 5 AI teams. These teams all play each other and it is here that I would like to suggest since the league has this structure then it would be nice if each team played the same team in all fixtures until the end. I think that way the AI team would adapt even better to our tactics and I think that would reduce the "luck" factor.
*I'm leaving a picture of a reply given by Jack Joyce (SI dev) on how you can reduce statistical error(luck) within the game.
What do you think?
It would be nice to have @Zippo's point of view here.
Nice to see so much involvement in this forum lately @dzek ! keep going
Gracolas said: Nice to see so much involvement in this forum lately @dzek ! keep going
Thank you! We can all contribute!
dzek said: Hi everyone,
I would like to make an input here on a change to the Testing League that would probably help even more to minimize the "luck" factor.
There are 5 controlled teams and 5 AI teams. These teams all play each other and it is here that I would like to suggest since the league has this structure then it would be nice if each team played the same team in all fixtures until the end. I think that way the AI team would adapt even better to our tactics and I think that would reduce the "luck" factor.
*I'm leaving a picture of a reply given by Jack Joyce (SI dev) on how you can reduce statistical error(luck) within the game.
What do you think?
It would be nice to have @Zippo's point of view here.
Hi @dzek , In a nutshell, my understanding is that Teams A to E share identical attributes and current abilities (CA). Currently, lets say they are facing off against an AI team using a 4-2-3-1 formation. The question is: why might the statistical error vary if Team A faces the AI team 4-2-3-1 five times, as opposed to each of Teams A to E playing against the AI team 4-2-3-1 once?
sponsorkindest said: Hi @dzek , In a nutshell, my understanding is that Teams A to E share identical attributes and player abilities (PA). Currently, lets say they are facing off against an AI team using a 4-2-3-1 formation. The question is: why might the statistical error vary if Team A faces the AI team 4-2-3-1 five times, as opposed to each of Teams A to E playing against the AI team 4-2-3-1 once?
That's a good question to discuss as I'm not sure. The only thing that comes to my mind right now is that the AI managers have some tactical/non tactical attributes that are mostly the same across all 5 but there are differences in a few others that probably(again I'm not 100% sure) affect the way each team plays regardless if they use the same formations or even the same players attributes.
Apart from the tactical attributes, tendencies of AI Manager play a big role (during the game). That is, which instructions they will give in the game, when they will give them and if they will give them. Will they instruct their players to play out from defense? Will they work ball into the box? Will they push higher up at some point in the game? And much more..
dzek said: That's a good question to discuss as I'm not sure. The only thing that comes to my mind right now is that the AI managers have some tactical/non tactical attributes that are mostly the same across all 5 but there are differences in a few others that probably(again I'm not 100% sure) affect the way each team plays regardless if they use the same formations or even the same players attributes.
Apart from the tactical attributes, tendencies of AI Manager play a big role (during the game). That is, which instructions they will give in the game, when they will give them and if they will give them. Will they instruct their players to play out from defense? Will they work ball into the box? Will they push higher up at some point in the game? And much more..
I get your point but I don't have enuf knowledge about test league setup to say one way or the other. I am sure the moderators will know more and that's why I brought up that question.
dzek said: Hi everyone,
I would like to make an input here on a change to the Testing League that would probably help even more to minimize the "luck" factor.
There are 5 controlled teams and 5 AI teams. These teams all play each other and it is here that I would like to suggest since the league has this structure then it would be nice if each team played the same team in all fixtures until the end. I think that way the AI team would adapt even better to our tactics and I think that would reduce the "luck" factor.
*I'm leaving a picture of a reply given by Jack Joyce (SI dev) on how you can reduce statistical error(luck) within the game.
What do you think?
It would be nice to have @Zippo's point of view here.
@Zippo Would you like to apply this to the FM-Arena Tactic Testing League? I can create the custom fixtures for you. I also want to add a few more things that could make it even more "full" in terms of options with more matches per run. So each run would include 1,100 matches so 3,300 for every 3 runs and so on.
Details below:
1. Increase AI teams from 5 to 11 and each team will have 1 of the 11 tactics present in the game. The reason is to have our tactics play against all tactics in the game regardless if one tactic is better than another by nature in the game. Like 4-2-3-1 compared to 4-3-2-1 for example.
2. Increase human controlled teams to 11 from 5 so each team can play 1 of the 11 AI teams always until the end of the season.
3. No Staff for any team - this prevents Team Talks / Shouts being given during the match. (EBFM)
4. Each team will play 100 matches per season.
• ((22 teams x 100 matches) = 2.200 matches per run for all the teams together) / 2 = 1.100 matches (for human controlled teams)