I am testing different things to lower the intensity and concede less goals for ZaZ 2.0, without losing much performance. The plan is to use that when winning to rest players and hold the result, but without risking too much. I suppose the same instructions can also be applied to Phoenix. Right now, I can say changing mentality to cautious and asking goalkeeper to slow pace down are effective measures to reduce intensity and concede less while keeping similar chance to win the match. Lower tempo, lower pressing intensity and deeper pressing line all reduce your chance to win, increasing the likelihood to concede goals. I am currently testing waste time often.
Mark said: @ZaZ A few interesting observations as I drill into the data.
Sides that had a lower number of players in the Squad were either worse off or more variable in their results. Weymouth had the smallest squad with just 16 players when I stopped them recruiting. It would only take a suspension or injury to impact them.
The stats that they all seemed to have that were consistently high for their sides were (in order):
Determination Natural Fitness Work rate Stamina Team work Acceleration
Of these there were 3 that increased in importance as you moved up through the grades ie you needed these to increase as you move through the leagues:
Work rate Stamina Team work Expand
I was thinking of checking team report to see the attributes of each of the ten teams, then list how much % of league average for each attribute. Then I could check what attributes are more important for each strategy.
skywalk3r83 said: What do you do for individual training? Player roles or playing position? General training by assistant? Expand
I do player roles for individual. For general training, I always have Attacking Movement, Defending Movement and Teamwork before every match. Before decisions, I also add Set Pieces Delivery. Other than that, I add Match Practice and Team Bonding for cohesion and happiness, and Recovery to prevent injuries. All the rest is usually optional, like one physical if I have full weak of training, overall or outfield.
Eric said: Usually, when my team leads by 2 goals or more in a match then I go for a very defensive tactic that implies very low tempo and time wasting to save the conditions of my players and I'm sure if you don't do something like that then you'll burn the conditions of your players very fast. Expand
I don't like to do that because I hate when my real team does that, putting the result at risk. But it's probably not a bad choice in game.
Tactic source is going to the wrong page. It should be either the first page of Blue, or the specific topic for Blue 2.0 (linked in the first page of Blue).
I usually just lower the mentality to cautious when winning by two goals or more, as it reduces 1/3 of the likelihood to concede goals as well as reducing intensity. I save a tactic on cautious so it trains the mentality and get tactical familiarity. I've tried other stuff like wasting time, lower tempo, lower pressing intensity and such, but all of them make the tactic way worse, putting the result at risk.
For resting, I would rather rotate the squad, give one or two days of resting from training, or just do lighter training sessions during a tight schedule (like just team bonding, recovery and some match preparation, like attacking/defending movement).
It just did. Better than Blue 1.0, but still behind Phoenix. It performs better in normal conditions, when players are not frozen in perfect state and make more mistakes. It also got the highest goal difference, which is always a plus.
Lapidus said: I think if you aren't ready to hear any criticism or a different opinion then you shouldn't post anything on the internet
Constructive criticism is a good thing, it helps evaluating and improving.
Expand
No point arguing like this.
healmuth said: In my testing, players, their stats, moral and condition is frozen. I also test 80 matches in a custom league per run. The basis of the league is provided by FMB and I’ve modified it further. Is it perfect? No, there’s no perfect testing but it certainly is more accurate than load tactic and press holiday. Expand
You are right if you mean loading a tactic once then pressing play. If you do the same thing for 1800 matches, then you start extracting useful information from the randomness. Either people like it or not, random facts have an impact on the tactics. Tests must aim to remove outliers, not to remove randomness. Freezing stats is one way to remove the outliers, but not the only way.
P.S.: Not saying holiday is the best way to test tactics, only that it is a different way, with different goals and outcomes.
Lapidus said: Pal, you're exaggerating here. I really don't see anyone tries "demotivate" Mark or saying anything negative about your Blue 2.0.
Guys were just pointing the weak and strong sides of different testing methods and that's all. I'm sure that such discussion can help people to improve their testing methods. Expand
I think they tried to discredit the results. One guy said "that proves you can't just plug and play and push holiday button", which is clearly a way of minimizing his methodology. The other guy said the difference of 4% was too small to mean anything and it was just RNG factors, while another was saying you can't say there is a clear winner, when there is clear statistical relevance to the results.
What I mean is that more people appeared to say something negative than to compliment the effort.
It's funny how it became something about Blue 2.0, when we should be appreciating all effort he put there. I think it's wonderful that people like you, @healmuth or him, among others, test tactics with different approaches. It really helps people improve the tactics. Today Blue 2.0 was on top, but maybe tomorrow there will be a better tactic. We need to be grateful that people are contributing to the community, not demotivate them by doubting their methods or results.
Droid said: For me that proves that you can't test tactics by just plugging them and pushing "On Holiday" button. I understand why people tend to test tactics this way because it the simplest way to test, it doesn't require any skill or any efforts but it also gives an inaccurate result even after 1600 matches because the impact of random factors is very significant in this case, any injuries of the key players in your team or any injuries of the key players in the opposition team have a great impact on the result and they can turn everything apart.
Also, the assistant manager is responsible for managing the morale, picking the starting eleven for the tactic but it's been proven many times that even the best assistant manager is very poor at piking the team for many different tactics and he isn't capable managing any morale issue so it's like play roulette. Expand
I think you are trying to compare apples to oranges. Holiday mode is a valid way to compare tactics when a significant number of matches/seasons is run. In that case, it tests how much a tactic is tolerant to player mistakes (caused by player out of position due to injuries, physical condition, morale or pressure, for example). It's not better or worse than freezing attributes, just a different methodology. Random factors do happen during normal gameplay, and the high number of matches make them behave as a statistical variance instead of as an outlier.
On the other hand, tests like fm-arena and fm-base are more suitable if you want to know how a tactic would behave when your squad is in perfect condition, which can happen often in practice if you are good at managing the squad.
Rince said: Nobody's minimizing anything. I see people just express their opinions on the subject, I think a quite normal thing... Expand
I am just saying that a difference of 4% (1.905/1.832) in 1800 matches is statistically relevant. We are talking about achieving 3-4 more points on average in every season, which can have even more impact in cups. For the conditions he tested, Blue 2.0 can be considered a clear winner. That doesn't mean it will be better in any conditions. Maybe Phoenix or Voila are better with stronger teams, or freezing stats, but that wasn't the objective of his tests.
For me, it would be really interesting if he had also tested Blue 1.0, or also tested on Premier League, but I can't ask him to do even more work after that.
Poacher said: IMHO, the changes between Blue 2.0 and Blue 1.0 would make a 1% difference at the most Expand
Maybe you are right. I am just saying that everyone that tested both prefer 2.0, since it gets better results. But it could very well perform worse in some test, for specific conditions. I would also say it's very hard to optimize a tactic when it's too close to the top, so you can't just minimize a 1% improvement as if it's something trivial to achieve.
P.S.: I don't understand why people are minimizing his results. They don't change any of the results from fm-arena, as they are not comparable. They use different methodologies and are meant for different purposes.
Poacher said: @Mark, if you'd ask me then I wouldn't say there's a clear winner
ZaZ Blue 2.0 tactic: Avg. points per game = 1.916
Phoenix 3.0 tactic: Avg. points per game = 1.849
The difference in PPG is less than 3.5%
I guess you tested without eliminating RNG factors like in the fm-arena testing, I mean freezing the morale and conditions and so on. In the fm-arena testing Phoenix tactic has 2.114 PPG and Blue tactic has 2.092 PPG so in fm-arena testing the difference in PPG is less than 1.1% and I'm sure if you were eliminating RNG then you also had like a 1.1% difference in PPG and you really can't say that there's a clear winner when the difference is less than 1.1% Expand
Blue 2.0 wasn't selected for testing in fm-arena. It doesn't appear in the table.
Also, the difference of 0.08 points per match is basically the difference between a 7.0 and a 6.8 tactic here, so it shouldn't be considered negligible.
Mark said: I dont disagree with your points about points per match being the most important stat @Nikko and @Milakus. This is why it is the main stat I have used in the summary results. However, I think winning the league means you had the best side that year in that league, so I think it is a supplementary indicator.
In terms of the variables and random factors, I think that the extent of the testing would reduce these somewhat. The higher the number of games played the lower the random impacts will be. For example. 1800 games testing should provide a much better picture than say 150 games.
I love FM Arena. The concept is brilliant and value to the FM community is huge. However I have often thought that by using teams with talent at the levels of EPL and Champions leagues, the testing doesnt always assist us lower league players. And I certainly like that there are differing views put forward on the forums.
Having said all that, I did this for fun and dont claim it to be definitive. Take what you want from what I have produced. I play the lowest tier in each save and will certainly be favoring ZaZ Blue 2.0 until the next major update from SI. Expand
There are also some factors that normal tests from here or fm-base won't account, like pressure added by a title run, injury of key players leaving no options for positions, as well as exhaustion for tight schedule. Those factors induce player mistakes, which can show which tactics are more tolerant to errors. I believe it's a very realistic scenario of what people encounter in practice, instead of always having players in the best shape to minimize their variation. It's just normal for players to oscillate during a season, and the high number of tests make up for the unwanted randomness.
I don't think he claims his results to be absolute, it's just a different approach, which might lead to different results. He is just considering the tactics in less ideal conditions.
Milakus said: Only the points per match are important and the final standing means nothing because the standing depends on the performance of the AI teams which can greatly vary and depends on many random factors so in one test some of AI managers can do very good and in other test they might do poor. Expand
The comparison was made using points per match. The number of titles was just an interesting fact. I believe extra data is always useful.
But when you look at your points per match table I don't think you can determine which tactic is a clear winner because there's like a 3.5% between the tactics and probably, if you remove all random factors like fm-arena does in its testing then you'll get less than a 1.5% difference between the tactic when you look at the points per watch and that's why all three tactics are rated 7.0 because a 1.5% difference is almost nothing Expand
He tested Blue 2.0, which is better than original Blue. That tactic didn't get selected for tests here.
@Mark , can you make a compilation of each teams average attributes, from team report? It would be interesting to know the first touch, determination and other attributes to find some correlation.
Mark said: OK, the results are in. Just under 1800 games for each tactic.
Points per match by league using top ranked and lowest ranked sides:
I think ZaZ Blue 2.0 is a clear winner for lower league management.
Points per match by club:
Weymouth are a bit of an anomaly. I feel inclined to take the challenge and try and win something with them on my next save using ZaZ Blue 2.0.
ZaZ Blue 2.0 Summary results and individual test results:
Phoenix v3.0 Summary results and individual test results:
Viola V1 Summary results and individual test results:
Expand
Probably Weymouth is much worse compared to other teams in the League, like they have stats of non-league competing with professionals. I will also try to play them since it seems to be hard mode.
About the results, first I want to thank you for the effort. It's always nice to have people testing tactics, but you brought it to another level with such a thoughtful report. The methodology was also sound, which makes me believe the results are as close as possible to what would happen to the average player. I am also happy to see Blue 2.0 winning, obviously, since it won't be tested here and has a high chance to not be tested on fm-base anytime soon. I kinda needed to know how it performs, because I always felt the better play style could be just because higher tempo is more enjoyable to watch.
Other than Weymouth, the worst position for Blue 2.0 was an 8th place. It also won che championship 18 out of 20 times, getting in promotion range 28 out of 40 times. For comparison, Phoenix won only 11 times and got in promotion range 26 times, very similar to Viola. Blue 1.0 would also fall in that range, probably, since the three have very similar performance.
Sides that had a lower number of players in the Squad were either worse off or more variable in their results. Weymouth had the smallest squad with just 16 players when I stopped them recruiting. It would only take a suspension or injury to impact them.
The stats that they all seemed to have that were consistently high for their sides were (in order):
Determination
Natural Fitness
Work rate
Stamina
Team work
Acceleration
Of these there were 3 that increased in importance as you moved up through the grades ie you needed these to increase as you move through the leagues:
Work rate
Stamina
Team work
I was thinking of checking team report to see the attributes of each of the ten teams, then list how much % of league average for each attribute. Then I could check what attributes are more important for each strategy.
Player roles or playing position?
General training by assistant?
I do player roles for individual. For general training, I always have Attacking Movement, Defending Movement and Teamwork before every match. Before decisions, I also add Set Pieces Delivery. Other than that, I add Match Practice and Team Bonding for cohesion and happiness, and Recovery to prevent injuries. All the rest is usually optional, like one physical if I have full weak of training, overall or outfield.
I don't like to do that because I hate when my real team does that, putting the result at risk. But it's probably not a bad choice in game.
For resting, I would rather rotate the squad, give one or two days of resting from training, or just do lighter training sessions during a tight schedule (like just team bonding, recovery and some match preparation, like attacking/defending movement).
It just did. Better than Blue 1.0, but still behind Phoenix. It performs better in normal conditions, when players are not frozen in perfect state and make more mistakes. It also got the highest goal difference, which is always a plus.
Constructive criticism is a good thing, it helps evaluating and improving.
No point arguing like this.
healmuth said: In my testing, players, their stats, moral and condition is frozen. I also test 80 matches in a custom league per run. The basis of the league is provided by FMB and I’ve modified it further. Is it perfect? No, there’s no perfect testing but it certainly is more accurate than load tactic and press holiday.
You are right if you mean loading a tactic once then pressing play. If you do the same thing for 1800 matches, then you start extracting useful information from the randomness. Either people like it or not, random facts have an impact on the tactics. Tests must aim to remove outliers, not to remove randomness. Freezing stats is one way to remove the outliers, but not the only way.
P.S.: Not saying holiday is the best way to test tactics, only that it is a different way, with different goals and outcomes.
Guys were just pointing the weak and strong sides of different testing methods and that's all. I'm sure that such discussion can help people to improve their testing methods.
I think they tried to discredit the results. One guy said "that proves you can't just plug and play and push holiday button", which is clearly a way of minimizing his methodology. The other guy said the difference of 4% was too small to mean anything and it was just RNG factors, while another was saying you can't say there is a clear winner, when there is clear statistical relevance to the results.
What I mean is that more people appeared to say something negative than to compliment the effort.
It's funny how it became something about Blue 2.0, when we should be appreciating all effort he put there. I think it's wonderful that people like you, @healmuth or him, among others, test tactics with different approaches. It really helps people improve the tactics. Today Blue 2.0 was on top, but maybe tomorrow there will be a better tactic. We need to be grateful that people are contributing to the community, not demotivate them by doubting their methods or results.
Also, the assistant manager is responsible for managing the morale, picking the starting eleven for the tactic but it's been proven many times that even the best assistant manager is very poor at piking the team for many different tactics and he isn't capable managing any morale issue so it's like play roulette.
I think you are trying to compare apples to oranges. Holiday mode is a valid way to compare tactics when a significant number of matches/seasons is run. In that case, it tests how much a tactic is tolerant to player mistakes (caused by player out of position due to injuries, physical condition, morale or pressure, for example). It's not better or worse than freezing attributes, just a different methodology. Random factors do happen during normal gameplay, and the high number of matches make them behave as a statistical variance instead of as an outlier.
On the other hand, tests like fm-arena and fm-base are more suitable if you want to know how a tactic would behave when your squad is in perfect condition, which can happen often in practice if you are good at managing the squad.
In short, they test different things.
I am just saying that a difference of 4% (1.905/1.832) in 1800 matches is statistically relevant. We are talking about achieving 3-4 more points on average in every season, which can have even more impact in cups. For the conditions he tested, Blue 2.0 can be considered a clear winner. That doesn't mean it will be better in any conditions. Maybe Phoenix or Voila are better with stronger teams, or freezing stats, but that wasn't the objective of his tests.
For me, it would be really interesting if he had also tested Blue 1.0, or also tested on Premier League, but I can't ask him to do even more work after that.
Maybe you are right. I am just saying that everyone that tested both prefer 2.0, since it gets better results. But it could very well perform worse in some test, for specific conditions. I would also say it's very hard to optimize a tactic when it's too close to the top, so you can't just minimize a 1% improvement as if it's something trivial to achieve.
P.S.: I don't understand why people are minimizing his results. They don't change any of the results from fm-arena, as they are not comparable. They use different methodologies and are meant for different purposes.
ZaZ Blue 2.0 tactic: Avg. points per game = 1.916
Phoenix 3.0 tactic: Avg. points per game = 1.849
The difference in PPG is less than 3.5%
I guess you tested without eliminating RNG factors like in the fm-arena testing, I mean freezing the morale and conditions and so on. In the fm-arena testing Phoenix tactic has 2.114 PPG and Blue tactic has 2.092 PPG so in fm-arena testing the difference in PPG is less than 1.1% and I'm sure if you were eliminating RNG then you also had like a 1.1% difference in PPG and you really can't say that there's a clear winner when the difference is less than 1.1%
Blue 2.0 wasn't selected for testing in fm-arena. It doesn't appear in the table.
Also, the difference of 0.08 points per match is basically the difference between a 7.0 and a 6.8 tactic here, so it shouldn't be considered negligible.
In terms of the variables and random factors, I think that the extent of the testing would reduce these somewhat. The higher the number of games played the lower the random impacts will be. For example. 1800 games testing should provide a much better picture than say 150 games.
I love FM Arena. The concept is brilliant and value to the FM community is huge. However I have often thought that by using teams with talent at the levels of EPL and Champions leagues, the testing doesnt always assist us lower league players. And I certainly like that there are differing views put forward on the forums.
Having said all that, I did this for fun and dont claim it to be definitive. Take what you want from what I have produced. I play the lowest tier in each save and will certainly be favoring ZaZ Blue 2.0 until the next major update from SI.
There are also some factors that normal tests from here or fm-base won't account, like pressure added by a title run, injury of key players leaving no options for positions, as well as exhaustion for tight schedule. Those factors induce player mistakes, which can show which tactics are more tolerant to errors. I believe it's a very realistic scenario of what people encounter in practice, instead of always having players in the best shape to minimize their variation. It's just normal for players to oscillate during a season, and the high number of tests make up for the unwanted randomness.
I don't think he claims his results to be absolute, it's just a different approach, which might lead to different results. He is just considering the tactics in less ideal conditions.
The comparison was made using points per match. The number of titles was just an interesting fact. I believe extra data is always useful.
But when you look at your points per match table I don't think you can determine which tactic is a clear winner because there's like a 3.5% between the tactics and probably, if you remove all random factors like fm-arena does in its testing then you'll get less than a 1.5% difference between the tactic when you look at the points per watch and that's why all three tactics are rated 7.0 because a 1.5% difference is almost nothing
He tested Blue 2.0, which is better than original Blue. That tactic didn't get selected for tests here.
@Mark , can you make a compilation of each teams average attributes, from team report? It would be interesting to know the first touch, determination and other attributes to find some correlation.
Points per match by league using top ranked and lowest ranked sides:
I think ZaZ Blue 2.0 is a clear winner for lower league management.
Points per match by club:
Weymouth are a bit of an anomaly. I feel inclined to take the challenge and try and win something with them on my next save using ZaZ Blue 2.0.
ZaZ Blue 2.0 Summary results and individual test results:
Phoenix v3.0 Summary results and individual test results:
Viola V1 Summary results and individual test results:
Probably Weymouth is much worse compared to other teams in the League, like they have stats of non-league competing with professionals. I will also try to play them since it seems to be hard mode.
About the results, first I want to thank you for the effort. It's always nice to have people testing tactics, but you brought it to another level with such a thoughtful report. The methodology was also sound, which makes me believe the results are as close as possible to what would happen to the average player. I am also happy to see Blue 2.0 winning, obviously, since it won't be tested here and has a high chance to not be tested on fm-base anytime soon. I kinda needed to know how it performs, because I always felt the better play style could be just because higher tempo is more enjoyable to watch.
Other than Weymouth, the worst position for Blue 2.0 was an 8th place. It also won che championship 18 out of 20 times, getting in promotion range 28 out of 40 times. For comparison, Phoenix won only 11 times and got in promotion range 26 times, very similar to Viola. Blue 1.0 would also fall in that range, probably, since the three have very similar performance.
Good job there!
My team won everything. Thank you.
You're welcome! Haaland must have scored like there is no tomorrow. He is really great to be retrained as SS.
P.S.: Just saw that he was the top scorer of all competitions. 63 goals in a year is not bad, it's more than Messi and CR at the peak.