Project M Recommended Ruleset

[DMG]

Set counts are not 1/1 because part of testing who is the better player is testing who is better for a period of time longer than 1 game. Otherwise we could just do friendlies, chip in $5, round robin 1/1 each other and call it a day for whoever has the best record. Skill is more accurately defined and acknowledged over multiple games.


Doing CP's first does not lead to a stable series of MU's, it leads to more RPS CP scenarios. If starters are believed to give the average cast member the best hope for finding a neutral playing field against most other cast members, it does not make sense to push this off towards the end of a set. Finding out who is the better player will best be served by not encouraging more CPing than the current format imo: if you were to pose the question of playing exclusively on starters vs exclusively on CP's, starters would likely win for finding more skilled and less imbalanced gameplay from the cast.


It also breaks from the traditional mold by not allowing a loser to retry CP. If a person fails his CP in this format at any point except for Game 1 sorta, he's put into more pressure that before since he lost the right to stage pick. A person who won the coin flip to CP first, in a 3/5 set, can have the opportunity to pick twice before his opponent has won on any stage. When this double pick situation is ALSO tied to something essentially like a coin flip or RPS at the beginning, it doesn't portray a more fair start than doing stage striking for Game 1. "Sorry I lost the coin flip, guess you get to pick the stage first during the most crucial part of the set (You get to pick stage first + this first CP is done prior to any losses in the set so it's not really an earned advantage).


If we have an issue with current format, then regulate the importance and impact of CP stages by limiting them, changing them, banning them, whatever. Changing the format to try and minimize the impact of CP stages, by that suggestion, impacts the relative neutrality of Game 1 and tries to push that neutrality to the very last possible moment.


TL:DR

Game 1 in the current format undergoes a much more fair process: nobody gets a direct stage pick, nobody gets to choose from CP stages, both players basically have equal weight on what happens. This is one of the best things to happen to Smash, and changing it to fix other potential issues in the stage selection process seems like the wrong way about fixing other problems.

 

[Kneato]

I know no one likes change so most would want to find issues with the system I drew up.

I KNOW it makes sets finish faster, and I think it makes set outcomes more reflective of player skill as they shold be. I could be incorrect on that part. But arguing it with you guys really won't prove anything either way.

So I'm just going to use it for the next few monthlies, see how it turns out and report here any issues or successes it had.

 

[Kneato]

Set counts are not 1/1 because part of testing who is the better player is testing who is better for a period of time longer than 1 game. Otherwise we could just do friendlies, chip in $5, round robin 1/1 each other and call it a day for whoever has the best record. Skill is more accurately defined and acknowledged over multiple games.

That was my point. The best player doesn't always win the first game. In a perfect world, sets would be Bo99 to truly determine who is the better player, but obviously Bo99 is not feasible.

And that's the issue I have, the better player may not win the neutral match, especially if the skill level is close, but the first match often determines the set outcome. Thats the problem to address, and I believe putting the neutral at the end makes it more likely that the better player wins it.

Again, obviously no one wants to change the current system without a good reason so I'll see how it works in practice.

 

[DMG]

Your biggest hurdle is convincing people that they should flip a coin and have someone start with CP advantage for Game 1 instead of mutual striking. Might have some merit for 2/3, but 3/5 I'm not sure it's going to be enjoyed. Only takes a couple of people to think "I'm getting counter-picked Game 1 because I lost a coin flip / RPS when I could instead be striking" to spoil the rest of the idea.

Idk how you would sell it but good luck. Defining better player is pretty subjective though. Better player losing Game 1 on a starter for example: given the fact that he indeed lost, is he actually the better player? If the better player loses during the CP phases in the suggested alternative format, maybe having CP only was imbalancing enough for him to lose and therefore having starters for Game 5 hurt his chances more than having it Game 1? You could spin it or think about it in many different ways.

 

[skellitorman]

I know no one likes change so most would want to find issues with the system I drew up.
I KNOW it makes sets finish faster, and I think it makes set outcomes more reflective of player skill as they shold be. I could be incorrect on that part. But arguing it with you guys really won't prove anything either way.

The problem here is not that people dislikes change. The problem here (aside from the flawed initial assessment of the “classical” method) is that the method used to come up with your data is flawed, and the conclusions based on such data are also flawed.

Then you ignore or fail to properly address relevant information that poses a problem to your proposed idea. For example, 4tlas asked a very relevant and important question “What are you going to do about character counterpicks?” which you have yet to respond to.

----------------------------------------------------------------------------------

Just for clarification on the flawed data: You used 100,000 set simulations (as in, they are data points that are not real) to come up with your data which you are using to proclaim facts. In the “chart” that you made you also have people winning above 50% (total) which logically makes no sense since for every winner you have, there is a loser (meaning that there should always be a 1:1 ratio).

Most of the arguments that you have made against the classic method can also be used against your proposed method because both methods have 1 starter stage and 2 counterpick stages (bo3). However the problems that DMG has stated also exist with your method, and you haven’t addressed them within your arguments.

If the better player loses the first game on a starter stage (current method), he does not suddenly become more inclined to win if the first game was on an opponent's counter pick stage (your method). The better player is also more likely to go to a game 3 when he has to play an opponent’s counterpick stage first which only occurs in your proposed method.

 

[Narpas_sword]

Momentum is a thing too.

Games in a set can't be thought of in a vacuum. If one player starts out really strong it can affect the other games.

Randomly giving one player a CP start is going to create this momentum.

 

[4tlas]

Momentum is a thing too.

Games in a set can't be thought of in a vacuum. If one player starts out really strong it can affect the other games.

Randomly giving one player a CP start is going to create this momentum.

This is more of what I'm talking about with counterpicks letting one player start with an advantage. I'm not even sure which player it would be, as maybe the first player feels pressured to win their counterpick because they won't get another chance but they have no info yet, or maybe the second player feels pressured for the rest of the set because they lost cleanly on their opponent's counterpick before they had any info.

 

[DMG]

First player to CP has best chance for advantage overall in practical terms, although this advantage becomes muted the more evenly matched and the more character variety the two players have. A weaker player going up against a stronger player can even the odds easier if he gets to CP first and win, even if it's a jank surprise win: in regular format he's more likely to lose to better player on a starter for Game 1 than on his own CP. Makes it easier for the worse player to create 1-2 set counts. It's also more likely for the player skill gap to grow or become more apparent if the better player gets a game or two of knowledge into your habits / etc, so avoiding that by getting to CP early in the process helps retain your CP advantage instead of partially or fully losing it because the better player has adapted to a certain point.

 

[Kneato]

The problem here is not that people dislikes change. The problem here (aside from the flawed initial assessment of the “classical” method) is that the method used to come up with your data is flawed, and the conclusions based on such data are also flawed.

Then you ignore or fail to properly address relevant information that poses a problem to your proposed idea. For example, 4tlas asked a very relevant and important question “What are you going to do about character counterpicks?” which you have yet to respond to.

----------------------------------------------------------------------------------

Just for clarification on the flawed data: You used 100,000 set simulations (as in, they are data points that are not real) to come up with your data which you are using to proclaim facts. In the “chart” that you made you also have people winning above 50% (total) which logically makes no sense since for every winner you have, there is a loser (meaning that there should always be a 1:1 ratio).

Most of the arguments that you have made against the classic method can also be used against your proposed method because both methods have 1 starter stage and 2 counterpick stages (bo3). However the problems that DMG has stated also exist with your method, and you haven’t addressed them within your arguments.

If the better player loses the first game on a starter stage (current method), he does not suddenly become more inclined to win if the first game was on an opponent's counter pick stage (your method). The better player is also more likely to go to a game 3 when he has to play an opponent’s counterpick stage first which only occurs in your proposed method.

You aren't reading the data correctly. The data gives theoretical chance of each outcome for the better player. The three columns are for three levels of difference in skill. The bigger the difference, the higher % chance of the better player winning. As stated on the page, the % chances are for the better of the two players.

As for not addressing certain arguments, it's rather hard to respond to everything everyone says when it's one of me trying to explain and about 5 people now all with different perceived issues.

For the character counterpicks, they would be done similarly to how they already are. The player counterpicking the stage also counterpicks character. Final match would be double blind character picks.

Once again, this should make it that each player has a good chance of winning their counterpick if they are close in skill. If there is a large difference in skill, the better player has a good chance of ending the match early by winning on the opponents counterpick.

Momentum is a thing too.

Games in a set can't be thought of in a vacuum. If one player starts out really strong it can affect the other games.

Randomly giving one player a CP start is going to create this momentum.

This sounds to me like the most relevant issue so far

 

[skellitorman]

You aren't reading the data correctly. The data gives theoretical chance of each outcome for the better player. The three columns are for three levels of difference in skill. The bigger the difference, the higher % chance of the better player winning. As stated on the page, the % chances are for the better of the two players.

"Each method for both types of sets (classic and your proposed set) was simulated over 100,000 sets in order to approximate the % chance of each outcome under multiple cases of skill difference between the two fighting players. These outcomes apply to the player who has the higher skill in each case."

All I have for determining the method used to come up with your data is the above, which is very lacking in details. I am unsure how the simulations were used to come up with the percent chances and would like the data on this so I can see exactly where the problems in the logic originate from. Now that I am reading your chart correctly, I will properly address certain problems that I can at this point.

As for not addressing certain arguments, it's rather hard to respond to everything everyone says when it's one of me trying to explain and about 5 people now all with different perceived issues.

You have failed to refute any of the arguments because the reasoning that you are using is weak (weaker than counterarguments and also because of being based on fabricated data).

Against one of the arguments DMG gave, you responded with, “Not true. I know I'm long winded and I post a lot to read through and I'm sorry for that, but I already posted statistics proving that theoretical win rates are unaltered with the new method.”

What DMG said is in fact very true though, and your fabricated data doesn’t even address his points at all. For reference, the particular argument was:

“Your setup is flawed because it hinges on the presumption that sets always go on to the final game. If a 3/5 set ends 3-0 or 3-1, every game played would likely be a CP. (If I play Lunchables on 4 CP's in a row, our set will be much more lopsided in gameplay extremes, than if we had 1 starter + 3 CP games)

Putting the starter tiebreaker at the end only gives balance when the players will reach final game. A lot of sets would not reach final game, and therefore you would be crafting a more imbalanced setup for anyone that doesn't make it to final game.”

To further explain one of the key points here: If a player is playing a Bo5 (in your proposed idea) and it goes P1 CP, P2 CP, P2 CP, P1 CP, NP, then if P2 wins the first three games he would have won the set having played 2 of his CPs which makes no sense in a competitive setting (winning a game while having twice the amount of counterpick matches).

Your fabricated data doesn’t refute this at all.

The whole argument that you are making (including the data) is based on a lot of assumptions. You are assuming that a player skill advantage is static (which it isn’t. Humans can get nervous for example in game 1 and be ok for the rest of the set) and will always yield an equal total % of games won in both methods because that’s what it means to have that level of advantage (65:35 for example). You are assuming that playing a “neutral” stage at the end of the match is the same as playing it in the beginning which it isn't.

Aside from arguments already given by others, here is another important one:

The info gained from the opponent necessary to make the best counterpick is not available if you have to counterpick game 1. The player could end up not playing the best counterpick he could have, had he had the information that you get from game 1 (this means that the second player counterpicking will have an unfair counterpicking advantage based on information). Suppose player 1 who “wasted” his counterpick barely wins game 1, but then loses game 2. Now he could lose game 3 based on picking a bad character (due to double blind pick) having been disadvantaged from game 1 throughout the entire set.

Furthermore, you only give the fabricated stats of the better player when he counterpicks first in your proposed method. The stats of him counterpicking last completely changes your percentages, creating the most likely situation for a game 3 (based on your logic), which in turn make games last longer.

The data points itself are also assumed, which makes the actual results lack the validity necessary to back up your argument.

-----------------------------------------------------------------------------------------------

Needless to say, none of your proposed advantages (that are relevant) actually exist over the current method.

It’s not faster (in fact, can be significantly slower as explained previously), nor does it ensure that the better player wins (in fact, can make results more skewed as explained previously). Having both players play on their counterpick in a Bo3 is of no significant value over the alternative. The premise of the Bo5 set being more emphasized on player’s skill due to playing on different stages also does not exist over the classic method. Even if it did, such emphasis on player skill would not be significantly measured differently due to such a factor.

------------------------------------------------------------------------------------------------

So instead of refuting the key points you have said this, “But arguing it with you guys really won't prove anything either way.” Rather than addressing the very relevant issues accordingly so that the argument can have some actual support, you dismiss everything.

EDIT: I am not trying to be sassy. Unfortunately, I just naturally speak this way. I am trying to be as clear as possible to avoid any misunderstanding. I apologize if it seems like I am attacking you.

 

[4tlas]

I would like your post skellitorman, but there is too much sass for my taste. Thus I will instead post and say you brought up excellent points and explained them very well.

 

[Kneato]

"Each method for both types of sets (classic and your proposed set) was simulated over 100,000 sets in order to approximate the % chance of each outcome under multiple cases of skill difference between the two fighting players. These outcomes apply to the player who has the higher skill in each case."

All I have for determining the method used to come up with your data is the above, which is very lacking in details. I am unsure how the simulations were used to come up with the percent chances and would like the data on this so I can see exactly where the problems in the logic originate from. Now that I am reading your chart correctly, I will properly address certain problems that I can at this point.



You have failed to refute any of the arguments because the reasoning that you are using is weak (weaker than counterarguments and also because of being based on fabricated data).

Against one of the arguments DMG gave, you responded with, “Not true. I know I'm long winded and I post a lot to read through and I'm sorry for that, but I already posted statistics proving that theoretical win rates are unaltered with the new method.”

What DMG said is in fact very true though, and your fabricated data doesn’t even address his points at all. For reference, the particular argument was:

“Your setup is flawed because it hinges on the presumption that sets always go on to the final game. If a 3/5 set ends 3-0 or 3-1, every game played would likely be a CP. (If I play Lunchables on 4 CP's in a row, our set will be much more lopsided in gameplay extremes, than if we had 1 starter + 3 CP games)

Putting the starter tiebreaker at the end only gives balance when the players will reach final game. A lot of sets would not reach final game, and therefore you would be crafting a more imbalanced setup for anyone that doesn't make it to final game.”

To further explain one of the key points here: If a player is playing a Bo5 (in your proposed idea) and it goes P1 CP, P2 CP, P2 CP, P1 CP, NP, then if P2 wins the first three games he would have won the set having played 2 of his CPs which makes no sense in a competitive setting (winning a game while having twice the amount of counterpick matches).

Your fabricated data doesn’t refute this at all.

The whole argument that you are making (including the data) is based on a lot of assumptions. You are assuming that a player skill advantage is static (which it isn’t. Humans can get nervous for example in game 1 and be ok for the rest of the set) and will always yield an equal total % of games won in both methods because that’s what it means to have that level of advantage (65:35 for example). You are assuming that playing a “neutral” stage at the end of the match is the same as playing it in the beginning which it isn't.

Aside from arguments already given by others, here is another important one:

The info gained from the opponent necessary to make the best counterpick is not available if you have to counterpick game 1. The player could end up not playing the best counterpick he could have, had he had the information that you get from game 1 (this means that the second player counterpicking will have an unfair counterpicking advantage based on information). Suppose player 1 who “wasted” his counterpick barely wins game 1, but then loses game 2. Now he could lose game 3 based on picking a bad character (due to double blind pick) having been disadvantaged from game 1 throughout the entire set.

Furthermore, you only give the fabricated stats of the better player when he counterpicks first in your proposed method. The stats of him counterpicking last completely changes your percentages, creating the most likely situation for a game 3 (based on your logic), which in turn make games last longer.

The data points itself are also assumed, which makes the actual results lack the validity necessary to back up your argument.

-----------------------------------------------------------------------------------------------

Needless to say, none of your proposed advantages (that are relevant) actually exist over the current method.

It’s not faster (in fact, can be significantly slower as explained previously), nor does it ensure that the better player wins (in fact, can make results more skewed as explained previously). Having both players play on their counterpick in a Bo3 is of no significant value over the alternative. The premise of the Bo5 set being more emphasized on player’s skill due to playing on different stages also does not exist over the classic method. Even if it did, such emphasis on player skill would not be significantly measured differently due to such a factor.

------------------------------------------------------------------------------------------------

So instead of refuting the key points you have said this, “But arguing it with you guys really won't prove anything either way.” Rather than addressing the very relevant issues accordingly so that the argument can have some actual support, you dismiss everything.

EDIT: I am not trying to be sassy. Unfortunately, I just naturally speak this way. I am trying to be as clear as possible to avoid any misunderstanding. I apologize if it seems like I am attacking you.

I truly don't understand the issue you have with my data. You sound like you think I made it up.

I wrote a script that simulates matches being played out. I linked the script next to the data so you can look at it yourself and run it to see the results. You probably didn't see it since it seems you dismissed the sheet before actually looking at it.

Obviously people don't have a static value that represents how good they are, nor can we quantify the advantage or disadvantage a counterpick gives. But in order to discuss rulesets and outcomes, we have to make assumptions best as we can.

I made some fair assumptions in order to run a simulation, and the outcome mirrors real life outcomes.

Here is the script for simulating Bo3: https://jsfiddle.net/n6LtvutL/5/ Once again this was linked in the data sheet.

If you compare the simulated data to a real life set analysis
(https://www.reddit.com/r/smashbros/comments/3hs3kd/a_statistical_look_at_best_of_3s_and_best_of_5s/)
the results are similar to real life outcomes. They found that winning the first match on average leads a player to win the set 82% of the time in Bo3 and 89% of the time in Bo5. My simulation data found that the theoretical set win rate was 88% for Bo3 and 85% for Bo5 (the inverse of Lost 2nd + 3rd for Bo3 and the inverse of the sum of Lost 2nd + 3rd + 4th, Lost 2nd + 3rd + 5th, Lost 2nd + 4th + 5th, and Lost 3rd + 4th + 5th for Bo5).
Aside from that though, I agree the fact that the first player's counterpick is weakened because he doesn't have a match worth of data to base his CP on.

My reason for the KTLT method was thus. Both players are given a guaranteed counterpick. These counterpicks were supposedly equal strength. They were both supposed to be "safe" matches where each person counterpicking is expected to win the match if the skill gap between players is low. Then it would go to a neutral match where both players use the data from the last two matches to try to win in a neutral setting.

If the skill gap is high, better player may beat worse player on his counterpick and end the set early.

The issue I had with the classic method is that the first match is played mostly blind. It could be decided strictly because one player has more experience with a certain matchup. There is no adaption taken into account. Yet both my simulation and actual data from real matches have shown that the winner of the first match has over an 80% chance of winning the set.

We play sets rather than single matches for the reason that we don't believe that a single match is truly reflective of the skill difference between tow players. But in the current system, the first match ends up having a large impact on the outcome of the set anyways. This is what I was trying to change. With my system, the first two games between two players near the same level of skill should end up going 1-1 and the final neutral game should decide the outcome, after equal amounts of data have been collected by both players.

But if in practice, player counterpick advantage ends up being overly uneven, this defeats the purpose of the new system.

But I stand by the fact that in theory, the new system runs faster. But perhaps not better.

 

[skellitorman]

If you compare the simulated data to a real life set analysis
(https://www.reddit.com/r/smashbros/comments/3hs3kd/a_statistical_look_at_best_of_3s_and_best_of_5s/)
the results are similar to real life outcomes. They found that winning the first match on average leads a player to win the set 82% of the time in Bo3 and 89% of the time in Bo5. My simulation data found that the theoretical set win rate was 88% for Bo3 and 85% for Bo5 (the inverse of Lost 2nd + 3rd for Bo3 and the inverse of the sum of Lost 2nd + 3rd + 4th, Lost 2nd + 3rd + 5th, Lost 2nd + 4th + 5th, and Lost 3rd + 4th + 5th for Bo5).

Your results are not comparable at all. Their results are based on players who have won game 1 that win the set. This means statistics based on the player winning the set by winning either game 1,2 or winning game 1,3. Their results do not account for skill level at all (because there isn’t currently a way to measure it in the first place) nor is it accounting for which games were won in the set (aside from the first game).

If you were to compare it to your results, then you would have to use the same variables. So although the inverse of losing game 2 and 3 is 88%, this number is not derived from similar variables being measured in the other analysis (winning game 1,2 + winning game 1,3= 82%).

Furthermore your results are based on 100,000 matches of people of equal skill level at all times (the other values having different numbers) as opposed to the variable skill level in the real analysis. Your results are quite unrealistic, and such statistics is not holding much external validity here. This is another reason as to why I am referring to your statistics as fabricated.

I truly don't understand the issue you have with my data. You sound like you think I made it up.

You made a script that simulates battles/ battle results. Hence, it is nothing more than a mathematical calculation which in this case is based on significant amounts of assumptions that do not account for the variables in realistic data. Battles that are simulated are not real. If the battles are not real, then the results are not based on real data.

Clarify this for me.

For clarification on my side: Although your results follow simple statistical logic, the values used (and the lack of control for variables) in the formulas do not reflect reality well. It does not hold external validity at all.

Obviously people don't have a static value that represents how good they are, nor can we quantify the advantage or disadvantage a counterpick gives. But in order to discuss rulesets and outcomes, we have to make assumptions best as we can.

It’s fine to make arguments based on logic. It’s not fine to use unrealistic (or invalid) statistics as support for baseless logic.

It also makes it difficult for others to argue for or against it, unless they take the time to read through all of the data and understand why it's flawed. This is very counterproductive, since you want criticism.

I made some fair assumptions in order to run a simulation, and the outcome mirrors real life outcomes.

The arguments presented by me and several others have shown that you have not made fair assumptions. The outcome does not mirror real life outcomes.

My reason for the KTLT method was thus. Both players are given a guaranteed counterpick. These counterpicks were supposedly equal strength. They were both supposed to be "safe" matches where each person counterpicking is expected to win the match if the skill gap between players is low. Then it would go to a neutral match where both players use the data from the last two matches to try to win in a neutral setting.

It’s quite clear the reasoning for your method. It should also now be clear (or at least after completely reading this post) as to why your assessment wasn’t correct, and to why your proposed method is flawed.

If the skill gap is high, better player may beat worse player on his counterpick and end the set early.

This is true for the classic method. As for the proposed method, it is only possibly more likely to happen over the classic method (based on your logic) if the better player gets counterpick first which is based on a 50/50. This is why matches will not be faster in your proposed method based on your logic.

We play sets rather than single matches for the reason that we don't believe that a single match is truly reflective of the skill difference between tow players. But in the current system, the first match ends up having a large impact on the outcome of the set anyways. This is what I was trying to change. With my system, the first two games between two players near the same level of skill should end up going 1-1 and the final neutral game should decide the outcome, after equal amounts of data have been collected by both players.

Similar logic can be made as counterpoints here. Having a blind counterpick match that forces a player to fight at such a disadvantage, devalues the match too much. Less data (or less valuable data) can be collected against the player when there is such a difference in advantage, and without having previous player data, the player may not even feel like he can win that match at all.

However, playing on the neutral stage first, has great value for both players since either player can win and gain the advantage, while at the same time collecting the most valuable data. Now the player at least has data to help them fight against the counterpick advantages. This makes the counterpick values not as powerful. In the classic method every match has more value in testing the players’ skill levels over your proposed method.

But I stand by the fact that in theory, the new system runs faster. But perhaps not better.

As already explained, it does not make sense for the new system to run faster, and most certainly not better.

 

[Kneato]

Your results are not comparable at all. Their results are based on players who have won game 1 that win the set. This means statistics based on the player winning the set by winning either game 1,2 or winning game 1,3. Their results do not account for skill level at all (because there isn’t currently a way to measure it in the first place) nor is it accounting for which games were won in the set (aside from the first game).

If you were to compare it to your results, then you would have to use the same variables. So although the inverse of losing game 2 and 3 is 88%, this number is not derived from similar variables being measured in the other analysis (winning game 1,2 + winning game 1,3= 82%).

Furthermore your results are based on 100,000 matches of people of equal skill level at all times (the other values having different numbers) as opposed to the variable skill level in the real analysis. Your results are quite unrealistic, and such statistics is not holding much external validity here. This is another reason as to why I am referring to your statistics as fabricated.



You made a script that simulates battles/ battle results. Hence, it is nothing more than a mathematical calculation which in this case is based on significant amounts of assumptions that do not account for the variables in realistic data. Battles that are simulated are not real. If the battles are not real, then the results are not based on real data.

Clarify this for me.

For clarification on my side: Although your results follow simple statistical logic, the values used (and the lack of control for variables) in the formulas do not reflect reality well. It does not hold external validity at all.



It’s fine to make arguments based on logic. It’s not fine to use unrealistic (or invalid) statistics as support for baseless logic.

It also makes it difficult for others to argue for or against it, unless they take the time to read through all of the data and understand why it's flawed. This is very counterproductive, since you want criticism.



The arguments presented by me and several others have shown that you have not made fair assumptions. The outcome does not mirror real life outcomes.



It’s quite clear the reasoning for your method. It should also now be clear (or at least after completely reading this post) as to why your assessment wasn’t correct, and to why your proposed method is flawed.



This is true for the classic method. As for the proposed method, it is only possibly more likely to happen over the classic method (based on your logic) if the better player gets counterpick first which is based on a 50/50. This is why matches will not be faster in your proposed method based on your logic.



Similar logic can be made as counterpoints here. Having a blind counterpick match that forces a player to fight at such a disadvantage, devalues the match too much. Less data (or less valuable data) can be collected against the player when there is such a difference in advantage, and without having previous player data, the player may not even feel like he can win that match at all.

However, playing on the neutral stage first, has great value for both players since either player can win and gain the advantage, while at the same time collecting the most valuable data. Now the player at least has data to help them fight against the counterpick advantages. This makes the counterpick values not as powerful. In the classic method every match has more value in testing the players’ skill levels over your proposed method.




As already explained, it does not make sense for the new system to run faster, and most certainly not better.

For all your assumed knowledge of statistics, it doesn't even take my simulation which you say is flawed to realize that sets would end faster with my method than the classic.

As stated before, for a Bo3 to end in 2 matches in classic, one player must win a mostly even match and then a match in their opponents favor.

In my method, a player must win a set in their favor and then a set in the opponents favor.

It is obvious that the more likely scenario is a player winning two matches with one in his favor than two matches where it is instead just even.

Anyways, this conversation has become rather vitiolic. When fielding the ruleset at the monthly yesterday, people were argumentative over just doing character first, so I didn't bother bringing the method up. I believe the issues with the method that people have brought up here combined with the unwillingness for change, even in the case where clear benefit can be had such as character first, leaves this method dead in the water.

Thanks for the feedback guys.

 

[DMG]

Doing characters first is a pretty decent idea, has been gaining traction past bit. I wouldn't give up on that: I've brought it to attention of DFW PM group in the past and reception is overall mostly positive.

 

[skellitorman]

As stated before, for a Bo3 to end in 2 matches in classic, one player must win a mostly even match and then a match in their opponents favor.
In my method, a player must win a set in their favor and then a set in the opponents favor.
It is obvious that the more likely scenario is a player winning two matches with one in his favor than two matches where it is instead just even.

The problem here is that you are making many assumptions that are not correct, as already previously explained. You are assuming that the data gained from the opponent has equal value when playing on a neutral stage first versus a counterpick stage first. You are also assuming that the better player is just as likely to beat an opponent’s counterpick without information, which is not the case.

The better player is more likely to win against the opponent’s counterpick when they have gained significant data from the opponent. If the better player plays on the opponent’s counterpick first, then he has to win the match with no previous data. Due to having no previous data, it is possible that the match could become more disadvantaged for the better player (especially since he has less information to make the best of his bans), making it more difficult for the better player to win the first match (meaning he is more likely to lose here as opposed to playing the neutral first and then playing against the counterpick). Furthermore due to such disadvantage it is also possible that the data gained from the opponent will have less value.

Not only that, but it is possible that the better player may not utilize his counterpick strength as strong of a counterpick as the worse player. He may play only one character and play on a neutral stage (due to feeling more comfortable on a neutral stage). However, he still has to deal with the momentum carried by the opponent if the opponent wins the first game with such an advantage due to the counterpick.

Likewise if the better player has to counterpick first, he may choose to play a secondary character who has “x” strength (let’s say a strong projectile game) thinking that the secondary will do better than his main against the opponent’s character that he is attempting to counterpick. The opponent could be abnormally strong at dealing with projectiles, but weak against normal fundamental play. The better player obviously wouldn’t know that since he would lack previous data that would be gained from the first neutral (playing his main) and could have made it more likely for the worse player to win the first match.

As for the cases when the better player's counterpick is advantageous. If the better player is playing against an opponent on a neutral stage versus his own counterpick, he is still most likely to win (meaning that the extra advantage is of negligible worth since he is bound to win in the first place). However, he becomes more likely to lose against the opponent’s counterpick without gaining proper information from the first round. Therefore it only becomes significantly more likely for the better player to lose a match when counterpicked against him first which only happens in your proposed method. This means that your proposed method is more likely to go to game 3 over the classic method, and is thus not faster.

Anyways, this conversation has become rather vitiolic. When fielding the ruleset at the monthly yesterday, people were argumentative over just doing character first, so I didn't bother bringing the method up. I believe the issues with the method that people have brought up here combined with the unwillingness for change, even in the case where clear benefit can be had such as character first, leaves this method dead in the water.
Thanks for the feedback guys.

I already apologized in my previous post if it seems that I am “attacking” you or being harsh. I just naturally speak this way, so that I can avoid misunderstandings by being as clear as possible.

Nevertheless, character first has already been tested by 4tlas at his local scene successfully (correct me if I’m wrong). Furthermore, character first has been already debated here heavily and has been seen as something worth testing prior to it being tested. Your method has been criticized heavily due to its weak support, and it has been shown that it is not something worth testing.

Nobody here is trying to attack you personally. The conversation has not become vitriolic (or shouldn't be seen as malicious in any way). You asked for criticism, and I am giving it to you. You can dismiss the criticism all you like, but it doesn’t change the reality of the situation. Your method is weak (including its support), and evidence has been shown to support that stance.

 

[4tlas]

Yes, we have tested Character First here and found moderate support from our players. We have implemented it for the foreseeable future with no backlash (surprisingly).

 

[mimgrim]

Character first just makes more sense in how it works and follows screen order. Which are both huge. Only time I ever see people have a problem with it is because it "limits" counterpicking potiential (even though that should be an easy fix by having the correct ratio of stages to bans, i.e. if 2 bans then 3 small stages are needed so characters who do amazing on small stages don't get screwed over and same for characters who like big stages).

 

[DMG]

No the concern usually was "if I have to lock chars and may not have enough bans for small/big stages, then being Bowser/etc is made more awful" AKA characters who are less stage-universal would see less play overall because you can't dodge stage bullets anymore.

With that said, taking out some of the ambiguity over what stages to ban and pick, due to the uncertainty of what both players will choose to play as, is probably a good thing in such a character/stage combo heavy game.

 

[4tlas]

No the concern usually was "if I have to lock chars and may not have enough bans for small/big stages, then being Bowser/etc is made more awful" AKA characters who are less stage-universal would see less play overall because you can't dodge stage bullets anymore.

Just elaborating on this.

As I mentioned in my analysis from running the trial ruleset, most players did not mind someone dodging the stage counterpick. They did mind having it inverted because they didn't know what their opponent played and they did mind having someone counterpick them AND then switch character.

 

[Kneato]

The problem here is that you are making many assumptions that are not correct, as already previously explained. You are assuming that the data gained from the opponent has equal value when playing on a neutral stage first versus a counterpick stage first. You are also assuming that the better player is just as likely to beat an opponent’s counterpick without information, which is not the case.

The better player is more likely to win against the opponent’s counterpick when they have gained significant data from the opponent. If the better player plays on the opponent’s counterpick first, then he has to win the match with no previous data. Due to having no previous data, it is possible that the match could become more disadvantaged for the better player (especially since he has less information to make the best of his bans), making it more difficult for the better player to win the first match (meaning he is more likely to lose had he played the neutral first and then be counterpicked). Furthermore due to such disadvantage it is also possible that the data gained from the opponent will have less value.

Not only that, but it is possible that the better player may not utilize his counterpick strength as strong of a counterpick as the worse player. He may play only one character and play on a neutral stage (due to feeling more comfortable on a neutral stage). However, he still has to deal with the momentum carried by the opponent if the opponent wins the first game.

Likewise if the better player has to counterpick first, he may choose to play a secondary character who has “x” strength (let’s say a strong projectile game) thinking that the secondary will do better than his main against the opponent’s character that he is attempting to counterpick. The opponent could be abnormally strong at dealing with projectiles, but weak against normal fundamental play. The better player obviously wouldn’t know that since he would lack previous data that would be gained from the first neutral (playing his main) and could have made it more likely for the worse player to win the first match.

As for the cases when the better player's counterpick is advantageous. If the better player is playing against an opponent on a neutral stage versus his own counterpick, he is still most likely to win (meaning that the extra advantage is of negligible worth since he is bound to win in the first place). However, he becomes more likely to lose against the opponent’s counterpick without gaining proper information from the first round. Therefore it only becomes significantly more likely for the better player to lose a match when counterpicked against him first which only happens in your proposed method. This means that your proposed method is more likely to go to game 3 over the classic method, and is thus not faster.



I already apologized in my previous post if it seems that I am “attacking” you or being harsh. I just naturally speak this way, so that I can avoid misunderstandings by being as clear as possible.

Nevertheless, character first has already been tested by 4tlas at his local scene successfully (correct me if I’m wrong). Furthermore, character first has been already debated here heavily and has been seen as something worth testing prior to it being tested. Your method has been criticized heavily due to its weak support, and it has been shown that it is not something worth testing.

Nobody here is trying to attack you personally. The conversation has not become vitriolic (or shouldn't be seen as malicious in any way). You asked for criticism, and I am giving it to you. You can dismiss the criticism all you like, but it doesn’t change the reality of the situation. Your method is weak (including its support), and evidence has been shown to support that stance.

I'm sorry, I guess terse would have been a better term than vitriolic. And that could have easily just been my perception.

Some of the criticism I think was based on a misunderstanding of the method, but there were enough legitimate problems that made me realize that the new method would be flawed as many have stated before and as I mentioned in my last post that it's "dead in the water"

Also, I don't think I made it clear but I am all for character first and we did eventually run it but not without some complaints from the entrants. I think it will just take some getting used to.

I apologize for mostly derailing this thread. I think we were last talking about stagelists.

What do people see as pros and cons of a large stagelist with more bans? Or a smaller stagelist with less?

I think small lists with less bans would be ideal if we had some way to ensure that every character had a certain number of stages that were good for them but that seems unrealistic with our current options.

Big lists are more likely to have polarizing stages in them, but more bans would help counteract their negative impact. More bans however could also really hurt characters with few preferred stages to play on.

 

[DMG]

Well a larger stage list would also likely include more stages for characters to prefer. If a character only has 3 good stages right now, going to say 20 stages with 5 bans doesn't have to mean that character got stuck with nothing new that was good. The main concern in larger list usually isn't about finding appropriate coverage for every character, but whether the game gets more polarized if a lack of enough bans occurs.


Example of this: say we run a 14 stage list with 3 bans for each player. We decide to expand the stage list to 20 stages, 5 bans each. If characters primarily gain advantages or disadvantages based on stage size, if you added 3 small 3 large, this would give the counter picking character 1 open stage that the additional bans do not cover. Say Bowser got 3 more small stages: the opponent only received 2 more bans in this leap and he can't cover all stage threats so to speak.


In reality, it can go beyond that though since character advantages and disadvantages often tie into more than just stage size (platform arrangement, blast zones, walls, etc). A character like Ike may find himself with 4 more favorable stages due to this combo: Small, Small, Small, Large (with walls).


One possible solution to this is to split up Small/Large additions by sprinkling in a decent number of medium stages. If we leap to 20 stages, but the balance is 2/2/2 instead, then your additional 2 bans may cover you adequately if the concern is Small/Large for your character. You could take it further and even do 1/4/1. So it may be feasible but at a certain point it becomes very hard to judge for 40+ characters and lord knows how many MU's PM has, on whether they have access to appropriate ban coverage for their MU's.

 

[mimgrim]

Example of this: say we run a 14 stage list with 3 bans for each player. We decide to expand the stage list to 20 stages, 5 bans each. If characters primarily gain advantages or disadvantages based on stage size, if you added 3 small 3 large, this would give the counter picking character 1 open stage that the additional bans do not cover. Say Bowser got 3 more small stages: the opponent only received 2 more bans in this leap and he can't cover all stage threats so to speak.

I think it is a good idea to leave these kinds of characters that perfer a type of stage size to anything with at least one open stage of a certain size so that they don't get crippled in the counterpick front (and being able to ban all stages of a certain size againt a character who likes that size indirectly nerfs them to an extent and takes away from the point of a counterpick).

Stage lists can very much affect the viability of (a) character(s) and how well they can preform (just look at Brawl for examples galore of that).

 

[Mc.Rad]

Hey Guys...



Why does the Project M Ruleset here dosn't have Wario Man banned but has Giga Bowser? I was always confused by that, aren't they supposed to be super powered versions of the characters with crazy hit-boxes and things like that???

 

[Narpas_sword]

Hey Guys...



Why does the Project M Ruleset here dosn't have Wario Man banned but has Giga Bowser? I was always confused by that, aren't they supposed to be super powered versions of the characters with crazy hit-boxes and things like that???

Because the ruleset was created before Wario man was selectable as an Easter egg. He, like other final Smashes, is definitely banned from play.

It needs updating.

@Umbreon

 

[Deleted member]

Because the ruleset was created before Wario man was selectable as an Easter egg. He, like other final Smashes, is definitely banned from play.

It needs updating.

@Umbreon

i intend to update this, but i have to address something else too first. i'll come back to this thread in about a week

 

[Kneato]

i intend to update this, but i have to address something else too first. i'll come back to this thread in about a week

The clock is ticking Umbreon. I haven't left my computer in anticipation. And I've only blinked like twice now.

 

[Kneato]

Hey everyone, Kneato here. You might remember me from bad ideas such as "Lets change how picks and bans work" and "Bowser's other Castle should be legal". I'm back to propose yet another radical idea, and this time, I can guarantee it maybe isn't ****.

Lately I've seen many people comment on how heavily character matchups play into the game and tournament/match outcomes. As PM develops, this seems like this will only increase as the effort to make every character viable will likely introduce very good and very bad matchups for every character.

This becomes an issue at events like Paragon. As a number of people have mentioned, that tournament was truly stacked for PM. And realistically, there were a number of good player who could have feasibly taken the tournament, but it ultimately boiled down to which players got the best bracket and were able to avoid terrible matchups.

I've also seen people suggest that in an effort to combat this, a new system closer to round robin should be implemented for the late stage of tournaments. The issue here lies mostly in time constraints. A major reason we currently do double elimination is that it quickly culls a large number of players in an exponentially smaller number of rounds. Round robin cannot be implemented as the number of rounds needed becomes unfeasible at larger than 8 participants (28 matches as compared to 14/15 in 8 man double elimination).

So, that leads us to Kneato's Next Stupid Idea! (KNSI™)

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Tournament plays out normally (pools into X man double elim bracket) until top 8 is reached.

Once only top 8 are left, begin a Swiss System with seeding/pairing set up like so:

Players are only paired with other players of identical score. Additionally, if two players have already met in bracket and different new players with the same score exist, they play the new players instead. Rounds would proceed like so:

Now at the start of round 3, you'll notice players 7 and 8 have lost two matches and have been eliminated. They tie for 7th place. In this sense, this Swiss system operates similarly to the double elimination bracket. But since these players made it into top 8, they basically reset their loses from the double elim bracket they have previously played. This combined with being paired against players they have not yet met gives players a better chance at not being knocked out by being unlucky with bad matchups. Now after round 3 plays out:

Players 5 and 6 have now lost 2 matches and tie for 5th. This leaves 4 players, three of which are tied, and one who has gone undefeated. 1st through 4th would be decided by a full round robin between these top 4 players with player 1 having one extra win added to his record to reflect his better score at the end of the Swiss System.

In a double elimination format, top 8 takes 9 or 10 matches. This new system takes 17. I believe Bo5 could be safely reserved for just the top 4 RR at the end due to the more forgiving nature of elimination in this format but I could be wrong. This would lessen the time impact the new system would have, but it would still take longer than a straight double elim.

I think any system that is more likely to favor the better player rather than the better matchup is worth it given the time cost is reasonable. What do you guys think about this system? Does it accomplish this goal? Perhaps is there some improvement that could be made or an entirely different method that accomplishes the same goal?

 

[nimigoha]

I like the idea of finding a new system to fit with PM's matchup-based game. It's definitely very creative.

However, having nearly double the amount of matches is a tricky thing.

Another thing is that there's not really a "grand finals". The four players play their matches and then you say "I guess Player 2 won all his matches so he wins the tournament". Although I guess the final match of the night can be arranged to be the deciding match.

The best part of this is the little system where you can change which players play each other if people have already met. I like that a lot.

Unfortunately this still only solves the matchup problem for the top 8 players. Pools are fine because it's RR, but as soon as you start the bracket to build to top 8 you're still left with these "middle men" who still have a chance of running into 2 terrible matchups, who otherwise could have won the tournament.

Also the fact that we barely won Character First counterpicks' inclusion into the official ruleset, which is such an obvious choice, means that getting this radically different system accepted will be extremely difficult.

 

[mimgrim]

So I've been thinking about stages a lot more recently.

So I'm sure some of you know how I feel about how ban to number of stage size needs to be not skewed against characters who like a particular kind of stage size.

2 bans overall seems to be the most popular (although I still see 3 beings used else where, including my region). With 2 bans I think a 12 stage list is best so we get 4 stages of each size (small, big, and medium) and 5 starters (since it is going to be impossible to get anyone to not use that it seems like) with 3 medium, 1 big, and 1 small.

If 3 bans then a 15 stage list with 5 of each size. Otherwise following the logic of above.

The problem becomes deciding just which stages to use.

So let's start with small stages.

Green Hill Zone, Fountain of Dreams, and Wario Ware are all easy ones that just about everyone agrees on more or less. The problem comes with the 4th stages. Yoshi's Story has some vocal opposition from certain players (most notable one being @Umbreon ) though from what I have seen they have never actually explained why the stage is so horrible. The other option I believe is Metal Cavern which also has opposition to it.

The medium stages we have an easy BF and SV for 2. After that other medium stages are either disliked to one degree or another, espically if to be put as starters (YiB and CS) or aren't really medium sized (PS2, DP).

Big stages we get PS2 and FD as two super easy ones to use. After that it gets a bit difficult. As all other legal worthy big stages have one dislike or another about them, it seems like, or in the case of one doesn't off much over FD (sadly). Dreamland is seen as having way to big blast zones and characters living too long, Delfino is seen as "jank" because of the way the platforms look, and alternate Bowser's Castle has problems with being a alternate stage right now (once alt becomes the main this shouldn't be an issue though).

Basically, we need better stages made to fill in certain gaps. :L

 

[Kneato]

I like the idea of finding a new system to fit with PM's matchup-based game. It's definitely very creative.

However, having nearly double the amount of matches is a tricky thing.

Another thing is that there's not really a "grand finals". The four players play their matches and then you say "I guess Player 2 won all his matches so he wins the tournament". Although I guess the final match of the night can be arranged to be the deciding match.

The best part of this is the little system where you can change which players play each other if people have already met. I like that a lot.

Unfortunately this still only solves the matchup problem for the top 8 players. Pools are fine because it's RR, but as soon as you start the bracket to build to top 8 you're still left with these "middle men" who still have a chance of running into 2 terrible matchups, who otherwise could have won the tournament.

Also the fact that we barely won Character First counterpicks' inclusion into the official ruleset, which is such an obvious choice, means that getting this radically different system accepted will be extremely difficult.

Yea the time thing is definitely an issue. But in a number of tournaments, top players are asking for all of top 8 to be Bo5. Worst case, that leads to 45 games. Worst case with the new system would be 51 games. Each game being ~4 minutes puts it at ~24 minutes slower than Bo5 for all top 8. Not terrible but definitely something to consider. This system could be implemented in situations where the schedule is not tight.

And the grand finals bit is true but on the flip side, it provides more top level matches for spectators. They get to see all top 4 play a set against one another which wouldn't happen in double elim.

For the middle men, there is feasibly nothing that can be done. There are just too many of them to have any kind of RRish system for, and any yet unthought of system that could be devised for them would likely be even more radically different.

The last bit about getting rule changes accepted, it doesn't matter. If there are issues it is important to discuss them and theorize solutions at least to get more people talking about them. Change is slow, but if it is for the better of the game, it is always worth it.

 

[Kneato]

So I've been thinking about stages a lot more recently.

So I'm sure some of you know how I feel about how ban to number of stage size needs to be not skewed against characters who like a particular kind of stage size.

2 bans overall seems to be the most popular (although I still see 3 beings used else where, including my region). With 2 bans I think a 12 stage list is best so we get 4 stages of each size (small, big, and medium) and 5 starters (since it is going to be impossible to get anyone to not use that it seems like) with 3 medium, 1 big, and 1 small.

If 3 bans then a 15 stage list with 5 of each size. Otherwise following the logic of above.

The problem becomes deciding just which stages to use.

So let's start with small stages.

Green Hill Zone, Fountain of Dreams, and Wario Ware are all easy ones that just about everyone agrees on more or less. The problem comes with the 4th stages. Yoshi's Story has some vocal opposition from certain players (most notable one being @Umbreaon) though from what I have seen they have never actually explained why the stage is so horrible. The other option I believe is Metal Cavern which also has opposition to it.

The medium stages we have an easy BF and SV for 2. After that other medium stages are either disliked to one degree or another, espically if to be put as starters (YiB and CS) or aren't really medium sized (PS2, DP).

Big stages we get PS2 and FD as two super easy ones to use. After that it gets a bit difficult. As all other legal worthy big stages have one dislike or another about them, it seems like, or in the case of one doesn't off much over FD (sadly). Dreamland is seen as having way to big blast zones and characters living too long, Delfino is seen as "jank" because of the way the platforms look, and alternate Bowser's Castle has problems with being a alternate stage right now (once alt becomes the main this shouldn't be an issue though).

Basically, we need better stages made to fill in certain gaps. :L

I think we went over something along these lines a few pages back. There aren't enough good stages right now to have a universally accepted list. Also, I think we need to settle on the stagelist size (more likely 12 than 15 if we want universal acceptance) and base # of bans off of that, rather than the other way around.

YIB I personally hate because it is Slants: The Stage. The issue with that being that the main stage slants are irregular enough that being a tiny fraction to the left or right completely alters the angle of most attacks, and there are numerous such spots across the stage.

 

[Kneato]

Alternatively, TOs could instead just aim to have more of the tournament done as round robin pools and have a smaller double elim end bracket like 8 man max size.

However, TOs can barely handle pools at large events as it is. It's always a plague of lost pool sheets, missing/late players, and illegible writing. I think this would only be possible if we had a better system for pools, like a challonge implementation where each pool was given a password so they could enter the results for their pool themselves, via their own phones or a tablet provided by the event.

 

[Deleted member]

ok so tl;dr i've been looking into this on the whole after paragon and i don't think we should be using brackets anymore. i may not get back to this for a while, so any edits that need made to the opening post, just list them and i'll edit them in. but right now this is a big ??? situation with no real conclusion yet so it may be a while.

 

[TheGravyTrain]

mimgrim
Can you elaborate on 12 stages? Why is ten not acceptable (3 small, 4 medium, 3 large) with 2 bans. Then you get one extreme cp and another moderate cp, if your cp's are based on physical size. Later I will discuss what I classify each stage as, but I don't have enough time at.

 

[mimgrim]

mimgrim
Can you elaborate on 12 stages? Why is ten not acceptable (3 small, 4 medium, 3 large) with 2 bans. Then you get one extreme cp and another moderate cp, if your cp's are based on physical size. Later I will discuss what I classify each stage as, but I don't have enough time at.

It is to take into account Bo5s.

In just Bo3s 10 stages are fine with that ratio because they only need one counter pick at the most.

In Bo5s though, because of whatever version of DSR is being ran, you'll still be out on your second CP.

 

[Bazkip]

Just do 2 bans in bo3 and 1 ban in bo5.

 

[TheGravyTrain]

mimgrim
I realize that, but that is why I said you get one severe cp and one moderate cp. I think that is better than adding in a few more questionable stages to the list. Or you could do what Bazkip suggested, that is fine by me.

 

[mimgrim]

I mean, I would prefer to keep the rule-set the same throughout if possible instead but that could work.

But even a 10 stage list comes into problems itself on what to use.

The Small stages are easy, WW, GHZ, and FoD. All are mainly universally accepted stages to use in any list.

Medium stages are still a pain with the only easy ones being SV and BF. YiB and CS still have alot of dislike towards them (especially CS) and any other suggestion are hated even more (poor Lylat, so misunderstood).

Big stages we have FD and PS2 as easy acceptable ones. The last one will lead to some kind of dissent as I outlined before.

I also dislike this number because of how it affects big stages.

Mainly in regards to FD. As despite it being a large stage, it doesn't really act like one in the normal sense for characters who like large stages.

 

[Bazkip]

Maybe instead of saying the stage number should be changed, you should realized that FD is a horribly polarizing stage and probably shouldn't be legal.

Except in doubles, it's fine in dubs.