Stock Count: APEX Case Study

[Dr. Tuen]

Purpose
This study seeks to characterize brawl matches via the APEX video data set. The winner of the first stock and the winner of the second stock will be compared to the winner of the match in an effort to characterize the effect of reducing the stock count. This study is limited in that it cannot account for altered character choice or play style with respect to the hypothetical stock reduction.

Motivation
Brawl is a slow game. Sets in this game have been shown to take an average of 18 minutes to complete. The nearly doubles a game which is known for it's slow projectile based play (Touhou IaMP) [1]. This description often extends to spectators of the game and is added to the ever-increasing public criticisms that this community faces.

Concentrate [2], a series held by Juice Gaming, has utilized a number of experimental rule sets, one of which includes the use of one stock matches. In the public eye, these sets have mildly changed the look of brawl. Even the most stubborn dissenters describe the game as "bearable" under that type of rule set.

The most obvious benefit outside the public perception of the game is the time spent when running an event. A standard set takes 18 minutes, while a one stock set can run in under 10 minutes (though an average time for this format has not yet been characterized). If the run time of events were reduced by half, the need for copious set ups may be alleviated, time for friendlies may be increased, and late events may become a thing of the past. "May" is the operative word, as the skill of the Tournament Organizer is still the most important factor.

Unfortunately, the community has shown little interest in the reduction of stocks [3]. The reduction is seen as a metagame altering effect, with Zero Suit Samus and Pokemon Trainer seen as two of the most benefited characters. There is notion that the ruleset would change smash's tendency to allow for comebacks (last stock outcome reversals). This study seeks to characterize this effect with respect to a confined data set.

Methods
The APEX tournament video archives for 2009, 2010, and 2012 were utilized in this study [4]. When the sets were decoupled, the total game count for this set of data is 198.

Videos were reviewed to determine the winner of the first stock (the first player to take a stock in a game), the winner of the second stock (the first player to take two stocks in a game), and the winner of the third stock (the winner of that game).

Results and Discussion
When the analysis was finished, the number of mismatches were tabulated. A Stock 1 mismatch is characterized as a difference between the "winner of the 1st stock" and the winner of the match. This is similarly characterized for a Stock 2 mismatch. The results are listed below:

Stock 1 Mismatch Percentage: 24.2%
Stock 2 Mismatch Percentage: 19.2%

If this measure were to accurately predict the future of a reduced-stock metagame, less than one quarter of the results would change in an event. To further investigate the "comeback factor", the number of "last stock reversal"s and "two lead changes" were tabulated. A "last stock reverse" is a game in which Player 1 leads by taking the first and second stock first. In the end of a "last stock reverse" game, Player 2 wins the game. A "two lead change" game involves two stock count lead changes.

Last Stock Reverse Percentage: 11.6%
Two Lead Change Percentage: 7.6%

If these were to be indicative of "intense matches", very few matches would count themselves within these categories.

The last category that was recorded in this study is timeouts. This occurs when the time expires in a match and the winner must be declared by the rules put forth by the Tournament Organizer.

Total Timeouts Recorded: 5 (2.5%)
Percentage Caused by a Specific Player: 80%

The second of the two numbers listed is based on the popular assumption that said player is the largest cause of timeouts. The name is not important to this study.

Conclusion
The results are for the community to interpret. The results imply a change of less than 1/4 of all matches with the hypothetical rule change, though that may also imply the complete loss of all "last stock reverse" and "two lead change" match types. To further characterize the change's impact on the community, a frequency analysis on the common victors before and after the rule change may be conducted.

It is in the author's opinion that the change would greatly cater to the spectating community (those that do not yet play a Smash Brothers game), and likely increase the player base.

====================================================
====================================================

References
1. TheRealBobMan - http://allisbrawl.com/blogpost.aspx?id=136229&page=2
2. http://allisbrawl.com/ttournament.aspx?id=12448
3. http://www.smashboards.com/showthread.php?t=318871
4. http://www.youtube.com/user/apextournament

 

[TreK]

Well I don't think anyone would be against actual test of this ruleset over mere theorycrafting. What stops TOs from testing this as a side event, or as a temporary replacement that could become the reference in case of success ?

Not bashing on your work, but most of this data and the conclusions that come out of it were already widely accepted (I was surprised by the timeout part but that's only because you guys are whiney people that make it sound like everyone and their sister gets time-out-'d in American Smash)
What we lack atm for this to be standard is community approval. Tbh, I wouldn't give away my opinion on this ruleset without experience to back me up and I'm probably not the only one.

 

[Browny]

Sets in this game have been shown to take an average of 24 minutes to complete.

O RLY?

Prove it.

 

[Ussi]

100% agree

I can't even watch brawl between 2 people i don't know and I play brawl!

I barely can watch high level of play without losing attention.

 

[Dr. Tuen]

O RLY?

Prove it.

Oh eff. I forgot to put in the edit. that's definitely supposed to read 18 minutes, and that is definitely speculation. I could go fish out an average time at some point.

 

[popsofctown]

"Less than a quarter" sounds very different from "24.2%", which is almost exactly a quarter.

As a player I don't want to lose to a weaker player a 1/4th of the time. "Less than a quarter" is really downplaying the impact ripping such a large fraction out of the competitiveness of the game.
A rate of a quarter illegitimate victories is equivalent to cutting the competitiveness of the game in half. You're going from
[SKILL][SKILL][SKILL][SKILL]
to
[SKILL][SKILL][CANDYLAND]

Going from a whole to 3/4s might seem like a 1/4 change, but think about it - if you double that number of illegitimate upsets, you have 50% of games going to the better player - a pure coinflip.
25% of games becoming illegitimate upsets is a 50% reduction in competitiveness. 50% of games being illegitimate upsets is a 100% reduction in competitiveness - a totally uncompetitive game like Candyland.

Of course the amount of competitiveness is a bit off because the 3 stock game isn't an authoritative answer on which player is better. But if your issue with my math is that 3 stock games are inadequate, it's hard to argue for stock reduction.

 

[#HBC | Joker]

Why are you arbitrarily doubling the 1/4 to 1/2? You say

"You might think going from a whole to 3/4 is only a 1/4 change, but if you double it, it's 50%"

but you don't give any reasoning for doubling it whatsoever. That doesn't make sense. It's blatantly 25%, not 50%. Based on this data, as flimsy as it is, the winner will remain unchanged 75% of the time. That means the better player is gonna win the set, despite what ruleset gets used.

 

[Dr. Tuen]

"Less than a quarter" sounds very different from "24.2%", which is almost exactly a quarter.

As a player I don't want to lose to a weaker player a 1/4th of the time. "Less than a quarter" is really downplaying the impact ripping such a large fraction out of the competitiveness of the game.
A rate of a quarter illegitimate victories is equivalent to cutting the competitiveness of the game in half. You're going from
[SKILL][SKILL][SKILL][SKILL]
to
[SKILL][SKILL][CANDYLAND]

Going from a whole to 3/4s might seem like a 1/4 change, but think about it - if you double that number of illegitimate upsets, you have 50% of games going to the better player - a pure coinflip.
25% of games becoming illegitimate upsets is a 50% reduction in competitiveness. 50% of games being illegitimate upsets is a 100% reduction in competitiveness - a totally uncompetitive game like Candyland.

Of course the amount of competitiveness is a bit off because the 3 stock game isn't an authoritative answer on which player is better. But if your issue with my math is that 3 stock games are inadequate, it's hard to argue for stock reduction.

Although there are some fuzzy points with your math (your skill/skill/candyland is divided in thirds), this has been addressed.

I'd rather make an interesting comparison. The community made its largest and most successful push to ban Meta Knight after the polls reached about 75% in favor of the ban. This shows that 75% of matches will remain unchanged if this video archive is close to indicative of the type of play that will be displayed in 1 or 2 stock matches.

Of course, I will never say that "75% of the community agrees" and "75% of the 2nd stock winners are the same as the 3rd stock winners" are mathematically equatable. However, I've seen time and time again that the community works emotionally, rather than logically. It would seem to me like an emotional appeal, say... some numbers reaching this 75% mark, would work well.

In the end, I'm not actually advocating for either side. This just happens to be the center of the community's focus right now and I wanted to provide some numbers. Interpret them as you will.

 

[popsofctown]

Why are you arbitrarily doubling the 1/4 to 1/2? You say

"You might think going from a whole to 3/4 is only a 1/4 change, but if you double it, it's 50%"

but you don't give any reasoning for doubling it whatsoever. That doesn't make sense. It's blatantly 25%, not 50%. Based on this data, as flimsy as it is, the winner will remain unchanged 75% of the time. That means the better player is gonna win the set, despite what ruleset gets used.

It's not arbitrary, it's just.. hard to explain.

Competitiveness isn't how often the better player wins, it's the margin between how often the better player wins and how often the worse player wins. In arm wrestling that margin is 100. In a coin flipping game the margin is 0. In a game where the better player only wins 3/4 of the time, the margin is 50.

My definition might seem arbitrary, but it's not. If competitiveness is how often the better player wins, then a coin flipping game has some amount of competitiveness, because the better player wins half the time. But we know a coin flipping game isn't competitive at all, so there's something wrong with the reference point. The important value is actually the margin between the winner's chances and the loser's chances.


Tuen's point about 75% is valid, 75% is, emotionally, a very acceptable number. Which is why I want to explain why 50% is the real number here, though it's hard to do so.

EDIT:
Don't confuse my arguments about the interpretation of Tuen's data as a protest about the collection of the data or his methodology. It was good of him to do so, and the methodology is fine.

 

[Dr. Tuen]

My definition might seem arbitrary, but it's not. If competitiveness is how often the better player wins, then a coin flipping game has some amount of competitiveness, because the better player wins half the time. But we know a coin flipping game isn't competitive at all, so there's something wrong with the reference point. The important value is actually the margin between the winner's chances and the loser's chances.

I don't understand the logic in its entirety, but I have another interesting argument to make. Your current interpretation is as follows: if the data above indicates the behavior of 1 or 2 stock matches, then the better player wins 75% of the time.

My point: who is to say the better player won 100% of the time under the current ruleset?

 

[popsofctown]

If 3 stock games are, say, only 80% accurate, with 20% of all games rewarding an arbitrary victor, going from 3 stocks to 1 stock won't affect that 20% of the games. So you would restrict the impact of the change to the remaining 80% of matches and lose half the competitiveness entailed in that. So 40% of your matches will be impacted instead of half.

So yes, if 3 stocks isn't fully accurate, the impact of the change is reduced by however often it fails.
But if your argument is that 3 stocks is very frequently failing to pick good winners, you really only weaken your stance further. If 3 stocks is already inaccurate, then we are even closer to the tipping point where people start quitting a game over the impact of variance (like Melee with items, or the poor drafting formats that crop up in MtG from time to time.) So there is even less safety in cutting the stock count.

 

[Dr. Tuen]

If 3 stock games are, say, only 80% accurate, with 20% of all games rewarding an arbitrary victor, going from 3 stocks to 1 stock won't affect that 20% of the games. So you would restrict the impact of the change to the remaining 80% of matches and lose half the competitiveness entailed in that. So 40% of your matches will be impacted instead of half.

So yes, if 3 stocks isn't fully accurate, the impact of the change is reduced by however often it fails.
But if your argument is that 3 stocks is very frequently failing to pick good winners, you really only weaken your stance further. If 3 stocks is already inaccurate, then we are even closer to the tipping point where people start quitting a game over the impact of variance (like Melee with items, or the poor drafting formats that crop up in MtG from time to time.) So there is even less safety in cutting the stock count.

I don't necessarily weaken my argument. I weaken it an strengthen it at the same time. A 'quantum argument', so to speak, heh. Since the rules in smash are so fluid, we have no way of saying which method more accurately tests skill. I have an idea for a "trip significance study", but we'll see how that pans out later. Since we cannot accurately state that one rule set favors the skilled player, one of two situations can occur:

1. 1 stock is weaker. As you stated, if we go with 80/20, the margin of error for the weaker player winning increases.

2. 1 stock is stronger. In this case, we reduce that 20% margin and maybe get a 90/10 split in favor of the better player.

Who knows? Not me.

 

[popsofctown]

Hypathetically one stock could be stronger than three, yes. But there's no data to back up either direction, while common sense strongly suggests 3 stock games will produce more consistent results.

 

[Ussi]

There is little data of one stock. We have some tourney vids with 1 stock, with ZSS footage too as well.

 

[teluoborg]

Where are the graphs ? If you want to call this science then there has to be graphs !

JK, good work, especially for having the patience to watch all the apex sets and for doing a proper scientific post with sources and stuff.

1 stock tournaments sound fun, like almost a different game from what we're having right now.

 

[Luigi player]

Oh, so if "only" 1 out of 4 matches have a different outcome that's not a big deal? Wow.

Also you can't just say that that's how matches would go with less stocks. Maybe someone tries harder in their last stock, and also matches won't always have the same outcome even if you wouldn't take that into account. You can't just say "well that's how it happened these times so it'll likely be like that other times".

You'd need much more data and in the end you probably can't even get a true percentage anyway.


I don't know what people have. Being able to stay at the top is what makes games good competitvely.
You say people want difference, because it get's boring or unhyped? You probably never thought how the people will feel that usually won till now. Consistency makes a good competitive game. If it's all random then there's no point in competition... it'd be more like a party game.
If people want to get better then we shouldn't let them randomize everything. How about you start practicing. The people that put the most effort in the game are the best. If you don't have the time for that then well sucks for you, but the people that put more time should be better. That's a good thing.

Maybe this sounds exaggerating. I know it wouldn't change that often/much, but if 1 out of 4 times you lose cuz of less stocks instead of 1 out of like 10, that's still somewhat of a big deal and annoying. And you could just prevent the "randomness" by playing more matches/having more stocks.

 

[The Ben]

I don't know what people have. Being able to stay at the top is what makes games good competitvely.
You say people want difference, because it get's boring or unhyped? You probably never thought how the people will feel that usually won till now. Consistency makes a good competitive game. If it's all random then there's no point in competition... it'd be more like a party game.

Variance != random. People want to see a wide range of variance because it makes the game have a broader appeal. A broader appeal means more competition. If every match was exactly the same the game would be bland.

 

[-Ran]

I recall doing similar research in the past on this, except my requirements were to find out relatively evenly skilled matches at the higher end of the spectrum. I did this, because I felt that the sand bagging of a superior player would skew the results. In my findings [~53 matches, 20 sets], I found that 86.7% of the matches had the same ending, if you had stopped the fight at the second stock.

 

[Vinylic.]

Instead of taking out stocks, why not lower the time limit?

 

[C.J.]

I think you mean increase the time limit? If stocks stay the same and the time limit is lowered, the game will get even campier as soon as one person has any lead.

 

[Vinylic.]

More campier, as people already do that in an 8 minute match?

 

[-LzR-]

If this game is campy then why should we even change the way it is played? I don't see any reason to change the timer or the amount of stocks as it has been perfectly fine so far. Also PT gets to use all the pokemon and Lucario can fully use aura in this ruleset too.

 

[Dr. Tuen]

If this game is campy then why should we even change the way it is played? I don't see any reason to change the timer or the amount of stocks as it has been perfectly fine so far. Also PT gets to use all the pokemon and Lucario can fully use aura in this ruleset too.

You get local events, which aren't quite as equipment ready as regional events or national events, which take 6 hours to run. It'd be much nicer to newcomers if we could wrap up on 2-3 hours and yield plenty of time for fun side events, opportunities for friendlies, and social time.

 

[#HBC | Red Ryu]

Eh...by that logic with Lucario he technically gets more with 4 stocks, like it was a long time ago.

Better to leave him out of discussion with stocks.

 

[#HBC | Ryker]

Vinyl's response made my eyes bleed.

Before I lose vision completely, what do you view as the sacrifice of moving to a one stock format and where do you draw the line on it being worth it?

Past that, discuss the effect of ZSS and PT for me. I live in the same state as the number one PT and near enough to see the number three (two?) ZSS on a regular basis. The ruleset buffs or nerfs these characters arbitrarily, but one will remain subjectively better seeing as you are talking about spectator value as your main selling point.

 

[hichez50]

This whole debate probably won't go anywhere without the community targeting T.O's, many of which are stubborn. They have legitimate reasons to be though as they don't want to affect there tournament attendance. Although I don't think they have noticed that compared to other main stream games that brawl attendance is underwhelming.

The solution is to find effective ways to test new rule sets. One of the problems is that naturally the community follows the highest level players. Being a high level player doesn't necessarily mean that you have leadership qualities. Things like the Rule-set committee attempted to solve the problem and only have mild success. As of now keeping the same rule-set is inevitable.

There are only a couple camps all of them having problems. The first camp is to keep the stock and time the same. The problem with this is that we have the same long tournaments that we have now. The benefit is that there will be no metagame change. You have the camp that wants to increase the time limit. This will make tournaments longer. The result will be less timeouts though. I don't see why people don't why timing out people as a strategy that should stay. Sure it may not be fun to watch but neither is prolonging an already campy game. I also remember seeing some where in the past that timeout only effects around 10% of all games. The 3rd camp is to decrease the the stock count. The fall back is that it changes the metagame. The benefit is that it changes the metagame. I don't see a problem with changing the metagame. The metagame is the core of the problem with campy games.

 

[#HBC | Ryker]

The assumption you make there is that having a campy game is a problem when the community developed around said campy game.

 

[hichez50]

The assumption you make there is that having a campy game is a problem when the community developed around said campy game.

True. I will accept that problem with my argument. In the current state though campy games leads to longer games. And long games are a problem. The only thing other things that prolongs tournaments are T.O's being too lenient with kicking people who cross the 5minute time limit they have to show up to a set. But this is irrelevant.

 

[#HBC | ZoZo]

Vinyl's response made my eyes bleed.

Before I lose vision completely, what do you view as the sacrifice of moving to a one stock format and where do you draw the line on it being worth it?

Past that, discuss the effect of ZSS and PT for me. I live in the same state as the number one PT and near enough to see the number three (two?) ZSS on a regular basis. The ruleset buffs or nerfs these characters arbitrarily, but one will remain subjectively better seeing as you are talking about spectator value as your main selling point.

If this wasnt Tuen's thread...

Regardless.
A game with a 50/50 outcome for the most skilled player is uncompetitive, like a coinflip.
The comp value equals zero.
A game with a 100/0 outcome for the most skilled player is competitive, like armwrestling.
The comp value equals hundred.

Thus when you alter 25% of the games, making the outcome 75/25, the comp value equals fifty. That is what was raised earlier on. Just elaborating.

50/50 - 0
75/25 - 50
100/0 - 100

 

[Dr. R.O.Botnik]

Stock 1 Mismatch Percentage: 24.2%
Stock 2 Mismatch Percentage: 19.2%

Because we're not sure exactly how many of these matches were the same, this could be anywhere from 24.2% to 43.4% of matches where the winner didn't have the stock lead the full time.

Last Stock Reverse Percentage: 11.6%
Two Lead Change Percentage: 7.6%

I honestly didn't know that there were so many 3 stock comebacks, but what does the other thing mean?

Total Timeouts Recorded: 5 (2.5%)

Really, that's it? I guess timeouts aren't that big of a deal, then.

It is in the author's opinion that the change would greatly cater to the spectating community (those that do not yet play a Smash Brothers game), and likely increase the player base.

WARNING: THE FOLLOWING RESPONSE IS NOT THAT IMPORTANT TO THE MAIN DISCUSSION, SO PLEASE, FOR THE LOVE OF GOD, DO NOT DERAIL THE THREAD TO ARGUE ABOUT IT.

I disagree with the idea that these things should be a goal we are actively trying to achieve for the following reasons:

1. Brawl is not a spectator sport, meaning that it is not made to be watched by people who don't know how to play. If outsiders watch it and like it, good for them, but that's not the purpose of the competitive scene. The purpose of the competitive scene is to create an environment where it can be determined who is the best at the game. Therefore, the people who the changes are made for should be the actual people playing the game.

2. If people need major changes in the rules to be able to play competitively, then they aren't the target audience. Pandering to potential outsiders who think that regular Brawl is boring will only result in a flood of scrubs who want to play Super Smash Bros. Blurns, which would be destructive to the already established community.

 

[#HBC | Joker]

I don't think you're reading the data right. Those %s are showing how often the winner changed if he reduced the stock count. That means that those %s are showing how innacurate the stock change would be, not how accurate it is. One can only infer "accuracy" based on the lack of "innaccuracy" in the data, if that makes sense.

 

[Dr. R.O.Botnik]

e: I see what you're saying, but my point was that putting it to 1 stock could change the outcome of a quarter to almost half of the matches, which is a gigantic change.

 

[#HBC | Joker]

He's measuring how often results of those matches would be altered if the matches ended after 1 or 2 stocks were lost instead of 3. He's trying to get a sense of how consistent 1/2 stock matches would be with current results.

 

[Dr. R.O.Botnik]

See above edit.

e: Actually, we also know that 11.6% of the time, the winner of the first stock and the second stock lost. That means that changing to one stock would, according to this study, change the outcome of at least 35.8% of matches.

 

[#HBC | Joker]

that's not right though, it would only change 11% or 7% of them. That's literally what the data is saying.

 

[Dr. R.O.Botnik]

Actually, it's not "at least". If the matches were all 1 stock, 35.8% of those matches would have different outcomes (24.2% [the number of matches where the first stock winner lost the match] + 11.6% [the number of matches where the first and second stock winner lost the match]). I think I've finally got this sorted out. You, on the other hand, I have no idea what your train of thought is.

 

[Roller]

You don't list any of the numerous possible confounding variables here. Your entire study goes out the window if you don't control for common **** like the fact that m2k has more vids featured on that channel than just about anyone. Obviously when a huge chunk of the matches in your subject pool are of outliars in the community who consistently flaunt the status quo for stuff like game duration, your data is going to be thrown off. The fact that most of the featured recorded matches are of top players could have an enormous effect on the data as well.


Character usage, Player skill, Matchup dynamics, Stage dynamics, Region, Tournament size, Rule Set, etc would all need to be controlled for in order to get anything close to resembling accurate data for this type of a study which could be generalizable to any type of tournament/scene.

Also, your methods section is a bit lacking.

My guess is you're some sophomore/junior college student who has taken a few courses on statistical analysis or media studies or some **** and think you are able to just instantly apply that to a real life situation; while ignoring half the issues that were raised in said class regarding the problems with misleading/poor studies.

 

[The Ben]

A game with a 50/50 outcome for the most skilled player is uncompetitive, like a coinflip.
The comp value equals zero.

This literally has nothing to do with competitive. Competitive simply implies that there is competition, two or more people willing to compete at this thing. A coin flip can be just as competitive as arm wrestling. Competitive isn't a numerical value, it's a mind set.

 

[Roller]

No, the more random, and less predictable the outcomes of matches, the less competitive the game is.

Think about it, if M2K plays a scrub, obviously he's going to win. Why do we know this? Because the game is skill based, and therefore competitive. As opposed to the scrub having a ~50% chance of winning. It would imply that outcomes could not be at all predicted by skill level, and that random factors played too big a role.

It's why Mario Kart for N64 is said to be more competitive than Mario Kart for the Wii.

 

[Dr. Tuen]

This got a lot of activity since the last time I looked. That's pretty neat. Discussion is good.

Actually, it's not "at least". If the matches were all 1 stock, 35.8% of those matches would have different outcomes (24.2% [the number of matches where the first stock winner lost the match] + 11.6% [the number of matches where the first and second stock winner lost the match]). I think I've finally got this sorted out. You, on the other hand, I have no idea what your train of thought is.

The numbers imply that 24.2% of matches would change if we switched to 1 stock, and 19.2% would change if we switched to 2 stock. That is using a very generous assumption of "there are no other confounding factors". And I do admit there are... such as a difference in playstyle.

The two lead change bit describes a game that goes like this: Dr. Robotnik vs Tuen! Tuen takes the first stock (Tuen leads), then Dr. Robotnik takes Tuen's first and second stock (Dr. Robotnik leads) but then Tuen turns it around by taking Dr. Robotnik's second and third stock for the win (Tuen leads).

That is one of the rarer occurrences in match descriptions, since it is far more common for the winner to be the person who took the first stock first.

You don't list any of the numerous possible confounding variables here. Your entire study goes out the window if you don't control for common **** like the fact that m2k has more vids featured on that channel than just about anyone. Obviously when a huge chunk of the matches in your subject pool are of outliars in the community who consistently flaunt the status quo for stuff like game duration, your data is going to be thrown off. The fact that most of the featured recorded matches are of top players could have an enormous effect on the data as well.


Character usage, Player skill, Matchup dynamics, Stage dynamics, Region, Tournament size, Rule Set, etc would all need to be controlled for in order to get anything close to resembling accurate data for this type of a study which could be generalizable to any type of tournament/scene.

Also, your methods section is a bit lacking.

My guess is you're some sophomore/junior college student who has taken a few courses on statistical analysis or media studies or some **** and think you are able to just instantly apply that to a real life situation; while ignoring half the issues that were raised in said class regarding the problems with misleading/poor studies.

Ooh look! Personal attacks! That's fun. How about I ignore those and get to the point that furthers useful discussion.

Well, pretty much all of those confounding factors are legitimate. It reminds me of the same kinds of factors that show up in education research, but with one important difference: it's much more difficult to get isolated data collection out of the smash community. I've worked on an automatic ranking program before, whose functions were limited because we don't even record set counts in this community.

To correct for things like that, I would have to get a region to cooperate with a long term case study. Tournament organizers would have to implement match slips (paper data collection) or get their video taking up to near 100% and give me that data for inspection. Even then, the generalizability of the study will only apply to that region, and will not legitimize change at the national level.

To address your attack in level manner, the facts are these: I am retired from this game. I do this as a service to the community and I do it out of personal curiosity. I have already done many analytical projects for the sake of community progression in the past, and none have really caused significant change. The community is stubborn and does not respond very quickly to logic which comes from the technical field.

That said, various parts of the 'report' are lacking because of said community's tendency not to respond well to technical information. The last time I produced something similar, it was a 6 page report on MK's dominance which detailed non-parametric statistics in a way that could hopefully bring graduate level information to the level of the average reader on smashboards. That didn't work. The first few responses were along the lines of "statistics is all smoke and mirrors, he can make it say whatever he wants". I even received the loving nickname of "dr. anti-ban".

So in essence, I am writing to my audience. If you would like a fully detailed technical document, I would welcome one authored by you.