Official: SSBPD unsupported; source code released.

[stelzig]

A placing is a placing. Period. The tournament host doesn't go "well you didn't play against the following people tied with or below you, go play those matches out now"

I can understand the problem with losing points even though you won and all, but i'm sure the elo can be tweaked to still consider the individual matches for both players.

Edit: To spell it out for you, it does matter because you can get the lead in shorter/longer sets (even in the same damn tournament), but then lose in the end and that somehow changes the result entirely? A set could really be pushed all the way down to a bo1 too, and I believe this has happened in pools at some tournaments before. Sure, the wins were unimportant for your performance at that given tournament, but they are still a relevant source of data for future tournaments.

 

[Fortress | Sveet]

Where did this results thing come in? I'm just going to ignore it since its irrelevent.

I already introduced the idea of using individual games, and it was shot down with good reasoning. Unless you can refute them with good reasoning, it will go by sets.

 

[Kal]

I have explained, twice, why I don't think it's appropriate to base rankings on your set record. I agree that, as a rule of thumb, and as far as effectively calculating ranks, it's true that losing 3-2 is better than losing 3-0. However, in tournament, you don't get bonus points for losing 3-2 instead of 3-0. Rankings based on a tournament should only use the same results the tournament recognizes.

As far as I'm concerned, asking to factor in that a player won 3-2 instead of 3-0 is like asking to factor in that a player won with Roy instead of Marth. Is it an indicator of skill? Probably. Does that mean it's fair to provide more reward in ranking to the Roy player? No. The tournament he's competing in certainly doesn't. Why should the rankings?

And holy **** Varist, look what I found by clicking your sig:

quoth Kal, good things are subjective

I AM FINALLY POPULAR

 

[stelzig]

Exactly, it is irellevant. But that is kinda what you bring up when you say that the impact on the tournament is all that matters. Anything that doesn't alter the results had no impact on the tournament. I'll admit it isn't a perfect comparison, but hey.

Kal: I guess we disagree on how rankings should be created then. I think the importance of rankings is to gauge how well players will perform if they were to participate in a tournament at this very moment, and not how many sets (with different number of matches too) they actually won at the previous tournament(s). For this reason, the roy point is completely unrelated.
And you once again say that they are rankings based on tournaments. What is stopping you from saying they are just based on tournament sets and not the tournament itself?

 

[Kal]

The issue I take, Stelzig, is that it's unfair to rank a player based on a performance that isn't part of the win condition. The tournament rules don't say that you're better off winning 3-0, so why should he get ranked differently for it? This falls back to my Roy example (which isn't irrelevant, I think you're just not understanding it): a player shouldn't get bonus points for winning with Roy, because the tournament doesn't reward him any differently for doing so.

To make this clear: if there were a round robin event, where winning 3-0 had an actual advantage over winning 3-2, I would be fine with using some sort of "match win percentage." However, in a bracket, there is exactly one win condition for progressing: winning a set. A player does not need to attempt to win 3-0 in order to win, so ranking him lower for doing absolutely nothing wrong within the context of the event makes no sense.

 

[ajp_anton]

ajp, the issue we take with counting wins, instead of only sets, is that in tournament your individual wins are unimportant. What matters is whether you win the set. So, it would be strange to have a tournament where "winning 3-2" and "winning 3-0" are identical, but to have rankings, based on said tournament, where they are not.

Whether you win the set or not depends on the individual matches.
3-2 and 3-0 give the same result, but ELO is based on statistics, and using "3-2" instead of "1-0" gives you more statistical data to predict how often someone will win a set. See the last three words, "win a set", not matches. The matches are only used for prediction.

Taking a game off Armada doesn't give you anything in *that* tournament, but it means that you actually have a non-zero chance of taking a match off Armada, and therefore a non-zero chance of taking a set off him. That one single match suddenly tells you that "in one out of 10000 sets, he will win against Armada" without the need for them to play 10000 sets.

 

[stelzig]

I'd argue that it is part of the win condition. The condition just involves more steps and the player didn't complete all steps.

However, do you really think it is more important to stand by these "win conditions" than to actually make as accurate a ranking as possible? (not to mention doing it faster, as AJP pointed out earlier).

The thing about roy/low tier btw, I guess makes sense if the winner used roy and doesn't normally do so (we could give him that extra reward for truly showing dominance or something, lol), but it isn't as simple to just calculate and we can't really know if the player just happens to be really good with roy. Who's to say that this player is more likely to perform better than someone who doesn't use a low tier? There is no way to really add this into an automatic calculation, and it could only really be considered if we make opinion-based powerrankings.

The biggest problem I see with considering individual matches would be the bracket matches often just being reported as win/lose and therefore individual match data isn't saved. A problem with the roy thing in all cases too btw.

 

[ajp_anton]

Too late to edit because of stelzig:

I do however agree with your point that because winning 3-2 has no advantage over winning 3-0, people may sandbag in sets they know they will win regardless.

 

[Strong Badam]

Tio doesn't allow you to put in whether the set is 3-0 or 3-2 is a pretty obvious reason why we can't.
Chess Elo ratings don't check to see whether you beat your opponent with all your pieces or with just a King and Rook, either. It's a win, a loss, or a draw. I think others have already given enough reasons for not including such a feature as well.

 

[Zivilyn Bane]

Stelzig and AJP, it is a bad idea to calculate elo based off set count. If you think about it more you will understand. And frankly it is never ever in a million years going to happen, so let's let it go. And even if we wanted to do it, the data simply just doesn't exist. And because of the fact that it doesn't exist, if we DID do it, I want to claim right now that every set I've ever lost that isn't uploaded to youtube was either 1-2 or 2-3. Please prove otherwise.

See what happens?

 

[Kal]

I'd argue that it is part of the win condition. The condition just involves more steps and the player didn't complete all steps.

But it's not. The win condition is a discrete, well defined concept: when you have won three matches, you have won.

Would I agree that winning two matches in a row without losing any would increase your chances of winning the set? Sure. Going Fox also increases my odds of winning the set. That doesn't mean that the win condition involves going Fox.

However, do you really think it is more important to stand by these "win conditions" than to actually make as accurate a ranking as possible? (not to mention doing it faster, as AJP pointed out earlier).

Yes, because it's plain and simply unfair. Grant me, for example, that Pichu vs. Fox is 20-80 in Fox's favor. It would be correct, then, to infer that I deserve more credit for winning against a specific Fox player with Pichu than with Marth. However, the tournament itself grants no such distinction: if I win with Pichu, or if I win with Marth, my reward is identical (I progress to the next round). If I lose with Pichu, or if I lose with Marth, my penalty is identical (I get knocked out of the bracket).

What you're asking is for players to recognize that, if I win with Pichu, this means more than if I win with Marth. And, while certainly true, the problem is that it's unfair: the rules of the event don't specify any reward for winning with Pichu over winning with Marth, but somehow the rankings, based of this event, do?

The unfairness is more glaring in the case of rewarding low-tier mains, but the issue is the same with considering matches. If you want to consider matches, your ruleset needs to account for it by having the win ratio on matches be relevant. If it's not relevant, as with a bracket, then it shouldn't be a factor.

Also, ajp, you're reiterating the same points, but clearly missing mine. As a rule of thumb, it's certainly true that taking one match off of Armada is an indicator that you are better than a player with the same Elo losing 3-0. However, the win condition of the tournament is winning three matches. Not winning three matches and minimizing your losses. Why should the player get rewarded in rankings for not actually achieving anything relevant to the actual event?

 

[stelzig]

Tio doesn't allow you to put in whether the set is 3-0 or 3-2 is a pretty obvious reason why we can't.

I have never done it myself, so I guess this may be true. I've just assumed it to be possible because TIO has a "show win-loss results" box you can tick.

Kal: Now you try to involve matchups. I can agree entirely that these can influence results, even player matchups do so as well on single results. This is still unimportant. The question is wether or not considering the individual matches will improve the ranking or not. You seem to agree that it would so i'm extremely puzzled to why you would oppose it.

The matchups (of all kinds) still puts up the problem that you CAN'T add it into the calculations anyway.


ps. I would like to point out that i'm still interested in making accurate predictions for future sets/tournaments and not make some weird skill ranking (I believe you used the word "skill" earlier). So wether the pichu is more "skillfull" than the marth or not isn't important to me. It's for tournament performance at any given moment.

Edit: Damn, I need to read these posts more closely, but on the other hand I guess nitpicking everything would make the posts unnecesarily long too... Anyway, picking fox is not a neccesary step in order to win the set. Winning 3 matches is neccesary and therefore winning 1 and 2 matches is as well. You are required to be able to win 1 and 2 matches in a BO5. You are not required to be able to use fox.

 

[Kal]

Yeah, thanks for that.

 

[KishPrime]

Just as another practical reason, a lot of old tournaments are all 2-0 because we don't have those details. Heck, we barely have brackets. It would drastically increase the inaccuracy of the system.

 

[FoxLisk]

3-2 and 3-0 give the same result, but ELO is based on statistics, and using "3-2" instead of "1-0" gives you more statistical data to predict how often someone will win a set.

no, it actually doesn't. it gives you more data on how likely that person is to win a game, and a game is below the resolution of how we define wins and losses.

 

[ajp_anton]

no, it actually doesn't. it gives you more data on how likely that person is to win a game, and a game is below the resolution of how we define wins and losses.

But with the probability of winning a game, you also get the probability of winning a set. These two are hard-linked together with a simple conversion formula.

Stelzig and AJP, it is a bad idea to calculate elo based off set count. If you think about it more you will understand. And frankly it is never ever in a million years going to happen, so let's let it go. And even if we wanted to do it, the data simply just doesn't exist. And because of the fact that it doesn't exist, if we DID do it, I want to claim right now that every set I've ever lost that isn't uploaded to youtube was either 1-2 or 2-3. Please prove otherwise.

See what happens?

I already accepted the sandbagging-argument (plus there's also counter-picking, altering the probabilities slightly), but in principle it would've still worked. Just weigh the results differently, and you can mix matches and sets (where match results don't exist) in the calculations.


Now, if we are done with the match discussion, what about the issue I originally brought up - the mixing of bo3, bo5 and bo7? That is in fact the same thing as your problem with mixing individual matches with sets. I could claim that my bo7 loss was a bo3 loss instead and lose less points than I should - oh wait, that's what the system is already doing.


edit:
And I realize there isn't much that can be done about this, so we might as well just keep going and ignore the slight inaccuracies. All I wanted was to point out that the system has flaws so people are aware of it, but then it blew up into a longer than necessary discussion whether what I was saying was even true...

 

[Zivilyn Bane]

The fact that the system counts a set win as a win, regardless of if it is a bo3, 5, or 7, is not a flaw. If you win your set you gain elo accordingly and it's as simple as that. If you try to over complicate things it waters down the entire system. Just because you don't think it should be done that way doesn't mean it's a flaw.

 

[Kal]

Well, ajp has a point, but the issue isn't really set length alone. It's that a different set length makes a different game, and so we're taking results from different games and compiling them into a single ranking system. The only real solution is for all tournaments to use the same rules.

You could similarly complain that some wins are done with Rainbow Cruise legal while others aren't. This all falls down to the ruleset. In my opinion, it doesn't have a large impact on the meaning of your Elo: a high Elo indicates a high level of skill in all rulesets, regardless of set length and stage list, and conversely a low Elo indicates a low level of skill in all rulesets, regardless of set length and stage list.

 

[ajp_anton]

The fact that the system counts a set win as a win, regardless of if it is a bo3, 5, or 7, is not a flaw. If you win your set you gain elo accordingly and it's as simple as that. If you try to over complicate things it waters down the entire system. Just because you don't think it should be done that way doesn't mean it's a flaw.

A win in bo3 is different than a win in bo5 in ELO, simply because the probabilities are different. I don't feel like calculating the difference, it might be small, and in our case we pretty much have to assume it's small because we don't know the match counts. But it's a fundamental part of ELO and it's not going away simply because you choose to ignore it.

Kal:
I don't think stages matter, as long as the same set of stages are available (as you mentioned, the tournament rules). Which stages that were played on is part of winning the set. Good Counterpicking and stage banning is part of your performance in that set.
The set length however will change and will *only* change the very thing that ELO measures. It changes the win/lose probability, with longer sets favoring the better player.

 

[JOS.fm]

so for this tournament, which i didn't attend.
http://www.ssbpd.com/tournament/Subway Fight Night
JOS is actually Anupam
"SOS is DQ'd" is a bye.

thanks!


and i think the issue of DQ's has been resolved?

also, i flagged Vash as Kevin. when the merge happens, the Kevin from
http://www.ssbpd.com/tournament/Test Your Skills Bar Fights 6
and
http://www.ssbpd.com/tournament/Winter Game Fest VI (Melee)
and
http://www.ssbpd.com/tournament/Subway Fight Night
who beat smiles and anupam and lost to jake and oscar will be Vash.

and Kevin (the second Vash) will hereafter be named Kevin or something else

and then i accidentally flagged armada as being from texas wtf.

 

[Kal]

ajp, when you have different stages, different set lengths, or anything different in the rules, you have a different game. So, from a pedantic point of view, you should have a different Elo for each game. Elo measures your probability of winning, relative to another player, for a specific game. It's a little crazy to pretend different stage lists are ok in this regard, but that different set lengths are not.

 

[Fortress | Sveet]

What about the NTSC/PAL differences? They are literally different games.

 

[TheCrimsonBlur]

God Kal why are you so awesome.

But with the probability of winning a game, you also get the probability of winning a set

This is not necessarily true. A Peach may beat a Marth on Dreamland 60% of the time but has a 0% chance of winning on any other stage. So even though they win 2-1 in a set against that Marth, they are not even close to winning overall; they just have a strong counterpick that can win them a game. They are actually LESS likely to win a set against that Marth than a Falco which has a 10% chance of winning on all stages.

 

[Zivilyn Bane]

Some of you are failing to realize what the aim of the DB is. It isn't to keep statistics of every single match. It's to show tournament results and associate an Elo rating with each player that can be used to approximate a players skill in the game of Super Smash Brothers Melee. Any tournaments that restrict a players right to open choose their character is banned.

So tell me, do you REALLY think the difference between a Bo5 or a Bo7 or a Bo3 would make the difference between Armada winning Apex 2012? He is simply the best player and I don't care what the set count is, he is the most likely player to win ANY given set with a balanced and competitive ruleset. It in no way means you need to completely streamline all tournaments.

Do ALL Nascar races take place on the SAME track? Do all football games use the exact same football? Do all Soccer games use the same referees? Do all chess tournaments use the same time control? You guys are getting a little crazy extreme with what you are interpreting as how this thing should be ran. Which makes me just a little more happy that FoxLisk is in charge of this whole thing because he has a great concept of not only Elo, but what makes it really work.

 

[TheCrimsonBlur]

One of the most exciting features we can add is detailed measurements for the size/growth of the community.

Like, we can get exact measurements for # of entrants across all tournaments in 2008 v. 2011. And then we could specify it by region. So we'd have stats for TOs and regional leaders to aim for year-by-year improvement.

(this is all assuming we can get everybody on board with submitting all results to the database)

 

[Rud Lisi]

But with the probability of winning a game, you also get the probability of winning a set. These two are hard-linked together with a simple conversion formula.

Couldn't the same thing be said for stocks for a game? This is an extreme example, but it involves the same reasoning as to why games aren't being counted.

 

[ajp_anton]

ajp, when you have different stages, different set lengths, or anything different in the rules, you have a different game. So, from a pedantic point of view, you should have a different Elo for each game. Elo measures your probability of winning, relative to another player, for a specific game. It's a little crazy to pretend different stage lists are ok in this regard, but that different set lengths are not.

Agreed.

This is not necessarily true. A Peach may beat a Marth on Dreamland 60% of the time but has a 0% chance of winning on any other stage. So even though they win 2-1 in a set against that Marth, they are not even close to winning overall; they just have a strong counterpick that can win them a game. They are actually LESS likely to win a set against that Marth than a Falco which has a 10% chance of winning on all stages.

Agreed.

Couldn't the same thing be said for stocks for a game? This is an extreme example, but it involves the same reasoning as to why games aren't being counted.

Agreed.


Really, are you people even reading what I've writing. You're arguing about things that I already agree with.

 

[FoxLisk]

But with the probability of winning a game, you also get the probability of winning a set. These two are hard-linked together with a simple conversion formula.

Absolutely not. Based on your response to TheCrimsonBlur, I'm not sure where you're confused - obviously you can't both believe that he is correct and believe that this statement is true. Let me know what you don't understand about why this is wrong and I'll try to make it more clear.

Now, if we are done with the match discussion, what about the issue I originally brought up - the mixing of bo3, bo5 and bo7? That is in fact the same thing as your problem with mixing individual matches with sets. I could claim that my bo7 loss was a bo3 loss instead and lose less points than I should - oh wait, that's what the system is already doing.

You would lose exactly the correct amount of points in each case because a set is an atomic unit of winning or losing.

A win in bo3 is different than a win in bo5 in ELO, simply because the probabilities are different. I don't feel like calculating the difference, it might be small, and in our case we pretty much have to assume it's small because we don't know the match counts. But it's a fundamental part of ELO and it's not going away simply because you choose to ignore it.

The probabilities have nothing to do with it. It's a bit. You win a set or you lose it. There's no probabilities involved, it's all or nothing, there is no grey area, there are no complicating details. You win a set or you lose a set. Anything below that resolution is meaningless. This is, admittedly, logistically required because of limitations in Tio, but also actually the correct decision.

The set length however will change and will *only* change the very thing that ELO measures. It changes the win/lose probability, with longer sets favoring the better player.

Whether or not longer sets favor the better player is not something I'm going to take for granted. I think it depends a lot on your definition of "the better player," for one thing.

In the melee community, what Elo scores are used to measure is the relative abilities of different players to win sets against each other. As Kal, I think, mentioned, different set lengths (along with different stages and different numbers of stocks and different timers and anything else) are actually a different game. Since it wouldn't help anyone to act accordingly, though, we're pretending that all the games we're playing are the same, in which case, as I have made as clear as I can bear, a set is a unit of winning or losing, and your complaint doesn't make sense.

 

[Scar]

this discussion is still going on?

Now for the actual problem:
I'm assuming that the vast majority of the sets that have been uploaded to the system are best of 3. The ELO points should tell us the probability that a player wins over another *in a best of 3 set*.
If a best of 5 set is uploaded, that result is biased towards the better player, and will not give an accurate result. Sometimes the worse player should win, but this happens less frequently in bo5.

in bold is the stated problem from ajp. does anyone agree that ELO should tell probability of Bo3? i submit that stage / ban ruleset would contaminate data in a similar way, should we then track each stage that was played, too?

bonus question (and the real kicker): if you agree, do you have a suggestion for how to collect the information and compile it? the cool thing about the SSBPD is that (thanks to neal, and foxlisk and friends) we have a single format containing all the information we care about, and an easy way to compile and analyze.

 

[ajp_anton]

Absolutely not. Based on your response to TheCrimsonBlur, I'm not sure where you're confused - obviously you can't both believe that he is correct and believe that this statement is true. Let me know what you don't understand about why this is wrong and I'll try to make it more clear.

Mathematically (or "in theory" if you will), you have a certain probability of winning each match. From this, you can calculate the probability of you winning a bo3 or bo5 set.
The two flaws are the human mind (sandbagging, or otherwise playing differently when you know you can afford to lose a match or two) and different counterpicks actually making each match different.

You would lose exactly the correct amount of points in each case because a set is an atomic unit of winning or losing.

The probabilities have nothing to do with it. It's a bit. You win a set or you lose it. There's no probabilities involved, it's all or nothing, there is no grey area, there are no complicating details. You win a set or you lose a set. Anything below that resolution is meaningless. This is, admittedly, logistically required because of limitations in Tio, but also actually the correct decision.

If the above "in theory" example worked, this atom could've been split. However, there are different kinds of atoms...

Whether or not longer sets favor the better player is not something I'm going to take for granted. I think it depends a lot on your definition of "the better player," for one thing.

I still think that the example above works to a certain degree. Sometimes people just aren't playing at their peak, and this is (in part) where the probabilities come from. A worse player always has a chance of winning a match, but the longer the set goes on (assuming the players don't get tired etc.), the less likely it is for these random fluctuations to alter the final outcome of the set.

 

[Rud Lisi]

Agreed.

Agreed.

Agreed.


Really, are you people even reading what I've writing. You're arguing about things that I already agree with.

I thought i did read ur stuff, my bad, i guess im just not clear, i did just sorta skim over it. The point is that you have to draw the line at some point, and tournaments (NOT FOXLISK) have determined that line is winning a set.... like someone else said, they don't give you any sort of benefit for losing 3-2 vs losing 3-0, so why should the system (based on tournaments) pick your deciding line instead?

I do see what ur saying about bo3 vs bo5 vs bo7, but i think the solution is to have a standardized ruleset, this problem shouldn't fall on this system to fix, but i do think, to make this system more fair, there need to be international standards

 

[Zivilyn Bane]

Take it from somebody who has played nearly a decade of tournament chess. When you play in a tournament, it could be game in 30 (30 minutes per player), game in 45, 60, 120, 180, 20 moves in 1 hour and sudden death in an hour. You might be black. You might be white. REGARDLESS it is the same game and the best and most consistent players will win the vast majority.

Elo is NOT what you are wanting it to be. It is not a mathematical formula to figure out who is going to win each match specific to the amount of games that are to be played in that set. It is a numerical approximation of a players skill level relative to the players that person has been played. And it's a great system to compare how good Player A from Texas is compared to Player B from Florida. But that doesn't mean we know who would win if they played, just a rough idea.

 

[Dr Peepee]

Mikemoney, the person that I beat at NC tourneys listed, is actually $mike. Mike$ or mikemoney, is a player from CT that is not the one that currently lives in NC lol. $Mike is an NC player who is pretty good. Issue is, $mike went to a few NE tourneys and maybe a no johns or two in NY when he went to school in MA for a bit, so that may be where the mixup came from.

I'd fix this myself but I can't seem to get into this thing. <.<

 

[ajp_anton]

Take it from somebody who has played nearly a decade of tournament chess. When you play in a tournament, it could be game in 30 (30 minutes per player), game in 45, 60, 120, 180, 20 moves in 1 hour and sudden death in an hour. You might be black. You might be white. REGARDLESS it is the same game and the best and most consistent players will win the vast majority.

Elo is NOT what you are wanting it to be. It is not a mathematical formula to figure out who is going to win each match specific to the amount of games that are to be played in that set. It is a numerical approximation of a players skill level relative to the players that person has been played. And it's a great system to compare how good Player A from Texas is compared to Player B from Florida. But that doesn't mean we know who would win if they played, just a rough idea.

It is a mathematical formula to figure out *the probability* of who is going to win. This doesn't change the fact that it still works well for how it's being used in practice, which is not calculating probabilities, which I honestly can't see anyone doing. It just is what it is.
Those different game types will alter the ratings "in the wrong way", but so will a bad day or just pure luck. Obviously the system still works well, which is why I've said that the difference between bo3/5/7 is likely small.

 

[Kal]

in bold is the stated problem from ajp. does anyone agree that ELO should tell probability of Bo3? i submit that stage / ban ruleset would contaminate data in a similar way, should we then track each stage that was played, too?

It should tell the probability of winning the game. What game is being played is the issue; are we playing Smash with best of three or best of five? Is Rainbow Cruise legal?

In a real way, every ruleset creates a different game. If you want to be especially pedantic, you would have an Elo ranking for each ruleset, the same way you would have one for Chess, and a separate one for Bughouse. However, we consider that most of these games are "similar enough" that we ignore this pedantry, choosing to combine these results under the assumption that a player good at one ruleset will be good at another, and a player bad at one will be bad at another.

In other words, what probability the Elo provides is entirely dependent on what game is being played. In Chess, the Elo is a ranking according to individual games of Chess, hence it would make sense to be able to calculuate the probability of winning a best of three or best of five in Chess by using the Elo. Here, our Elo is less well-defined. We use it for a variety of different (albeit similar) games, hence it's not going to necessarily indicate any real probability. Ideally, we would have everyone playing Melee with the same set-length, with the same stages, and with the same overall ruleset, and then we would be able to claim that the Elo is meant to measure the probability of a player winning a set at that set-length. Which it will do, by construction.

Right now, I think we should accept the Elo for what it is: an estimate of who is better than who, with a number attached to give you a skill range. Intending for it to be an actual, rigorous Elo is simply not possible, seeing as we aren't all playing exactly the same game at each tournament.

 

[Zivilyn Bane]

Mikemoney, the person that I beat at NC tourneys listed, is actually $mike. Mike$ or mikemoney, is a player from CT that is not the one that currently lives in NC lol. $Mike is an NC player who is pretty good. Issue is, $mike went to a few NE tourneys and maybe a no johns or two in NY when he went to school in MA for a bit, so that may be where the mixup came from.

I'd fix this myself but I can't seem to get into this thing. <.<

Ok thanks for the clarification. We were talking about this earlier and I couldn't really figure it out. My knowledge of EC players is really low.

Ok so I currently have each player under these tournaments:
MikeMoney: Pound 4, Tunes Biweekly Revived #1, Saffron City #1, Pound 3
$Mike: Smash Rehash 3, Smash Rehash 5

Let me know if any of these are wrong and I can fix it.

 

[Scar]

It should tell the probability of winning the game. What game is being played is the issue; are we playing Smash with best of three or best of five? Is Rainbow Cruise legal?

In a real way, every ruleset creates a different game. If you want to be especially pedantic, you would have an Elo ranking for each ruleset, the same way you would have one for Chess, and a separate one for Bughouse.

my point exactly, except that i wouldn't consider them to be of varying degrees of pedantry. if you're going to make one distinction without the other then you're being inconsistent. so there's no point in making the Bo5 vs Bo3 distinction without distinguishing between rulesets.

 

[Deleted member]

The system is only as intelligent as the data it has. Go upload tournaments from back in your day if you want to see your rating increase in accuracy.

2 things:

1. i wasn't even at genesis 2. i've never even been further went than south bend for FC for smash. also the only other tournament i'm listed under, several of those losses are also incorrect. that's why it's wrong.

2. the vast majority of tournaments that I won were under smurf names. I've won tournaments as mew2king, DA_RAIN, X Japan, DireVulcan, NJFoxMasta007, The_Cape77, and tons of other smashers that already exist. So yeah I don't see that working out lol

 

[Kal]

my point exactly, except that i wouldn't consider them to be of varying degrees of pedantry. if you're going to make one distinction without the other then you're being inconsistent. so there's no point in making the Bo5 vs Bo3 distinction without distinguishing between rulesets.

There's a sort of irony to making the distinction between the degrees of pedantry, despite that having very little to do with my post.

 

[Scar]

yeah i thought my post made it clear that, besides on a side-issue, i agreed with you completely

sounds like you just missed that i was referring to it as such