Match-up based character ranking

SuperDoodleMan · Jun 11, 2008

I believe an accurate match-up chart can provide all the information needed to make an accurate tier list. The problem of character population can be solved by assuming that better characters will be used more often.

I'll say right now that since this is based on IvanEva's Brawl Character Match-Up chart, it is only as correct as the chart is. I'm posting here partly to open discussion of how to interpret the chart, and partly to draw awareness to the importance of match-ups in ANY tier list.

Enough of that, on to the making of the list.

First, every character is assigned a point value based on the number of (dis)advantageous match-ups they have. +2 for a major advantage, +1 for an advantage, and the opposite for disadvantages.

The worst characters are removed, and the remaining characters' scores are changed to reflect the new pool. Removing fewer characters at a time is more accurate, but takes longer. The bottom is chipped away until there's a list!

Snake
Meta Knight
Pikmin & Olimar
Marth
Toon Link

Game and Watch
Zelda
Falco
ROB

Pit
Lucario
Wolf
Pikachu
Ice Climbers
Ivysaur
Wario
Kirby
Zero Suit Samus

Fox
Lucas
Sheik
Squirtle
Diddy Kong
King Dedede
Donkey Kong
Ike
Link
Charizard
Luigi
Ness
Sonic
Peach
Mario
Samus

Jigglypuff
Bowser
Yoshi

Captain Falcon
Ganondorf

The breaks are where there was a notable margin. Individual order DOES matter. For any given character, they are the worst out of themselves and everyone above them. Within the top ten characters or so, I resorted to the tiebreaker of more neutral matches being better.

Thoughts on all this?

Kiwikomix · Jun 11, 2008

SuperDoodleMan said:
I believe an accurate match-up chart can provide all the information needed to make an accurate tier list.

That's almost true... you do still need tourney results. Also, we don't have an accurate match-up chart. I appreciate the time put into IvanEva's chart and agree that it's a great idea, but the chart is too inaccurate at this point to base a tier list on.

Browny · Jun 11, 2008

SuperDoodleMan said:
Thoughts on all this?

search

this has been done many, many times before

bman in 2288 · Jun 11, 2008

Wow. Post in SDM topic.

And I think the topic is great (bearing in mind the little weight my opinion carries). Though is looks very much like a tier list, I can only guess that tournament placings are needed to form proper tier lists, but then again, we're much too early to determine that...

But I digress. I'm going to admit I expected Falco to be higher up, but I trust the arguments in IvanEva's topic are valid. Oh, and Snake is too good.

Winston · Jun 11, 2008

I really like this approach. As you said, the approximation of just discounting all the bad characters isn't as accurate, but the list you came up with shows it works pretty well. Since it's purely mathematical, it will be able to be refined as the matchup chart is refined. Working on evaluating individual matchups is obviously more precise than saying "well, I think character x is better than character y".

What I'm thinking about now, (though I don't know if you care enough to do this), is what if you wrote a program to do many iterations with small steps? You'd have a numerical metagame represented by the percentages of people playing each character at any given timestep. (It would start evenly divided). For each successive timestep, the change in percentage for each character would be based on your system of advantages and disadvantages, weighted by the current percentage per matchup.

I guess it's not that relevant but I just like the idea a lot.

I might do this if nobody else wants to and if anybody actually cares about this.

petrie911 · Jun 11, 2008

Kiwikomix said:
That's almost true... you do still need tourney results. Also, we don't have an accurate match-up chart. I appreciate the time put into IvanEva's chart and agree that it's a great idea, but the chart is too inaccurate at this point to base a tier list on.

You make the chart from tourney results. Although a tier list should somewhat reflect tourney results, it need not, as how good a character is does not always determine how popular a character is.

Anyways, it looks good, but PT shouldn't be separate. As interesting as it is to see the three pokemon individually, they need to be together. My suggestion is to average PT's best two pokemon of a matchup to create his overall matchup, as you can get by with only 2 pokemon.

Yuna · Jun 11, 2008

Not surprisingly, it's pretty much what I've been saying all along. Except I thought Bowser was a bit better. Other than that, no big surprises for me.

Seanson · Jun 11, 2008

If you want the tier list to be based solely off of tourney results then just look at tourney results. This method takes into account matchups, and therefore actually has more meaning and reasoning behind it.

This method as said, compares essentially each character to all characters that have the chance of being higher ranked. In real tourneys, you don't see many "lower tier" characters because they're all losing to "higher tiers". It can get tough to compare the 10 characters near the bottom because they all lose to higher tiers, so perhaps the best way is to compare them to eachother (not in tournament standings, because there aren't many lower tier players, but compare the matchups).

SuperDoodleMan · Jun 11, 2008

Another problem with tourney results is that they only show the best of the best. The top ten characters might be very easy to place, but beyond that it's no help to look at tourneys.

I'm not saying flat out that tourney results have no place in a tier list (except to provide knowledge for people to change individual character match-ups). But I will say that they have no place in THIS list. How could they even be taken into account? Deciding how to weigh characters brings in too much opinion.

Xiivi · Jun 11, 2008

An easy way to add the tournament results to this would be instead of chipping away based on lowest match-up score, chip-away on lowest tournament output as you are less likely to see the characters with the lower scores.

I'm not saying you should, I'm just saying you could incorporate the two together.

noobslayer · Jun 11, 2008

This thread fails.

Just from a peek at the Match-Up Chart thread, I can see that G&W is supposed to be best.

petrie911 · Jun 11, 2008

^Just because you have good matchups with awful characters doesn't make you good. Sheik could destroy almost the entire lower half of the Melee tier list, but was she Top? no.

FlashGearz · Jun 12, 2008

petrie911 said:
^Just because you have good matchups with awful characters doesn't make you good. Sheik could destroy almost the entire lower half of the Melee tier list, but was she Top? no.

She was top for atleast 2-3 lists. And within the top 3 everytime iirc.

This looks right, and probebly will be about there, but I'd rather see people like Dedede, Falco, G&W a little bit higher. And olimar a little bit lower

doodmahn · Jun 12, 2008

Some things a matchup chart CANNOT account for:
Edgehogging tether recovery up B's.
Chaingrabs

Yes, these do REALLY affect the list.

Seanson · Jun 12, 2008

Aren't those exactly the kinds of things that would be taken into account in match-ups? The fact dedede cant cg some people but can cg others.

doodmahn · Jun 12, 2008

Seanson said:
Aren't those exactly the kinds of things that would be taken into account in match-ups? The fact dedede cant cg some people but can cg others.

Still, look at Oli's tier position. That's because of his easily gimable recovery.

Charizard92 · Jun 12, 2008

http://www.smashboards.com/showthread.php?t=168966

My first thread and essentially why you have to stop putting Charizard on the low tiers.

petrie911 · Jun 12, 2008

doodmahn said:
Still, look at Oli's tier position. That's because of his easily gimable recovery.

IIRC Falco's recovery in Melee was pretty bad, yet he was still Top tier.

Winston · Jun 12, 2008

@ doodmah, charizard92, and others who aren't getting the point:

>.> The point of this is that this is the best way to do a theory-based tier list. Given perfect knowledge of every character matchup in the game, this process would develop the correct tier list.

Thus, with this process down, we just have to work on improving the accuracy of the matchup chart.

If you want an empirically-derived list, Ankoku's working on that in a separate project:

http://www.smashboards.com/showthread.php?t=165954

SuperDoodleMan · Jun 12, 2008

I've considered combining this list with Ankoku's, but I just can't believe that it will show any significant accuracy within the bottom half, unless much more information is gathered from each contributing tournament.

Then there's the problem of how exactly to combine them. Frankly, I'm not sure if it's possible, or even necessary. This list and Ankoku's are each simplifications of a much larger trove of information, from completely different sources. Looking at either list can give some definite insight, but a conglomerate list would be a little of both, and so not as indicative of its source info.

Speaking of combining, I like the idea of combining PT's best two pokemon in any given match-up. The current list would need to be redone, but it's definitely something to do in the future.

Yuna · Jun 12, 2008

doodmahn said:
Some things a matchup chart CANNOT account for:
Edgehogging tether recovery up B's.
Chaingrabs

Yes, these do REALLY affect the list.

How does the matchup chart not account for this? It's obviously counted in when the matchups are written out. "How well can Ivysaur fare against X-character", including the fact that Ivysaur can be easily gimped.

petrie911 said:
IIRC Falco's recovery in Melee was pretty bad, yet he was still Top tier.

That's because it took a lot to get him off stage. Also, it was nowhere near as bad as Olimar's and Ivysaur's.

Winston · Jun 12, 2008

SuperDoodleMan said:
I've considered combining this list with Ankoku's, but I just can't believe that it will show any significant accuracy within the bottom half, unless much more information is gathered from each contributing tournament.

Then there's the problem of how exactly to combine them. Frankly, I'm not sure if it's possible, or even necessary. This list and Ankoku's are each simplifications of a much larger trove of information, from completely different sources. Looking at either list can give some definite insight, but a conglomerate list would be a little of both, and so not as indicative of its source info.

Speaking of combining, I like the idea of combining PT's best two pokemon in any given match-up. The current list would need to be redone, but it's definitely something to do in the future.

Well, I wasn't suggesting combining the lists; I think it's great that there are two separate projects with two completely different approaches.

Was just mentioning Ankoku's thread for those who were talking about tournament results and etc.

soloSHADOWROB · Jun 12, 2008

Just let SDM do his stuff.
There's a reason why we can't base a teir list off tourney results, why? Take a look at Ken for instance, he played Marth and almost always placed first. So why was Marth not the best player in the game?
There are certain advatages and disadvatages that will all take awhile to find out.
Olimar although easily gymped can be struggle to play if you use him correctly, and if so a smart Olimar player is hard to get off the ledge.
Same as Ivysaur, I play him very defensive but agreesively when I have to and win alot.
SDM is just doing it based on a chart and although I agree with alot of things I believe there some things to be changed like King DeDeDe should be higher along win Falco and Fox.
I'm not saying its completely accurate but it is a great teir list conducted so far.

Mr. Escalator · Jun 12, 2008

Just doing a quick check shows that G&W is better than Olimar and Toon Link in terms of matchups.
He may very well be top of top top, but I didnt check.

SuperDoodleMan · Jun 12, 2008

selective quoting for the win

Mr. Escalator said:
Just doing a quick check shows that G&W is better than Olimar and Toon Link in terms of matchups.

petrie911 said:
^Just because you have good matchups with awful characters doesn't make you good.

SuperDoodleMan said:
For any given character, they are the worst out of themselves and everyone above them.

And now for something new: On the left is each character's original score, and on the right is their eventual placement in the list. Once there's a consensus on the accuracy of the source chart, I may even make a graph showing each character's relative placement as each character is removed. That would be a lot of work, though.

DanGR · Jun 12, 2008

nice work!

fkacyan · Jun 12, 2008

Good work, but... Results are all that matter, so I kinda doubt the usefulness of this.

DanGR · Jun 12, 2008

I've been arguing for Olimar>Snake this week, so it might change soon, unless someone else comes in to disprove me.

TheKneeOfJustice · Jun 12, 2008

noobslayer said:
This thread fails.

Just from a peek at the Match-Up Chart thread, I can see that G&W is supposed to be best.

Match-up charts don't reflect tier lists all the time.

petrie911 said:
^Just because you have good matchups with awful characters doesn't make you good. Sheik could destroy almost the entire lower half of the Melee tier list, but was she Top? no.

petrie911 said:
IIRC Falco's recovery in Melee was pretty bad, yet he was still Top tier.

Sheik was always one of the best characters. No question there.

Falco's recovery might have been poor, but on stage, he dominated 90% of the cast.

petrie911 · Jun 13, 2008

TheKneeOfJustice said:
Sheik was always one of the best characters. No question there.

Falco's recovery might have been poor, but on stage, he dominated 90% of the cast

Erm...yes. That was the point. Sheik is good, yes, but she's not Top. Much like GW. He has amazing matchups with many low tier characters, but that means nothing next to his lesser matchups with the high tier ones. Sheik had the same in Melee. Thus, GW is high, but not Top.

Similarly, Falco was in a similar position as Olimar. He had a quite awful recovery, but his amazing on stage game put him in Top tier.

I think I may be misunderstood here. My posts were in response to the ideas that since GW has a ton of big checks, he must be the best, and since Olimar's recovery is easily gimped, he can't be in Top. I was giving examples as to how those ideas are wrong.

KernelColonel · Jun 13, 2008

You should use Ankoku's frequently-updated character rankings. Your list syncs well together with that.

Maybe this thread's list, Ankoku's list, and IvanEva's list can be put together for even moar SOOPER TIER LIST MATHS!

I'm not going to do it though because I'm lazy

SuperDoodleMan · Jun 13, 2008

Uh...this list is already solely based on IvanEva's chart.

Mikezor · Jun 13, 2008

Good work. Last Tier War?

Binx · Jun 13, 2008

Yeah man, last tier war! it was SOOO amazing. I'd go gay for it. Anyhow I would like to see Broom and such get involved in helping us create a correct match up chart so we could really see where we go from here.

KernelColonel · Jun 13, 2008

SuperDoodleMan said:
Uh...this list is already solely based on IvanEva's chart.

Well, I was referring to a tier list in general. You did a good job of yours on account of basing it on Ivan's chart. I was just saying, if someone (anyone, could even be Yuna XD) used your mathematical findings in conjunction with the placement of characters and their frequent usage (i.e. Ankoku's list)....the only data missing, to form a proper tier list, is the individual quirks of each character.

We're pretty close right now, with all the info we have.

petrie911 · Jun 14, 2008

OK, so I did a ranking of my own. It used a different method than SDM, which would probably take too long to explainin detail. The short version is that it involves using game theory to minimize a player's expected value of a match. In the end, it ends up assigning every character a score from 0 to .5, and all characters that score .5 an expected percentage use. Ties were broken using Ankoku's character list, since I really don't have anything else to go on in those cases. The ties generally result from the current coarseness of the matchup chart.

Anyways, onto the list

Top--all characters score 14.3%
Snake
Meta Knight
Mr. Game and Watch
ROB
Toon Link

Upper--all characters score 7.1%
Marth
Donkey Kong
Olimar
Zelda

High
Falco--.4857
Wolf--.443
Pokemon Trainer--.436
Pit--.4286
Kirby--.4143
Wario--.400
Zero Suit Samus--.400

Mid
King Dedede--.3857
Pikachu--.3857
Ice Climbers--.3714
Link--.3714
Samus--.3714
Lucario--.357
Fox--.357
Diddy Kong--.357

Low
Shiek--.3286
Ike--.3143
Lucas--.3143
Peach--.3143
Bowser--.3143
Yoshi--.3143
Luigi--.300

Lower
Ness--.2857
Sonic--.2714
Mario--.2714
Jiggs--.2714

Bottom
Ganondorf--.2286
Captain Falcon--.157

The most glaring difference between this and Ankoku's list is the placement of Pokemon Trainer. Ankoku's list has him in Bottom tier, while this list has him near the top of High tier. Additionally, Toon Link, Zelda, and Olimar are much higher on my list, while King Dedede, Lucario, and Wario are noticably higher on Ankoku's list. My list also seems to place Link and Samus unusually high. Many of these, especially King Dedede and Wario, are probably attributable to inaccuracies in the matchup chart, but I believe the discrepancy in the placement of PT is due to his high difficulty of use.

Match-up based character ranking

Smash Ace

Smash Ace

Smash Hater

Smash Journeyman

Smash Master

Smash Journeyman

BRoomer

Smash Ace

Smash Ace

So much for friendship huh...

Smash Apprentice

Smash Journeyman

Smash Apprentice

Smash Journeyman

Smash Ace

Smash Journeyman

Smash Champion

Smash Journeyman

Smash Master

Smash Ace

BRoomer

Smash Master

Smash Journeyman

G&W Guru

Smash Ace

BRoomer

Smash Hero

BRoomer

Smash Lord

Smash Journeyman

Smash Journeyman

Smash Ace

Smash Journeyman

Smash Master

Smash Journeyman

Smash Journeyman

Information

Network