Actually, the last time I played Hawk on HDR a few weeks ago, I was having a little bit of a tougher time beating Sim than I normally do. My Hawk was a little rusty but I think some players are figuring out that match more.
If I had to give a number I would normally say it’s in Hawk’s favor 5.5-4.5 (Yes, I like fractions!) but I’ll just call it even: 5-5.
Note: I haven’t done this match offline so for those of you that say playing online is a factor in this match, keep that in mind. I personally don’t think it would make a difference.
I see your point ZAS (not Zass), but I would like to politely disagree. An inch (when applied to human height) is a substantial length. Now, to take it to quarter inches or eighths would be overkill. Similarly, I think including fractions is a bit overkill. It goes into the nuances of a match that really are ancillary to the necessary information. I mean, are there really two characters, when ordered into a tier, that are separated by half a point?
Street Fighter is a game of pixels and frames: nuances.
The ranking system as it is now (taking into account how the scale is used) when configured to whole numbers boils down for 99% of matches to:
loses badly
loses
even
wins
wins strongly
A 5-point scale, which wastes 1 point for even, and wastes 2 more points because the remaining points mirror each other with simply the order of the characters in the matchup swapped.
Leaving us with 2 points to indicate degree of matchup imbalance in uneven matches.
To do this does not convey information relative to the nuances of the differences of the matchups for me.
And WHY has it been expressed that people want the fractions removed and the nuances lost?
Pretentious.
Overkill.
I don’t have time for people who are driven to keep things pretty, simple, and tidy except when communicating at a novice level.
If fractions weren’t material, then people would not be suggesting them.
I am gonna keep using fractions, to whatever degree I feel is justified in the matchups.
If I get charts with whole numbers, I am gonna regard them as crude data.
Major: Business Economics (emphasis in accounting)
Minor: History
Because we are using the scale as a “# matches win”, we end up not using most the scale.
Why?
Because (according to players in Thelo’s chart) 99% of the matchups fall between 3 to 7 (3-7, 4-6, 5-5, 6-5, or 7-3), so (like my height example for heights like 1 foot) we end up not using part of the scale. (10-0, 9-1, 1-9, and 0-10 are not used; and 8-2 and 2-8 are only used in 1% of the matchups)
That’s how it gets reduced to 5 data points within which almost all the matchups are defined.
If I have a height scale to 100 ft, and people cap out at 8’, then all data points from 9-100 are not helping and for all intents are not real/occurring data points on the scale.
Also, all the non-5 numbers have an opposite counterpart that represents the EXACT same degree of variation from 5… but in the opposite direction.
This means that if you believe Honda beats Dictator, and you subscribe to the convention that he can only beat him by a 6-4 or a 7-4, then you only have 2 ways to specify the degree to which Honda beats Dictator: 6-4 or 7-4.
That’s why I saw the 10 points scale without fractions only gives us 2 choices to use in defining how much advantage the favored character has: 1 point off even or 2 points off even.
IF we can define a character’s matchups in an accurate way that indicates that some are more difficult than others (4.5-5.5 vs. 4-6), then why choose a ranking system that obscures this and instead communicates that the matches are even (4-6 vs. 4-6)?
I lose data.
Q: "What is Dictator hardest matchup?"
A: "Chun Li, it is 6-4 in Chun’s favor."
Q: "Isn’t Ryu vs. Dictator also 6-4 vs. Ryu’s favor?"
A: "Yes, but it’s a tougher 6-4 vs. Chun."
Q: “Obviously.”
When I was taking physics in High School, we learned that the acceleration due to gravity on the Earth is 9.8 meters per second per second.
But there was a basic physics class that rounded that number to 10 in order to make the math easier.
This greatly complicated experiments for the basic physics class, since the math did not match the experiment results… but on the plus side, at least they weren’t being pretentious.
I only hope the 3.14 was whittled down to a modest 3.
Lol, yeah basic courses make the math easier pfft.
What you should do is show by comparison by graphing with and without fractions. If you show them graphically speaking the difference it may be easier to understand, for those that still don’t or want to argue about it, :wgrin:
Also, what if you altered the scale to still reflect out of ten matches, and plotted your ten points between 8 and 2. That way you are using your two most extreme points, and the level of detail between them would be greater, for explaining differences between.
Of course that requires fractions though.
Why does everybody hate fractions? I like fractions.
Or why not just collect data on about 4000 matches, factor in a way to objectively measure skill comparatively across data points and do a multivariate analysis based on binary variables for character with cluster mapping?
I just want to make sure the data set is large enough, besides, I know I can get a large part of the data from WWL matches (of course there’s situation where numerous characters would literally have no entries… where’s the love for Ryu?) but even that would take weeks and weeks… I do agree that a real quantitative analysis would be of some scientific inquiry, I was personally skeptical as to how to do it w/o pre-test bias problems but if you could actually get people to self-identify their talent level correctly according to a very rigid and well defined set of values, it would be possible, then we could turn theory fighter into STATA or SAS fighter and let the best Standard Deviation win…
I was looking into your WWL thing and I was wondering, knowing how exact you are, why don’t you use that bracket maker thing, as well as an IRC channel instead of forum posts?
Your argument is getting so ridiculous zas, I’m running out of ways to reply. If the edge numbers aren’t being used on our 10 point scale, that means the game is approaching balanced. Following your argument, why not use tenths or hundredths? Scale is relative. In your absurd examples, precision is necessary. In measuring matchups based on sets of 10, not so much. I can see that in effect you want to trim the unused portion of the scale, and zoom in on the remainder, at which point the differences become too insignificant to matter. For your records, fine, but as a standard, no. It’s overkill for the majority and makes you look like a washed up mathematician searching for relevance in a kindergarten classroom.
Because we’re working on a website now to do it, but SRK is the place to be and its a convenient location, people have helped by providing webspace for our spreadsheets and stuff but we’re building, I suggest you check out our youtube channel if you want to see some great matches! But that’s completely irrelevant to this thread and I apologize… now back to theory fighter…
You missed his point about matches being “relative” on difficulty.
He made the comparison of two people potentially having the same “6-4” matchup, but the matchups aren’t the same difficulty. So he’s advocating the use of fractions (in .5 divisions) to further clarify difficulty.
“I can see that in effect you want to trim the unused portion of the scale, and zoom in on the remainder, at which point the differences become too insignificant to matter.”
And then you have two 5.5s. And then two 5.25s. When does it become insignificant? After the decimal point.
Exactly…and I’m still having trouble understanding why people have a problem with this. Statistics are supposed to be as exact as possible. If we have the ability to narrow things down in a way that doesn’t completely muddy the waters, why not use it?
When I first joined the website after my fighting game hiatus, using half decimals to describe matchups was foreign to me because, to my recollection, that wasn’t done back in the alt.games.sf2 days. But, I quickly understood that using whole numbers to differentiate certain matchups was necessary. I find that it’s especially necessary when you’re talking about the same character having the same matchup against two different characters. That’s the time when you’d really need to specify any nuances that may make one matchup more difficult than the other.
With what I’ve seen do far, I think this has been done with reasonable clarity via a half decimal system and I think that it can continue to be done.
I agree with damdai about not going into quarter decimals, though, simply because we’re not going to have enough data to justify making those disctinctions. I mean…every game and sport stops recording averages at a certain decimal place. For this particular brand of game, I think one decimal place is as far as it should go. The key is to stop recording when the difference isn’t readily discernible. I think that’s why National Football League stats rarely ever go beyond one decimal place, while Professional Bowling Association stats can go to two or even four decimal places. It depends on how much minutia (I hope my spelling bee experience paid off there) the game has.
I believe that in this game, there is an interesting difference between a 5.5 and a 6, but that there is none between a 5.5 and a 5.75. Half-points are the closest level of precision that I think still mean something. So that’s why I like half-points here.
If we really, really wanted to not have decimals, we could always see a 5.5-4.5 as a 11-9, hah. No decimals! :clown:
So that’s what it comes down to. Whether or not we need a 10 or 20 point system to convey matchup difficulty. Given the subjective nature and history of matchup charts, why don’t we try to establish a whole number consensus before we attempt to break it down further, at which point 99% of us can walk away with a yawn.
