Super Sundays -- ST Ranking Battles

-=- OVERVIEW -=-

This is a Super Turbo league hosted on 2DF (www.2dfighter.com) with player ratings that represent an approximation of skill level. The higher the rating, the stronger the player. It starts each Sunday at 6 p.m. Pacific / 9 p.m. Eastern and lasts 2 hours.

In to the handicap setup used, you usually race to a different number of game wins depending on each player’s rating, so one player might need 5 game victories to win the match while the other might need only 2. (The emulator automatically keep track of the number of wins for each player.) Ratings are adjusted by +25 points for the winner and -25 for the loser.

You’re allowed to play whoever and whenever you like. This means you can show up late, leave early, miss weeks entirely, play only opponents that you know you have a good connection with, or whatever. And because the ratings are based on a version of the Elo system, you can’t “farm” rating points, because playing a lot of games doesn’t give you an advantage. (Unless it helps you learn to play more skillfully!)


-=- CURRENT RATINGS -=-

See the current player rankings. There are too many to list on one page, so click the numbers at the bottom to see different pages of player listings.

If a player has been inactive for a long time, s/he may be removed from the active rating list. Any such player may begin playing again anytime with their old rating.


-=- SETTING UP ON 2DF -=-

All you need to do is get on 2DF and jump in the Super Sundays room at the appropriate time. Be sure to put the Mode menu option (at the top-left) to Private.

Here’s the stuff you’ll need to play on 2DF:
– Windows operating system
Microsoft .NET 3.5 (or 3.0 works too, if you have that already)
Create a 2DF account
– Super Turbo game ROMs, World version. (I can NOT tell you where to find these, sorry.)
– Port 27886 UDP open and forwarded if you want to be able to host games and challenge people. (You can still be challenged if this isn’t open.) --> Port-forwarding thread
– (If using FireFox) FFClickOnce add-on


-=- JOINING THE LEAGUE -=-

Why do I need a starting rating? With the handicap system in place, it’s important that each player’s rating accurately reflect their ability so that matches are fair. We don’t want any “ringers” (rated way too low) or “lame ducks” (rated way too high). That’s why we can’t just throw people on randomly with any old rating.

How do I get a starting rating? You get one by playing a few exhibition games against one of the trusted members of the Super Sundays community so that they can make an estimation of your skill level. If you haven’t bumped into one of us in casual games ahead of time, you can just show up at the event and say you need help getting a starting rating. We try to be available 15 minutes or so before Super Sundays starts to take care of this.

People with the ability to give starting ratings include Raisin, Little Mac, damdai, brian, JMS, djfrijoles, rashreflection, krost, and possibly others.

We’ll try to estimate your initial rating as accurately as we can. Don’t worry if you think it’s slightly off at first; your rating will quickly self-correct itself and gravitate to your true level as you play ranked matches.


-=- LEAGUE RULES -=-
[list][]Game Speed: Set to Turbo 2. (F2 = default service menu key, submenu #3 = system configuration, game speed = Turbo 2)
[
]Who Picks First: Higher-ranked player has to choose character first for the initial game.
[]Character Switching: The loser of a game is allowed to change characters between games. The winner has to stick with the same character, at least until he himself loses.
[
]Banned Character: Akuma is a no-no.
[]3P and 3K Buttons: These control settings are allowed.
[
]Play It Out: Once a game starts, it’s on for good. Play through any lag or stuttering, and type at your own risk.
[]Cancelling Matches: After being started, a match can only be cancelled if both players agree to it for a good reason (really bad connection, one player has an emergency come up, etc.) Otherwise, leaving an unfinished match will result in a forfeit. If you’re unsure about the quality of the connection with someone, have someone mess around in 1P mode against the computer for a minute to gauge it.
[
]Freezes: If a match freezes mid-game, the players are to replay that individual game no matter who appeared to be winning at the time of the freeze. Any finished games that happened before freeze still count towards match victory.
[*]Sportsmanship: Players are expected to display sportsmanship. Joking between friends is one thing, but things like excessive taunting or deriding your opponent aren’t allowed.[/list]

This setup is provided by and for the ST players’ community, and they are expected to conduct themselves with honor. Don’t smurf, throw matches, or cherry-pick opponents with the intention of artificially inflating your rating. In the case of a dispute, talk to Raisin, Little Mac, or damdai for arbitration. We reserve the right to modify the rules as we see fit.


-=- HOW THE RATINGS AND MATCH WIN REQUIREMENTS WORK -=-

Feel free to skip this section. You do NOT have to memorize or even understand this list to participate in Super Sundays. It’s simply provided for those who are curious. Remember that ratings are based on the Elo system for the most part. The important difference is that…

In normal Elo:
–The competition itself is balanced. (Win the chess game, outscore the other team in a soccer/football match, etc.)
–The ratings adjustments are handicapped. (A 3:1 favorite might risk 30 rating points if it lost and only win 10 points if it won. And that 1:3 underdog would only lose 10 rating points if it lost and stand to gain 30 points if it won.)

While in Super Sundays:
–The competition itself is handicapped. (A 3:1 favorite needs 3 times as many game wins to win the match.)
–The ratings adjustments are balanced. (Each side always stands to win or lose the same amount [25] regardless of the outcome.)

Chart format is: (difference in rating) – (Elo’s expected winning percentage of games by the higher-rated player) – (Listing of games required to win the match)

Note that: (higher-rated player’s goal number of victories) / (sum of higher-rated and lower-rated players’ goal victories) = as reasonably close to Elo’s expected winning percentage as possible.

000 – 50% – Both players need 4 victories
025 – 54% – Higher-rated player needs 5 victories, lower-rated player needs 4 victories
050 – 57% – Higher-rated player needs 4 victories, lower-rated player needs 3 victories
075 – 61% – Higher-rated player needs 6 victories, lower-rated player needs 4 victories
100 – 64% – Higher-rated player needs 7 victories, lower-rated player needs 4 victories
125 – 67% – Higher-rated player needs 6 victories, lower-rated player needs 3 victories
150 – 70% – Higher-rated player needs 7 victories, lower-rated player needs 3 victories
175 – 73% – Higher-rated player needs 5 victories, lower-rated player needs 2 victories
200 – 76% – Higher-rated player needs 6 victories, lower-rated player needs 2 victories
225 – 79% – Higher-rated player needs 7 victories, lower-rated player needs 2 victories
250 – 81% – Higher-rated player needs 8 victories, lower-rated player needs 2 victories
275 – 83% – Higher-rated player needs 10 victories, lower-rated player needs 2 victories
300 – 85% – Higher-rated player needs 6 victories, lower-rated player needs 1 victory
325 – 87% – Higher-rated player needs 7 victories, lower-rated player needs 1 victory
350 – 88% – Higher-rated player needs 8 victories, lower-rated player needs 1 victory
375 – 90% – Higher-rated player needs 9 victories, lower-rated player needs 1 victory
400 – 91% – Higher-rated player needs 10 victories, lower-rated player needs 1 victory

What’s this list mean? Well, the important figure is the first column, the difference in player ratings. You subtract the lower-rated player’s rating from the higher-rated player’s. If Joe (rated 1200) wants to play a match against Corey (1200), they have the same rating, so their ratings difference is 0. Now if Cathy (1750) wants to play Mike (1625), the ratings difference for that match is 125.

Now you take that ratings difference and look at the chart. Say two players have a ratings difference of 0. Elo predicts that each player will win about 50% of the games in the long run. Since they’re expected to split games, they play to the same goal number of wins when they have a match.

Suppose a very slight ratings difference of 25. On the chart, you can see that Elo predicts that the higher-rated player will win about 54% of the time. So the number of games required to win reflects this. The higher-rated player needs 5 wins before the lower-rated player gets 4. If you do some quick math, you can see that 5/9 is 55.5%, which is in line with what Elo expects.

Now assume a ratings difference of 150. Elo’s expected winning percentage for the higher-rated player is 70%. Here, the higher-rated player needs 7 wins and the lower-rated player needs 3. Checking the math, you can see that 7/10 lines up perfectly at 70%.

In Super Sundays, 400 is the largest allowable difference for players to play a ranked battle. It is possible to accurately reflect matches with bigger ratings discrepancies, but the problem is that they become so lopsided that their length isn’t very practical. (For example, a rating gorge of 600 means the stronger player is expected to win 97% of the time, and requiring the stronger player to win 33 times to the weaker player’s 1 is going to lead to some marathon matches.)

(super-duper old and dated post no longer relevant)

T H A N K Y O U ! ! !

For so many reasons. I’m forever begging for matches on GW and it is so hard to find a match
sometimes. Also for sacrificing yourself and being an admin instead of playing.
And for taking into account the hour and being able to play when you want.
I can’t give you enough praise yo.

Now tell me…

  1. I’m a turbo 3 speed player. If me and my opponent agree to it can we play on 3 speed?

  2. I think the ranking is a lil lop sided in favor of the lesser skilled player. Isn’t it like a fake rank (inflated)

  3. I’ll pm you the rest lol

I think this system is a great idea.

I like the fact that you can play against anyone and have a meaningful, intense match to decide the ranking change.

I also like that its restricted to p2p and that its on turbo2. (my impression is that default turbo2 is actually the same as free-select turbo3… correct me if I’m wrong on that)

dj I dont see why you would say that the system is lopsided in favor of lesser players- the ultimate rank ordering of players is not really influenced by the system they’re using. The points gap between players might not be the same as in a normal elo system, but I dont see why that matters.

Sounds like fun.

If Brian is doing it then I wanna do it too

Macumazahn who are you?

btw its worth pointing out that this system does have its flaws. For example, if there’s someone who plays a character that counters your main character, you could just avoid playing them, unlike in a tournament. Also, it seems hard to make the required wins ratio just the right number depending on the points difference.

But the main thing is that it sounds like a fun system to play in, for all the reasons mentioned in the OP!

We played a few times on GW, it’s good to see decent players are willing to do this, that’s all I’m saying

Sure thing, updated the rules. I just think that in general, it would be good for the game if there was a standardized speed. I don’t even particularly care what that speed happens to be as long as everyone agrees on it. Turbo 2 seems to be the most popular, and I guess you could argue that it’s kind of in the middle and a sort of compromise over game speed and not something that would take a ton of adjusting to if you were used to something else. So I went with Turbo 2. But if you and your opponent agree to play on something else, go for it.

If you mean that it handicaps things against the higher-ranked players, you’re completely right, and that’s actually part of the idea.

Think of it this way… Imagine a situation where there were no built-in handicaps. Let’s say players just played best-of-7 and won or lost 25 rating points depending on the outcome. What would happen is that strong players would be encouraged to avoid each other (why risk points?) and prey upon the weaker players all the time (easy victories). Meanwhile, you have an extremely frustrating situation for the weaker players, because they keep getting challenged and pounded by the better ones all the time.

There is also one final problem with this setup in that time investment starts to be important. If a good player wins 75% of the time overall, he’s going gain the highest ranking in this system just by playing tons and tons of games to pile up points. The poor guy who’s just as skilled but ended up being able to play less (he had to work some weekends and attend some stuff in real life) ends up with a lower ranking just because he had less free time.

In a handicapped environment, these problems disappear. Imagine a moderately-strong player in a handicapped environment like Super Sundays. He can either challenge a stronger opponent and have the win conditions be easier, take on an opponent of equal skill and be on equal footing, or take on a lesser-skilled player but be required to win more often. In any of these scenarios, the risk/reward is weighted, so his chances of success should be equal no matter which he picks. There’s no more preying on lesser players and dodging of higher ones. It encourages player freedom to challenge whomever they want.

Well, it’s my intention that the relative difference in ratings should be standard Elo. In other words, a difference of 200 points means the better player should win about 3/4 of the games. Not 3/4 of the matches (which should always be 50-50 in this system if each player’s ranking is accurate), but 3/4 of the individual games. A difference of 125 points means the better player should win about 2/3 of the games, etc. I’ll update the first post with some info on this soon when I get more time.

Yeah, good point, and that’s part of why I included the part about cherry-picking your opposition. Thing is, I don’t really know how to solve this problem without compromising players’ ability to freely come and go, and I’m hoping it won’t be much of a problem anyway. Character-switching during matches is allowed in any case, so let’s see what happens.

I dont think you should worry about the little problems much… it sounds like a fun system. I really like the handicap, because it will make playing against weaker players challenging and fun, instead of it just being boring.

Very good point.

To answer your first post toward me… I didn’t (dont) really understand the point system embarrassed but I’ll take your guys word for it. I definitely wont be counter picking any one on purpose because you can put me down as using Gief from here. This is gonna be a lot of fun!!!

only thing I would suggest is an alternate emulator if you want things to go smoothly. with .64 you’ll have the looming threat of random crashing.

In case it wasn’t clear, Super Sundays does NOT start tonight. :looney: It starts next week.

For the future, I have a plan for how I’m going to add new players to the ladder. If the person is a regular of the ST community, it’s probably easy enough just to make an educated guess about their skill level based on their past performances. For relatively new or unknown players, there’s probably going to be a provisional rating system. Details to be decided and explained later. For now, if you’re a regular player who plans on participating, you can join by PM’ing me some of your win percentages as explained in one of my earlier posts. I’d actually encourage you to do it ahead of time, since it needs to be done sometime anyway, and you may as well get it out of the way ahead of time. :tup:

I wouldn’t either if I wasn’t familiar with it. The Elo rating system (which is not exactly what we’re using, but close enough) can seem confusing if you’re not used to it. I’m pretty confident in the long run it’ll work out great, though.

There seems to be a lot of unclarity in general about how the rating system works, so I’ll write more about how the ratings work soon.

OK, I modified the rules a bit so that you can play on any emulator if both players agree to it. As far changing off of .64 as the default, I feel I could only do that if there was a big majority of people who agreed to use something else.

ill do this

edit: are there any optimal setting for minimizing input lag?

Sounds cool.

why mame32k 64 as default? why not mame32++ .117

I was hoping no one would say this because my comp. can’t handle 0.117

If you guys change to 0.117 I wont be able to participate. =(:sweat::sweat::sweat:

edit

now Kawaks 1.58 is another story. I actually prefer it.

add me
time to kill someone with deejay sagat and guile

all good… i spoke to the catz team the other night on westcoas wonderland and made a lot of points regarding the different versions. although i prefer ++.117, they tell me 32k .64 is best because not everyone has a good computer. recommended requirements were like at least pentium 4 2ghz and geforce 4. i dont think i will be participating yet cause my stick isn’t complete. :sweat: whoever’s played me knows i cannot pull any moves off my logitech dual action pad :sweat:

This would be awesome if done aswell for EU peeps, What if I would run EU ST Ranking battles with your concept ? Would you mind/like/approve of that?

Also wouldn’t it be better to have everyone start out with a set amount of points ? like 1200 ?

stop saying bullshits, ma nizzle :slight_smile: