KAoS 2010 Data and Statistics

[RoboDestroyer]

... Not counting dropouts? I guess technically winddragon won KAoS then! Congrats winddragon!

/endsarcasm

[RDFMASTER]

... Not counting dropouts?

Let's see next time we meet :P

From 13th to 3rd/4th in the second AHP tourney. Not bad at all xD


We need more OK, Emp, and Shantotto.

[Veysey]

First up, thanks everyone. I'm just happy to help (and play with numbers XD).

I completely agree that we need more accurate data Suppi, and as such, I did design everything so that I can extract separate tourney data (like this) but also everything can be compiled collectively. So after a few tournies, we'll start seeing the data take a more reasonable shape and better represent what's happening. Though if the community wants to do something like what you mentioned, it won't be hard to incorporate the data - I'd be concerned about the seriousness behind some players in such an atmosphere though. I think it'd need to be carefully executed.

As for the battle points, I actually wanted the communities feedback on that since I feel its a little TOO lopsided for bracket winners. For those who are curious (like Draec) currently it breaks down like this:

Per round, you get 50 for fighting someone equal rank and if you are ranked 50+ higher than your opponent, you get 100. If you are ranked 50+ lower than your opponent, you get 1. The difference isn't linear, it's on an exp bell curve type of approach thats designed to have a larger spread as opponents get closer in rank (so one rank difference from say rank 49 to 50 versus 48 to 50 might mean an extra 6 points while 49 to 90 versus 48 to 90 might mean an extra 1 point). That's the most complicated part.

This is what you get for wins in certain situations
Normal Round Winners Bracket/Starting Rounds 5
Quarter-Finals Winners Bracket 150
Semi-Finals Winners Bracket 250
Finals Winners Bracket 500
Normal Round Losers Bracket 0
Quarter-Finals Losers Bracket 50
Semi-Finals Losers Bracket 75
Finals Losers Bracket 100
Grand Finals 250

Grand finals is like that not only because it can take a long time, but there is someone competing who was in the losers bracket.

If you perfect (no loses in a round) then you get 25 bonus. 50 for finals and grand finals since they're best of 5.

Add all those together for skill experience (based on wins)

Now you get 3 exp for each match you participate in as played experience.

Take 30% of the skill and 20% of the play for the final values. They're modified this way to reign in the values for appearance and weight and then added together for battle points.

Now what you see is actually a temporary placeholder though. I'm considering taking away the quarter/semi/final etc bonuses and implementing a skill exp modifier based on the round you're participating in and the bracket. I thought this would keep the values more in check and give a better representation of where people place with less bias towards finals participants. Before going through with it though, I wanted to see if the community had anything to say on it all and get some feedback on the system and how it might be used (the need will dictate how it's presented).

[RoboDestroyer]

Let's see next time we meet :P

From 13th to 3rd/4th in the second AHP tourney. Not bad at all xD


We need more OK, Emp, and Shantotto.

Love your quote RDF. It made perfect sense

[melveys]

Great job! I know you put a lot of work and effort into this and it really turned out great!

[Draec Khin]

Thanks for the info Veysey. I was thinking that maybe we can somehow incorporate Elo rating system (http://en.wikipedia.org/wiki/Elo_rating_system) which is used for chess and other computer games. It is a lot more complex but this system was in use for a long time and had a lot of phd mathematicians improve it. If it is too complex and much of burden I guess we can find something a lot simpler which is in use or make something up. So far the one you have seems okay but I think it like you said puts too much weight in the last rounds.

[Bloodsword83]

If we could integrate the ELO, that would be interesting. But I think I like Veysey's system, too.

What is the meaning of the Matchup Strength data?

[Repliica]

Oh wow, this is awesome.. Thanks for this Veysey, I'm mainly interested in Squall buaha.. awesome.. he had a pretty decent win average. <3

[Veysey]

If we could integrate the ELO, that would be interesting. But I think I like Veysey's system, too.

What is the meaning of the Matchup Strength data?

Matchup Strength shows us who has the best overall odds against the Dissidia cast in general. Basically, it just takes the matchup odds from the 21 opponents and averages them out, showing us who is strongest/has the fewest weak matchups.

As for my system, I actually personally think it can be improved on and I'm in the process of drafting up a few attempts. I'll be posting the results somewhere once I'm finished and seeing what the community prefers. Currently, I have two methods finished and almost done the Elo ... whatever everyone wants is fine with me since the systems are already basically finished. Thanks for the information Draec! Personally, I like the concept behind Elo the best and think it would work best since it's fairly versatile.

[Stupid Aquarius]

Dude, wat. I usually would never say this, but if I had the chance to bear Veysey's kids if just to continue the line of Amazing that Veysey is, I totally would. In a heartbeat.

Are you also gonna take in the data from the xlink Kai tournaments? More specifically DNC.