Profile of MilkaCow

ELO is a mathematically sound and proven way to measure a persons performance over time. It's nowadays used in almost all competitive games in some way or another. The standard algorithm can be found here:
http://en.wikipedia.org/wiki/Elo_rating_system#Theory

No idea how exactly Relic implemented it. The standard algorithm is obviously to a degree flawed here, as CoH2 does not have the ability to end in a draw. It still works of course, but it means it's less stable.

Now ELO is just one part of the system. For a good matchmaking you need to 1.) be able to accurately judge the performance of a player and 2.) be able to match them against similar opponents.

For the pure ELO it doesn't matter too much what you get matched against, but of course for the player experience it. You could be matched against quite a lot stronger players and only win one out of 5 games, but still stay at the same ELO. You lose little for a loss, but gain a lot for a win.

Now the big problem is the matchmaking, but that's not a problem that's easy to fix. If the pool of players is small, the game has to match you against players that differ a lot from your skill. So even a perfect ELO system can still create a bad user experience when the player pool is so small.

Oh hey - http://www.coh2.org/topic/25716/just-saw-99-axis-vs.-1-allied
You are one of that guys who rages about stuff they don't understand. Well, I guess you won't read & understand what I wrote anyways. Since I already put the time in I'll still post...

To confirm what others are saying: you have ELO tracked seperately for each faction in each game size. Your Ostheer ranking (for example) has no bearing on how you are matchmade as Soviets.

Actually even more.

For random team (aka 'solo queue' or RT): (1v1/2v2/3v3/4v4)x(Sov/USF/Ost/OKW)
For arranged team (aka 'multi queue' or AT): (2v2/3v3/4v4)x(Allied/Axis)

I am not entirely sure how it works if you queue with 2 people, but play 3v3. It's neither a full arranged team, nor a full random team. Having it in the 2v2 AT stats with your mate is as weird as having it as a RT. So it might create a 3v3 AT for you, but this would mean pretty much each game is a seeding game. Same way if you queue with 3 but play 4v4.

So there are at least 16 RT ELO values plus up to 2 (each side) for every arranged team.


For any unset ELO the first 10 games define your initial value. In order to do so you can be matched against any opponent.

http://www.coh2-stats.com/explosive_weapons/schu_mine_mp
http://www.coh2-stats.com/explosive_weapons/tm-35_mine_mp

Exactly the same damage.

As much as I had my problems with Relic lately, they did do a great job with Observer mode. I am also quite happy they now presented that feature. If there ever was something where they used the feedback they got, it was obs mode.
A lot of the features were shown, but just to recap a few:
- Persistent player colors: If you switch between players their colors stay the same. Also, they are the same for every person spectating this games.
- Viewing parties: Syncs automatically for all viewers and the host of the party can accelerate / pause for all. Great for doing casts together, one person streams and all are in a viewing party & Skype call.
- Viewing options: Set the delay, password etc
- Ingame stats: You know the after-action stats you get? You can see them get updated in real time.
- Resizeable Hud: Well, self explanatory

I'm just curious how it's gonna look in 4v4 games, when in 2v2 the players screen takes half the screen

http://www.coh2.org/file/6152/obsmode-preview.jpg

Doesn't really look like half the screen
3v3 and 4v4 were thought about. Don't worry.

I guess when it gets released you'll see for yourself. For me personally I was quite sceptical at first, but the tool has come a long way and it's still in development.

Thanks for the presentation

Bye and have fun down under.