2 coms win rate

Depends on the size of the team. I won most of the time I have 2 coms (base on recent records). I have issue with elo calculation though, it seems I'm stuck in a specific elo range. I either get straight wins then straight loss or equal wins with equal losses, floating between 2200 to 2300 elo.

Well, time well spent; I looked at 4 different average players recent game results, so that they are not likely to be the people getting the 2nd com and influence the result, and also excluding games where I was involved, and in 56 battles the team with 2 coms lost 2 battles more. So that doesnt tell much.

And psaniac, your recent 33 uneven games resulted to 2 coms team losing 1 more battle.

Maybe its somehow fixed already? But whateverz.

quote: 35 games and 2 com team lost 3 more battles than won.

So 16 vs 19. That's not very good evidence of double com being OP, but it's adequate evidence of the double com mechanism being accurately understood by the balancing system.

quote: And spying on one more person involved in this thread(not me), 75 recent games: 44 wins and 31 losses where 2 coms were involved.

This seems the only one of the people-samples that diverges from roughly 50% chance. Interesting, but attributable to that guy being on a roll at that point, and even then the effect size seems fairly small.

I guess the real takeout here is that double com team is not more likely to lose, it seems.

I think it depend on team size. 5 vs 6/6 vs 7 can be more problematic with two coms then 12 vs 13 or even 15 vs 16. Its possible that smaller team sizes could be larger fail rate with two coms. At least i have seen it. Too much factors to make obstructive statement about two coms.

Hm, i think if we pick all non-Custom teamgames, we can infer the second commander from the teams being uneven with 1 player difference. Then the whole op could be one database query. I'll look into this later.

quote: Feels like this is a balance thing thats not really getting attention it should already get.

I don't remember how much progress was made with it or whether it is in fact already implemented, but there is a plan to put one player in the waiting list when the room is small and the number of people is odd.

So the problem has gotten some attention, but instead of a "game balance" problem it's being treated as a "just don't have these games happen" problem... which seems reasonable to me.

I'd check game sizes 4v3 to 2v1 for win chances. I suspect that as game sizes grow, the difference that one player missing makes on win rate diminishes. I wonder how that would change if 2 comms could place their commanders separately.

Anarchid
how big is the chance that the balancing-algorithm gets a degree in psychology?

Aquanim
maybe you could also make those games count less for the rating-algorithm. I suggested that to DeinFreund originally for games that have very high internal rating-differences in a team. this of course wouldn`t solve the problem of people feeling bad about playing those games at all, but it might be an element to soften the negative feelings. I personally wouldn`t feel so disincentivised to play those games for example.

Is there any meaning to aim for a statistic of 50% win rate? Are players meant to be equal to each other? Doesn't that defeat the purpose of elo? Isn't it more important to make sure elo actually reflect individual level than to create a "balanced" game which also possible by allocating wrong elo?

For example, I have seen one player play about the same level as the other player yet the elo gap between them is 900. Don't make sense. Why would their elo gap be so wide when they play the same?

quote: Is there any meaning to aim for a statistic of 50% win rate? Are players meant to be equal to each other?

Imagine two teams that are perfectly equal in strength, and they play 100 games without changing even one player, but of course adapting to each other and continuously changing strategies.

What should be their win rates?

quote: For example, I have seen one player play about the same level as the other player yet the elo gap between them is 900. Don't make sense. Why would their elo gap be so wide when they play the same?

In the narrow sample where you observed that the two players could well play the same despite the rank gap, it's plausible that for the 900 elo higher it was an unfamiliar or just unfavorable situation. However, consider that there are many situations and environments in ZK that all influence the "casual" rating. It is likely that your higher-ranking player is significantly better in many of those.

Even the "lobsterpot" room continuously oscillates between modes such as 3v3 and 16v16 - those are entirely different settings with drastically different strategies and skills needed.