But different stats is something that all calculate before game where it show balance %.
Yes. However, it's using trueskill for this. It's taking all the individual players in one team and comparing to the other, not lumping them all together as one like ELO would.
When all can lose/win the same mount then team would be on one boat, it is more clear for what are fighting for.
Why is this preferable? If you have a higher deviation then you should lose more/gain more, or it makes the whole system less accurate. That's one of the advantages trueskill has over ELO.
when it is separate rating for each player it can lead to low player plaing with high player would stuck on low rating because will get only litle bit from bet but will probably play(and rise) better as low rated face only low rated.
Actually, it's the exact opposite.
In ELO, A low rated player and a high rated player would get lumped in as medium vs another lumped together team. In this scenario, it's actually more likely for the lower rated player to gain rating slower. Trueskill would take these games into account and increase deviation, which in turn allows the lower rated player's rating to increase faster (if he continues winning).
I was trying to say that on every rated game is some bet what is same for bout side (differences about balance of course).
Like 16 point and if one team win 9other team lose 7and together it is still same mount.
Already on true skill it dont work, one gain 6 other lose 2 and balance is cca same true skill work with much more statistic level and no one know of what are fighting for.
Again, this is a case of trueskill just being a more accurate system than ELO. It makes sense for people to gain/lose differing amounts of rating, and arbitrarily changing that won't help at all.
because truskill is not ideal for ranking, that is answer on next question too. True skill is primary for balance game, it have they own mechanism what are probably very accurate on infinity amount of people, but unaccurate on some concret situation.
for example i was play extreme chees with freind 1v1 balance was 60% and i lose 42 rating for 1 lose game. When i win against him next game i win 4 point. In finally i must 10time win againt him for posible to one time lose. Or i must win cca 10 well balance team game in row for one lose against 1400. Or he can lose in raw about 7 game for one win in 1v1. 1v1 is about 4time more rated as team games, and much more on balance differences. So that i want to say is that trueskill is not a good for use as rating because on this way its not fair for so big differences of point distribution.
I do not understand your problem.
I'm not sure of exact numbers of how much rating you'd lose, or gain, but really, it makes sense for you to lose a large amount of rating if you lose an unbalanced game.
If I had a 90 % chance of winning a game, I'd expect that one lost game to take away as many points as I'd gained from 9 won games (of course, it's done statistically, but my point stands).
In terms of 1v1 being more "rated" than team games : It's not. It's just that the system takes into account that it's just you playing, and not you and a team, which is why it affects your statistics more, and so you gain/lose more points. Again, an example of a more accurate system than ELO.
In terms of ranking,elo or trueskill rate players, they don't rank them. There is obviously a rank you can use from the rating : the player with the most points can be considered rank 1, and really no rating system is better than any other for this apart from the accuracy of the rating itself... which leads us back to using trueskill
Statistics: Posted by Gorton — 19 Nov 2014, 16:33
]]>