I won a game on the rt ladder and I lost a point >.<

I started with a rating of 1651 and ended up with a rating of 1650. My opponent started with a rating of 529 and ended up with a rating of 530...

I know the ladder is based on trueskill, but it's ridiculous that a player loses a point when he wins a game. I think this should be fixed.

don't you think that has been discussed many times?

Didn't notice the last ones.

Besides that, I don't care that it has been discussed too many times. I think this is ridiculous. It shouldn't be that hard too make sure the points gained when you won a game doesn't become negative.

well, basically Tjoex, the whole community agrees with you. It is dumb.

Ideally the RT ladder should only throw together players within 1000 rating points.

Edited 7/25/2016 22:00:36

I used to agree but I don't anymore :p

True skill mean is more important then ladder rank imo and that one increases each time you win.

Win with that (or simmilary ranked) player a few more times and your rank will grow (and faster then it did before)

The entire community agrees!

Wait... Why has uservoice not been brought up yet?

RT Ladder is to play dank LD and commanders games, not for good rating system.

What Semicedevine said was only posted for discussion not for trolling purposes

Edited 7/27/2016 07:21:47

[OFFTOP]
@Green Turtle
Less then 0.1 sec?
That is impossible for a human being.
That must have been done by a computer.
Don't you consider that cheating?

Also no offend please, I like you, just asking.

@Sultan it's from when you could use autopilot to make your picks super fast.

Edit: Also nice to see everyone criticising the best rating system on any of the ladders, I think half of the forum just likes to complain for the sake of complaining.

Edited 7/27/2016 13:14:36

Oh no! Buns is starting to catch on!

So, for the confused:

The RT ladder uses a ranking system developed by Microsoft called TrueSkill. It's got a very solid statistical foundation and your rating is actually not just one value but two.

Each RT ladder rating has two values- a mean (what it thinks your "real" rating most likely is) and a standard deviation (how sure it is about your rating).

For example, two players rated 1850 might actually have very different ratings- one could be (2000, 50) and another could be (1910, 20). Notice that the "rating" you see is actually your mean - 3 * your standard deviation, so the ladder is actually 99.7% sure that you're better than the rating you see on your profile.

Tjoex won a game, but it was an unexpected victory- so the ladder's less sure about his rating now. Since the published rating is very conservative, it dropped a point. But in the long term, Tjoex's win will still help him- although it's going to matter a lot less if it turns out to be just an outlier.

^

Tjoex won a game, but it was an unexpected victory

Huh? Is a 1651 rated(not trueskill mean) player not expected to beat a player with a rating of 529? I get that this rating doesn't 100% accurately represent their trueskill rating, but surely Tjoex should be expected to win the game by the system?

Just to clarify: I like this system, and I'm not surprised Tjoex shown rating went down from him winning a game against a much lower rated player, but I would think that would be because his standard deviation increased due to his opponent being an outlier in the skill level he usually plays against, not because he isn't expected to win the game?

markus: was a typo by knyte.

I don't think so.
Tjoex's opponent had just huge TrueSkill Standard Deviation so the game didn't expect anything. ;p
Ofcourse saying about expectations is just a metaphorical statement.
I don't know the exact algorithm but I know what standard deviation is. And when you add to the set of your opponents a new one which is very different from the mean your SD will grow. This is what happened.