[Suggestion] Reset Ranked Standing

What do people think about keeping ratings hidden? They are basically just for decoration beyond the matchmaking system. Maybe you could retain the icon from your highest rating if people still care about that.

 

Only the icon from highest rating wouldn't be enough. Maybe highest achieved rating would be better. Or average rating over last 100 games? Then you don't have to worry about losing a lot of rating in 1-2 games.

 

I think that's precisely what he was suggesting.

 

One of problems in ranking is -> when you get paired with lower ELO player.. you can also lose.. what, he gets few OBs or something.. its not like you play chess ..

and for example .. I just won a fight over 1200ELO player .. I got 3 points .. had I lost, I would have lost what, 30-40 points? (I'm @1603)
Ok, dont give us more that 3 points in that situation .. but, dont take 30 points if we lose..

Does anyone know the math beside that +-ELO ?

 

I've written down all my ratings after every ranked game I've played and the rating for each of my opponents before the game. Some things I've observed. It seems that:
*You can't be 100.0% sure what your elo will be after a win/loss when only looking at one number: your_rating - opp's_rating. For example I've won having 54 points higher rating 6 times: 4 times I got +14 and 2 times I got +13.
*If you win x rating points, opponent will lose x points unless he's a very new player. Their ratings are calculated in a different way.
*If you win x rating points, you would have lost (32 - x) points.
*If you win, you win at least 1 point (has happened to me 5 times, winning exactly 1).
A small code I have that uses my data suggests that 1603 vs 1200 would be +3 or -29. Though I'm quite positive that I have some small typos in my data.

And as I've said 1000 times beofre: the problem in CH is not in elo system but in small amount of (higher-rated) players.

 

If I had to hazard a guess it would be that people's actual ratings are not whole numbers but what we see are rounded versions of the stored data. This would explain the difference if the equation for +/- uses the un-rounded form of each player's elo.

 

I was thinking the same, might well be.