lichess.org
Donate

My opinion of the new rating system.

Also, @Cynosure, it looks like Standard is just the weighted average of Bullet, Blitz, and Standard, and the pool ratings aren't part of the weighted average, but are counted as whichever of Bullet, Blitz, and Standard they would fall in.

I've checked a couple players' pages before and after 1 0 games, and both their 1 0 pool and Bullet game counts incremented by 1.

So avoiding the pools wouldn't really help, as it turns out, but sticking to your strongest time control certainly would.

Now that I realize it's a weighted average, I agree that it doesn't make so much sense. Either just removing it, and letting Bullet, Blitz, and Classical speak for themselves or changing it to behave like Global did would make the most sense to me.

Every other guessing seem just fine to me ;)
@51 Thanks for checking that out - so to clarify, if you played a bullet game of 1+0, and won, you'd get +1 bullet time game, +1 pool game (but no rating for the pool)?

Dunno if that's entirely true, as then if you played no pool games, you'd have something like 1500 / "number of 1+0 games" which would drive you lower.

Sticking to your strongest time control is how DontBeMad, Some Patzer, etc, seemingly gained rating points - and why others such as abyssus dropped away...Guess which ones had played more than just bullet?

Maybe just being done with the rating altogether is best.
@Cynosure:

You would get +1 bullet time game, +1 pool game, your bullet rating would change based on the result and you and your opponent's bullet ratings, your 1 0 pool rating would change based on the result and you and your opponent's 1 0 ratings, and your standard rating would change based on how the bullet rating change affected the weighted average.

DontBeMad is a perfect example of what I'm saying. He's played exclusively bullet, and almost all of those bullet games have been 1 0 pool games. You can see that his Standard rating just is his bullet rating, with no dragging downward from his 1 0 pool rating.
I am horribly confused. In two days my rating has gained 200 points for no reason and I now have a "top %1 player" trophy next to my userinfo. I don't know if I deserve this as I believe I'm a 1750 player (which I was before) I think it may just be an inflated rating illusion. Maybe this is how Thibault is getting back at me for criticizing the site "let's inflate his rating so when it drops back to normal and his trophy gets removed he'll think he's getting worse at chess." Haha just kidding Thibault but still, I'm confused about this whole thing and don't think I deserve this high of a rating.
It's because you have 158 classical games at 1933 and 8 Blitz at 1740, and no bullet.

So you effectively have 97% of your rating weighted at 1933, and 3% at 1740, so it averages to 1950.

On the other hand, I lost my 1% trophy and 250 points, so swings and roundabouts?
Bullet's probably harder to get a high rating in. I have a feeling you're the better player than me Cynosure. You can have my 1% trophy because I feel elitist and mean having it, but more importantly, I don't know if this is my real rating or if the trophy is even legit. I'm not sure of anything anymore. I wrote earlier I feel like a bodybuilder who walks into the gym to find a prankster has changed the numbers on all the weights. "Hey I can bench 100 lbs more than normal....oh wait.."
You're way too focussed on that four digit number. Just keep studying the game and have fun playing chess. You act like your life depends on that four digit number. I think it's safe to say it doesn't.

Once you break the 2400 barrier here you can safely say you're a player to be reckoned with. Nobody cares about anyone rated under 2400 anyway. Until then keep studying and playing and pay little attention to that four digit number and trophies handed out on online chess platforms.
@SeaTurtle: That's a natural feeling to have, but the trick to remember is that ratings have never been like that.

Ratings have never been numbers the absolute values of which have any meaning. The way ratings work, only differences matter.

If everyone's ratings went down by 3000 points, or went up by 3000 points, all is the same as far as the rating system goes, since all that matters for predicting results and changing rating is the difference between your rating and your opponent's.

Even in rating systems that aren't just recalculated from scratch, the same thing happens, it just happens more slowly, as rating pools are prone to deflation and inflation as more players above average or below average enter the pool.

Some engine lists try to keep a standard by keeping the rating of one engine fixed at 3000 or some other number, and since engines are just the strength they are and don't improve or deteriorate, that means that the rating of 3000 always corresponds to a fixed strength.

In human chess it's not quite possible to do something like that, unfortunately :)

Something that would be interesting, actually, would be to calculate average error for players in analyzed games in a way like Guid and Bratko, and others have done.

That would give some sort of fixed, objective strength measurement, despite all the issues with that sort of thing, which have all been discusses ad nauseum following Guid and Bratko's work.

Either way, I'm getting far afield. Short version: Ratings are always relative to the other players in the pool, and the absolute numbers don't tie to any objective level of play, so the absolute numbers can't be relied on.

Simplest thing to do is just to notice if your rating is trending up or down. If it's trending up consistently, you're probably improving :)

This topic has been archived and can no longer be replied to.