Are variants "real" chess?

@polylogarithmique said in #40:

@piazzai yeah but what I mean is I believe the rating system is designed in such a way that there is a formula which gives you the expected outcome as a function of the rating difference (at least for the elo rating).
In an "ideal" world, this formula would match the empirical data exactly. Now because elo (and glicko) are based on some mathematical assumptions that are not necessarily verified in the real pool of players, a priori there is a difference between the average outcome of every game where the rating difference is exactly 100 (say), and what is predicted by the formula.
I would be curious to know if there is actually a big difference between them.

I see. I don ́t know much about the rating system and I'm not aware if either Elo or Glicko allows for such a calculation. I believe the main purpose of both systems is simply to rank players from best to worst, and Glicko improves on Elo by factoring in volatility when ratings are updated. They don't seem designed with the intention to give you information about what happens in a future match.

I could not find references to such a formula in the Glicko and Glicko-2 white papers (http://www.glicko.net/glicko/glicko.pdf and http://www.glicko.net/glicko/glicko2.pdf). If you happen to find anything, I'd love to know.

@Rookitiki said in #39:

10 rating points cant be a WR difference of nearly 8%, something mustve surely went wrong there! nearly 20% just because rating differs by 25 points? did you lose a zero or something?

€: sorry i havent exactly read what your model is, i thought were talking about glicko, so nvm if its not.

@piazzai

I am not sure what you mean by WR: my percentages refer to changes in odds ratios. But is it really that surprising that higher-rated players are favored to win?

@cFlour said in #37:

Antichess helps the most in standard chess, if studied deeply. It helped me improve from 1000 to 1400.

Also, you don't even play antichess. How can you make such statements? Just because the rules say something, it doesn't mean that the strategy is that.
This isn't my main account, on my main i've played thousands of antichess games
To your first point I would say that's true in some situations for sure, and when I think about it my standard ratings did improve as well as my antichess ratings from a while ago. I guess I hadn't thought of it from that perspective to be honest, probably because I don't study much theory in anti or standard, I kind of just pick up what I learn from playing.

@piazzai said in #43:

I am not sure what you mean by WR: my percentages refer to changes in odds ratios. But is it really that surprising that higher-rated players are favored to win?

WR = win rate

if were talking lichess/glicko2, then 10 points dont amount to 8%. so if by "odds" you mean winning odds/chance, your math seems off (this is probably just a misunderstanding, thatswhy i am asking what those odds refer to). btw i just read your blog post, but it is still inconclusive for me. if you were talking about glicko2 youd be saying that on average a 1500 vs a 1525 has below 32% winrate (-18%). that obviously cant be true, thatswhy i am asking what those odds are. btw i know what a factor is (and i played alot of poker when i was younger), so you can give me the ratio any which way - again im assuming that were just not talking about the same stuff or i have a misconception about how to convert those odds into winning chances.

@piazzai for Glicko I don't know. For Elo, the details can be found there https://en.m.wikipedia.org/wiki/Elo_rating_system
under section "theory".

@piazzai said in #42:

I see. I don ́t know much about the rating system and I'm not aware if either Elo or Glicko allows for such a calculation.

There is such a formula in this document on glicko ratings. See page 10 and look for the Standard winning expectancy:

http://www.glicko.net/ratings/rating.system.pdf

The number of dislikes for my #2 reply has been increasing incredibly. Why is it so?
Racing Kings-- Doesn't make much sense...just taking your king to the 8th rank. It's simply nuinsance.
Horde -- How do you compel with a disgusting army of so much of pawns?
Antichess -- Exchange off!
Atomic Chess, Crazyhouse -- Yes, they are crazy.
Chess960 -- This makes some sense. They help you in reverse chess planning and arrangement.

I am guessing people disliked because you made a negative comment without backing it up with any reasoning on why. Your comment#48 doesn't really make a big effort in that direction either. But this time you got a like! So its already better.

@Rookitiki said in #45:

WR = win rate

if were talking lichess/glicko2, then 10 points dont amount to 8%. so if by "odds" you mean winning odds/chance, your math seems off (this is probably just a misunderstanding, thatswhy i am asking what those odds refer to). btw i just read your blog post, but it is still inconclusive for me. if you were talking about glicko2 youd be saying that on average a 1500 vs a 1525 has below 32% winrate (-18%). that obviously cant be true, thatswhy i am asking what those odds are. btw i know what a factor is (and i played alot of poker when i was younger), so you can give me the ratio any which way - again im assuming that were just not talking about the same stuff or i have a misconception about how to convert those odds into winning chances.

Yes, this is a misunderstanding. I believe the mistake is that you are reading about odds ratios but thinking in terms of probabilities. As I mention in the post, odds ratios are a bit tricky to interpret: if they increase or decrease, probabilities increase or decrease, but they are not the same.

As you are interested in probabilities, I used the model to predict the probability of winning in two different games: (1), where the player has a rating of 1500 and the opponent of 1500, and (2) where the player has a rating of 1500 and the opponent of 1525. In both cases, I assume that the player is playing white, that s/he has played 20 games over the past week, that 100% of these games were with the same time controls as the game whose outcome is being predicted, and that 100% of them were in standard chess. Note that I necessarily must fix the value of these other variables in order to give you a probability. It is not necessary to do so if we talk about odds ratios.

The probability of winning in (1) is 45.6%. It is not 50% because of the possibility of a draw. The probability of winning in (2) is 40.9%. If game (1) is repeated 1000 times, the player is expected to win 456 games. The remaining 544 will be drawn or lost. So the odds of winning are 456/544, or 0.838. If game (2) is repeated 1000 times, the player is expected to win 409 games. The remaining 591 will be drawn or lost. So the odds of winning are 409/591, or 0.692. The value of 0.692/0.838 is 0.82, or -18%, which is the odds ratio.