Feature request: "purchasing power of a rating"

About the rating, the ELO or its variants are only as valid as the playing pool is constant. Otherwise it is subject to inflation/deflation. Besides the percintile of players, it would be nice to have an accuracy associated with the various ratings, ie purchasing power.

So a feature request: somehow show that say given the games of last month, a rating accuracy is X, and that it is different, given the active playing pool, than it was a month ago.

It is really frustrating to get opponents of a rating that all of a sudden means nothing, given that I face a difficulty level -- if, and that is a big if, if the profiles are human and honest, not smurfing nor hansing -- so a difficulty level I have in mind for 2100 not 1550, and indeed the CPLs do say that much.

Just looking at how I fail to improve I wouldn't be so quick to scream inflation!!!!!

Adding frature requests (regardless of how sesible these are) and deleting your account afterwards, doesn't seem like a good strategy at all...

Have fun anyway.

@dogmatic42 said in #1:
> About the rating, the ELO or its variants are only as valid as the playing pool is constant. Otherwise it is subject to inflation/deflation. Besides the percintile of players, it would be nice to have an accuracy associated with the various ratings, ie purchasing power.
>
> So a feature request: somehow show that say given the games of last month, a rating accuracy is X, and that it is different, given the active playing pool, than it was a month ago.
>
> It is really frustrating to get opponents of a rating that all of a sudden means nothing, given that I face a difficulty level -- if, and that is a big if, if the profiles are human and honest, not smurfing nor hansing -- so a difficulty level I have in mind for 2100 not 1550, and indeed the CPLs do say that much.

I don't know what this means and I am afraid to ask. Yes, there is value in having an accurate rating system so players can get good pairings.

One could argue that a good rating system should predict outcomes of future games, then that predictive accuracy should be used to change the rating system. That argument, while logically invalid, may have some merit.

It is an interesting topic actually. I doubt the pool changes timewise so dramatically, suddenly and at a clear cut moment so one can calculate different pools per month, as OP suggests, but I would expect there are real clusters of players, who play more often against each other, forming different sub-pools depending on other criteria.

I've had similar experience playing chess on another site several years ago. I started noticing that when playing against people with same rating, but some coming from US/Canada and others from Russia, I would definitely feel those from Russia significantly stronger than those from North America, even though they had the same rating. So back then I came up with a similar theory, that maybe there are clusters of people who more often play against each other, which makes sense, given different timezones and assuming people all around the world still play in similar local hours so more likely to play against players nearby.

Wonder what other clusters there might be except geographical - e.g.:
- people more likely to play on weekends vs. workdays
- mornings vs. lunchtime vs. evenings
- arenas vs. lobby games
- inc vs no-inc games

and how lets say morning players compare with evening players with similar rating for example