Congratulations
I Finally Got Out of 100 ELO!
Sort:
tiffanyperera2 wrote:
YAYYYY
YIPPEEE
Sep 25, 2024
0
#24
basketstorm wrote:
Wrong. Rating of bots (engines) is the correct FIDE rating. Bots are well calibrated. Chess.com uses Komodo. Written by GM Larry Kaufman and others. Most engines including Komodo were rated not arbitrarily but precisely within the pool in which actual chess players and grandmasters competed and FIDE rating of those players was used to get proper relation to strength of humans.
Weaker bots were scaled down from stronger bots and their rating is not arbitrary either, it's all statistically proven.
Source? Strong bots get rated like that but I don't think Chess.com rates their bots that way.
GooseChess wrote:
basketstorm wrote:
Wrong. Rating of bots (engines) is the correct FIDE rating. Bots are well calibrated. Chess.com uses Komodo. Written by GM Larry Kaufman and others. Most engines including Komodo were rated not arbitrarily but precisely within the pool in which actual chess players and grandmasters competed and FIDE rating of those players was used to get proper relation to strength of humans.
Weaker bots were scaled down from stronger bots and their rating is not arbitrary either, it's all statistically proven.
Source? Strong bots get rated like that but I don't think Chess.com rates their bots that way.
Well, it's just what you think. But in fact Komodo was incepted long before it was acquired by chess.com. Chess.com didn't "rate" it.
Read Arpad Elo (creator of Elo rating system) book to understand what Elo means and you will see that it's trivial to rate unrated weak bots by making them compete against rated strong bot or player. You might need 2 or 3 transitional weakened bots to create the whole palette. Basically you run tournaments and adjust engine settings until win-loss ratio matches the desired Elo difference. It's all just math.
Two separate engines scaled down in strength to 1400 in completely separate efforts compete against each other with apparent equal strength. That proves the point.
It depends on how often can you beat that 1800 bot. If that's roughly 50% of games, consider 1800 is your rating as well.
Lol, this is what I mean by "lying to make people feel good about themselves". Even chess.com ratings are massively inflated compared to otb ratings
They're inflated on higher side and deflated on lower side.