Originally posted by @nevare
Your true online rating is your "opponent average rating"... look in your profile and you might be surprised. Their are some highly rated chess players who play much weaker opponents. I will try to keep mine accurate.
Originally posted by @mchill
Their are some highly rated chess players who play much weaker opponents. I will try to keep mi ...[text shortened]... gained besides an inflated number next to their name? They are pathetic and insecure creatures.
Two friends meet to play chess once a week. They keep track of the games and their win loss record. After months of play the record shows one friend winning 99 out of every 100 games on average. The friends should have ratings that differ by about 800 points.
Is it your contention that the two friends should be rated the same? The stronger friend is only playing the weaker friend, low "opponent average rating". Is the stronger friend not likely as strong as any other player that could compile the same record against the weaker friend?
Let's create, in thought experiment, two large groups of chess engines. One group we limit, by various diverse weaknesses in play, to ratings of 1400 against the RHP population. The second group we similarly limit to playing as diverse 1700 players.
Any engine in the first group by definition must maintain an even win loss record in the first group. Any engine in the second group by definition must maintain an even win loss record in the second group. They can not deviate from as they were created. There can be anomalies, one engine always losing to another and so on but all must maintain average in their group!
Now any engine from either group, plucked from its group and rating masked, would be identifiable by its play against either group. A 1700 engine played against the 1400 group would win 85 out of 100 games and thus reveal its 1700 rating. A 1400 engine played against the 1400 group would win 50 out of 100 games. A 1700 engine played against the 1700 group would win 50 out of 100 games. A 1400 engine played against the 1700 group would win 15 out of 100 games.
Our user names are a black box, however their inner workings produce the output of moves that lead to their ratings, against any sub population of rated those same ratings should emerge, on average. It is possible that for very human reasons some ratings would rise and others fall but it is as likely that any player that has played a weaker population for his stronger rating is underrated!