The shown Elo in my test are preliminary and therefore not 100% precise. They show a direction rather than an aim.LetoAtreides82 wrote:Currently just 4 elo under Rybka 4 in the Ipon tournament, and their respective matchup is currently 50%. It's almost like their identical in strength.
1. The Elo for the opponents are Bayeselo
2. The Classic GUI is calculation more straight, maybe more like ELOstat - so there is a mismatch
3. The Classic is simply calculation an average ELO of the opponents, while in reality the opponent ELO is build with the outcome of the games. The Classic is considering the opponent elos as static while they are dynamic.
Right now (1200 games) the real ELO calculated with Bayes is 8 Points lower than it is displayed.
Bye
Ingo