How about Elo range normalized to time? If someone is 100 Elo points higher than me they win 64% of the time—if we play 3 games they’ll win best of 3 71% of the time, which is like 160 points higher rather than 100? So you figure out this function and then rank games by Elo per time
How about Elo range normalized to time? If someone is 100 Elo points higher than me they win 64% of the time—if we play 3 games they’ll win best of 3 71% of the time, which is like 160 points higher rather than 100? So you figure out this function and then rank games by Elo per time
This was one of my first ideas. It fixes the deca-chess bug, but not the coin-chess bug, I think.