how about an "average place" stat, which could be given any denominator. ie: someone finishes 36th in a pond of 96, and then finishes 2nd in a pond of 16. We would probably make 16 the base denominator (although others are possible) and from the above example give the player an average placing of 4/16, meaning thus far they have finished 4th on the mean for every 16 players in their ponds. Of course, not all examples will be whole numbers, so we could round off to whole or 1 or 2 decimals. That way, the average placing out of 16 participants could replace a cumbersome 4-digit number and wouldn't have to be weighted either. Of course, show the wins and total ponds and other cool stats too. But this one would, like the usual BKR ratings, give a reasonable expectation of how a 5.0 and 9.6 would do in the same pond...the 5 would be expected to finish 4.6 places ahead of the 9.6. And of course, as the sample set increased, the numbers would flatten out to more accurate measures. I believe comparing expected placements is better than BKR because it is not head to head by any stretch of the imagination. Also, extremely easy for Fencer to program.