But how about leaving the rank the same but change the algorithm for how the games affect the rank for 9x9 and 13x13. Correct me if I’m wrong but the algorithm here is based on the Elo system. So 2 people of the same rank should have a 50% chance of winning and the number gets higher or lower from there as the rank disparity increases. What I’m saying is to reduce the impact a 9x9 and 13x13 board will have on a rank as to a 19x19. Change the expected winning rate for a 5 rank difference for a 19x19 from eg. 90% to 65-70% because I’m sure stronger ranked players here can agree they tend to lose to lower ranked players more on a smaller board. There’s a lower chance for a stronger player to gain an advantage and a weaker player to lose it.
I can play in a 9x9 tournament and play 6 rounds winning 5 of them against lower ranked opponents and losing 1 against an equal/stronger opponent and the net result will still be a negative for my rank simply because of that 1 loss. And it gets a lot lower should I lose even once to weaker opponents.
I can safely say that I have won more than 50% of my games against equal/stronger ranked players but I somehow still get stuck at my current rank for a while because of all the losses to weaker opponents that I’ve only occasionally lost to on smaller boards. I think it’s because I lose more ranking in the few matches with weaker opponents than what I gain with the equal/stronger opponents. Probably because I win 70 percent when I probably have to win 90 percent (rough guess) according to the current system against 5 stone weaker players. To illustrate my point, I have consistently won/lost close games with 7-12 stones stronger opponents as a SDK and my wins with weaker player,though usual, are often close even though they are 5 stones weaker.
PS. I almost exlusively play on smaller boards only
It makes more sense to either leave smaller board sizes unranked or to have separate ranks. 9x9 and 13x13 just confound the rating system that is supposed to represent someone’s 19x19 level. I find the argument “But we only confound that rating a little bit” to be quite unconvincing.
Since I believe that players would tend to have similar win/loss ratios regardless of the board size, I think there is information that can be generated by play on all board sizes. I agree that there is more noise in the outcomes on the smaller sizes which is why I agree with their lower weighting.
Any rating is an estimate based on observed outcomes. The key is consistency. While I can win a game on a smaller board against a stronger player, I don’t think I can do so consistently.
More input to the rating system yields better ratings.
I think it would be an interesting scientific experiment to run a simple test on all past games of different sizes (1919, 1313, 9*9) and time settings (blitz, live, correspondence), and compare the outcome of ratings (either combining them together or separately). Then we can see why we should do this way or that way.