Number of games necessary to get approx PR
Posted By: Matt Cohn-Geier In Response To: Number of games necessary to get approx PR (Keene)
Date: Wednesday, 12 February 2014, at 12:06 a.m.
In Response To: Number of games necessary to get approx PR (Keene)
Another thought here - it is far easier to play against XG than it is to play against a human opponent. XG doesnt make ridiculous plays (as often..), it doesnt get into silly positions (as often..), and as soon as the doubling window opens, its cubes you. ... A human will get into ridiculous positions that you dont see in practice. ... Long story short. Being able to play a 4 vs XG is not the same as being able to play a 4 vs random human.
No, I don't believe it.
So, let me say again: Your PR reflects your understanding of backgammon positions. If you understand backgammon at a 4 level, you play at a 4. If you don't understand backgammon at a 4 level, you don't play at a 4. It's ultimately that simple.
There are lots of explanations and long stories about why PR varies from day to day. But in the long run none of those stories add up to anything more than a minute difference.
Now sure, if you play an opponent who likes to slot everything and get into crazy priming and back games, your PR will tend to be higher than against XG. But as a long-run average vs random opponents, if you play a 4.0 against XG, it's hard to see how playing against humans would cause you play worse than say, 4.2.
If someone says to me, "I play a 6.5 live, but I play a 4.0 at home vs. XG," based on that statement alone, I would consider myself over 99% confident that they do not really play a 4. There is likely either bias in the sample (cherry picking, for example), or some positions are checked (maybe they're not checking this decision, but just checked their last checker play and saw that it was a blunder, and change their game plan accordingly), etc. Somewhat more unlikely but still possible is that either sample is too small (maybe only 3 matches). The odds that someone plays 6.5 live but 4.0 vs. XG are virtually impossible.
I'll leave open the possibility that if your PR is high or if you regularly play opponents whose PR is high, then there is more room for variance in the PR, because the standard deviation tends to increase significantly as PR increases.
Messages In This Thread
BGonline.org Forums is maintained by Stick with WebBBS 5.12.