| |
BGonline.org Forums
The view I posted may be confusing...this should be clearer
Posted By: Chase In Response To: The view I posted may be confusing...this should be clearer (Timothy Chow)
Date: Thursday, 18 March 2010, at 6:12 p.m.
I've been trying to improve my OTB estimates. I'm still not very good at it, but I've found that even poor estimates can be helpful. For example, in a money game, let's say I estimate the player on roll wins 30% gammons, but he only wins 25, and I estimate his opponent wins 10% gammons, but he actually wins 15.
With a gammon value of .5 for both players, I'd subtract half of 10 from half of 30 to get a 10% gammon adjustment. I know the gammonless take point is around 22% (taking recube vig into consideration), so I'd add the 10% in to get a take point of around 32%. Then I'd ask myself if the taker can win 32%.
Using the actual values would give 25/2 - 15/2 = 5%, for a take point of only 27%, but this is an extreme example. I can usually come within 2 or 3% in my initial estimates, and when I'm off, I'm often off in the same direction, so the errors will tend to cancel each other out. So I can often estimate poorly and still come pretty close to finding the actual take point. And even if I'm off by 5%, as in this extreme example, it's still not hopeless. I always check my result against my gut impressions about the position, and it often happens that, even with the wrong numbers, I end up erring in the right direction and making the correct decision anyway, or perhaps I'll make an equally bad estimate of the takers winning chances and that will "realign" my errors. Kit is right when he often says "these numbers may be way off, but we have to use something."
As far as the actual estimates go, I should probably start collecting reference positions. In the meantime, I simply ask myself, "If we played this position out 5 times would I expect to lose a gammon at least once?" If so, that's 20%. Next question, "If we played it 4 times, would I expect to lose a gammon more than 1 time?" If so, I'm up to 25%. "If we play it out 3 times, would I expect to lose a gammon more than once?" Now we're up to 33%. Usually, I'll find that this closes me in on a fairly tight range. If I decide I lose about 28% gammons, I'll usually do a sanity check and say "If I play this 10 times, will I lose a gammon more than 3 times?" By asking the question in different ways I begin to hone in on where I think it should be. Then I compare it to my initial reaction.
As I say, I'm still not very good at all this, but just training myself to ask these questions every time I look at a cube decision is helping me improve.
Perhaps the taker's wins will be so far
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.