| |
BGonline.org Forums
Luck vs. Skill
Posted By: Maik Stiebler In Response To: Luck vs. Skill (phil Simborg)
Date: Sunday, 7 December 2008, at 9:21 a.m.
the major reason that two players of relatively equal skill play a given match at different ratings is because of this [luck] factor,
I really like that and agree with it. I think too many backgammon players, after botching a match according to analysis, will feel down and think their game is deteriorating, instead of acknowledging that the match was unusually hard for them.
However, if you choose to call it luck and really mean it, this observation is irrelevant to trichard's premise that there's to much luck in the game, or wouldn't you agree?
I don't believe we have any method of accurately measuring this factor, particularly since it is difficult to determine what is "challenging" to one player as opposed to another.
We can still get a feel of how large this kind of luck factor is. Take a bunch of your matches against a set of opponents who you consider your equal on average. Look at how much the Snowie-favorite-in-%-number fluctuates.
I believe that this, more than anything else, determines the winner when you have two players who are very close to the same skill level.
I disagree. I have never done what I describe in my last paragraph, but from my recent experience (an error-rate-five-point-match-tournament between intermediate?advanced?Snowie 7ish! players) and some back-of-the-envelope calculations I conclude that, while it's annoyingly common for players in this category to play matches of nearly double their average error rate, it's rather rare that one of them (given they are usually equal) is a 75% favorite in any match. 75%, according to the implicit definition used in the thread, would be the threshold where one might say that this luck factor is equally large as any other. (Caution, technical digression: I disagree with that as well, I would rather say one luck factor is as large as another if it contributes the same amount of variance. Half of the total variance in mwc is (50%/sqrt(2))^2=(35.5%)^2. So, I would really only agree with Phil's statement if we routinely see the player who is confronted with the easier decisions as a 85% favorite).
I'd love to see data for longer matches (should work towards your hypothesis) and stronger opponents (should not), though.
Another thing that's often been argued is that it's usually the better rolls that are botched (think doubles vs. dances). So, you might find that being a favorite according to the Snowie statistic is correlated to losing the match.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.