Author: Stephen A. Boak
Date: 19:07:33 08/24/01
Go up one level in this thread
On August 24, 2001 at 10:56:16, Uri Blass wrote: >I understand but I think that it is still better to have a better prformance >than having nothing. The goal in a Swiss tournament is to score the most points. That is the way to be the best program in the event--by winning more points than the competitors. Incidentally, that is also the best way to obtain the highest possible TPR. In another post, Diepeeven indicates he always tries/wants to win. I applaud that effort. Wins are needed to put a program on top, by any measure. There is no secondary prize for having a relatively high TPR. Usually the program that scores the most points will have the highest TPR. Since the ratings of the opponents each program faces will *limit* the possible Performance Rating you may achieve, a program/author can't solely take credit or blame for the ultimate TPR that the program achieves. Part is dependent on the chance inherent in the pairing system in a Swiss event. If you don't win more points than your opponents, you have nothing (no tournament win), and all the consolation talk about TPR is meaningless. Besides, a short event is not appropriate for placing such undue emphasis on results (either points or TPR). You either did best or not, in the event, compared to your opponents. However, the inherent relative strength of a program is not guaranteed to be evidenced by the small set of results (measured by points or TPR). This is due to randomness in performance (not just due to pairing randomness). Elsewhere, in another posting, Christophe Theron highlights the unavoidable effects of randomness (in results) with typical 'flippant' humor. :) With a coin flipping remark, he illustrates that, due to randomness, a program may have off games, and an off event. Most posters don't understand the import of his analogy (don't understand randomness), let alone the simple application of mathematical probabilities in the context of a specific tournament. Truly that isn't a big deal--just normal statistics at work, even with an occasional loss or two to much lower rated programs. Pointing out the TPR statistics is fun, I do it also, but it isn't a big deal in a single, relatively short event. It's just an interesting, but limited, feedback--about that event alone! Now tell me the TPR of a program across hundreds of games and I will take more interest (although the changes in programs & hardware during that period may make the interepretation more challenging). --Steve
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.