Appendix B  --  Bounds on the ``Best Change'' Probabilities

The ``Best Change'' behaviours of chess programs represent typical count data for a binary-valued random variable in the terms of standard statistics. The count probabilities of binary-valued random variables generally adhere to binomial distributions. For large enough sample sizes n and success counts m with m > 4 and n - m > 4, however, corresponding normal distributions provide practically sufficient approximations of the awkward to handle binomial distributions. Classic engineering statistics [88] derive the following lower and upper bounds of the success probability P for given values of m, n and any desired %-level of confidence as specified by the single-sided percentiles z of the N(0, 1) normal distribution.


\begin{displaymath}P~~>=~\left(m + z^2/2 - z * \sqrt{m * (1-m/n) + z^2/4}\right) * (n + z^2)^{-1}
\end{displaymath} (1.1)


\begin{displaymath}P~~<=~\left(m + z^2/2 + z * \sqrt{m * (1-m/n) + z^2/4}\right) * (n + z^2)^{-1}
\end{displaymath} (1.2)

With the help of these formulas we determined 80%-confident (z = 0.8416) and 90%-confident (z = 1.2816) bounds on the ``Best Change'' probabilities of BELLE (n = 447), CRAFTY (n = 343), and DARKTHOUGHT (n = 343). For BELLE we calculated the success count m from its ``Best Change'' rates of Table 1.4. For CRAFTY and DARKTHOUGHT we used the absolute ``Best Change'' numbers of Table 1.5 and Table 1.6 as their observed success counts m. The resulting bounds clearly discriminate the drops of the ``Best Change'' rates below 20% for all three programs with at least 80% confidence (see Table 1.13 where  >= denotes lower bounds and  <= upper bounds).



 
Table 1.13: Confident Bounds on the ``Best Change'' Probabilities.
Search Belle Belle Crafty Crafty DarkTh. DarkTh.
Depth C=80% C=90% C=80% C=90% C=80% C=90%
2 - - - -  >= 36.59%  >= 35.46%  >= 33.14%  >= 32.05%
3 - - - -  >= 34.57%  >= 33.47%  >= 37.45%  >= 36.32%
4  >= 31.26%  >= 30.32%  >= 28.56%  >= 27.52%  >= 29.70%  >= 28.65%
5  >= 31.26%  >= 30.32%  >= 28.27%  >= 27.24%  >= 27.42%  >= 26.40%
6  >= 25.99%  >= 25.11%  >= 25.43%  >= 24.43%  >= 22.59%  >= 21.64%
7  >= 27.75%  >= 26.84%  >= 22.59%  >= 21.64%  >= 19.48%  >= 18.59%
8  >= 24.24%  >= 23.38%  >= 20.61%  >= 19.70%  >= 23.16%  >= 22.20%
9  >= 20.97%  >= 20.16%  <= 20.19%  <= 21.20%  >= 19.76%  >= 18.87%
10  <= 19.24%  <= 20.10%  <= 18.98%  <= 19.97%  >= 22.31%  >= 21.36%
11  <= 19.70%  <= 20.57%  <= 18.38%  <= 19.35%  <= 19.29%  <= 20.27%
12 - - - -  <= 18.68%  <= 19.66%  <= 17.17%  <= 18.12%
13 - - - -  <= 16.25%  <= 17.19%  <= 18.38%  <= 19.35%
14 - - - -  <= 17.17%  <= 18.12%  <= 15.34%  <= 16.26%
 



Created by Ernst A. Heinz, Thu Dec 16 23:28:11 EST 1999