Author: Michael Henderson
Date: 14:26:44 09/19/04
Go up one level in this thread
On September 19, 2004 at 16:44:18, Robert Hyatt wrote: >On September 19, 2004 at 13:23:32, Michael Henderson wrote: > >>I'm confused about the definition of this stat. Does one count null moves and >>hash hits in this? My fh numbers vary dramatically based on what is counted. >>Thanks. >> >>Michael > > >No. After you generate moves (which is after the null-move search, hash probe, >repetition check, etc) you watch for when you fail high at this ply. For every >fail high, increment a counter total_fail_highs, if you are searching the first >move at this ply, increment another counter fail_highs_on_first. When the >search terminates and you print statistics, print fail_highs_on_first * 100 / >total_fail_highs. 90+ is a good number for most cases... I can get mid-upper 80's mostly with my program using crafty-style move ordering. just some things I noticed: this stats tends to be higher when you have a steady PV more acute eval makes it higher--leads to steady PV and easier cutoffs inaccurate root move ordering has a bad effect on it--i do have this problem
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.