Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: % of first move beta fails question.

Author: Michel Langeveld

Date: 05:38:09 03/09/04

Go up one level in this thread


On March 09, 2004 at 05:22:45, macaroni wrote:

>How do people measure their percent of first move beta fail rate.

if (val >= beta)
{
  betaCounter++;
  if (legal_moves == 1) beta1stCounter++;
  return val;
}

> Do you include null-move?

No

> hash returns?

No

And also QSearch ... : no

>cheers

I see typically values between 86%-96% for a move.

I heart that:
>= 90% is reasonable
>= 95% is good

The average FH of Nullmover of an entire game is typically between 89 and 91.

Moveordering (killers, see, history heuristic, use the move from the hashtable)
can influence these values significant. As you can see I have some work to do.

>Tor



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.