[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

Understanding Gnubg MWC

Posted By: Rich Munitz
Date: Saturday, 28 February 2009, at 4:56 p.m.

In Response To: Understanding Gnubg MWC (Barry Silliman)

"EMG does not take into account the match score, so the same move from the same position will always produce the same EMG numbers"

This is certainly not true since the "correct" move at one score (EMG = 0) can certainly be an error at another score.

The difference (simplified) is that EMG represents the error's effect on winning chances in the current game, whereas MWC represents the error's effect on winning chances in the match.

While maximizing MCW is ultimately, the sole arbiter of correctness in match play, I find it completely unintuitive in trying to analyze play, exactly because the absolute value of the same error changes radically as the match progresses. I'd much rather look at EMG which is normalized for the current game. If an oracle could tell us that the match would eventually wind up at DMP, then errors early in the match are in effect just as costly as errors at the end. It would be like trying to state the size of the error in money play in terms of the number of dollars of equity loss (i.e. considering stake and cube level in the equation). Sure the same mistake at a 16-cube is likely to have a greater impact on the wallet, but we all know that the stakes and cube level should never be a factor in decisions for money play, so it only confuses things to include this in the discussion of "error size".

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.