Here are the details I promised of my last (very badly played)
tournament.
I would welcome comments on the discrepancies in the comparison, and
particularly on the large discrepancies in comparison of individual
moves.
Adding the normal manually-calculated (from Maven suggestions)
endgame figures to the Quackle kibitzer figures up to but not
including the endgame, the comparisons look like this.
Game 1) Maven sim 121
Quackle kibitz 140
Game 2) Maven sim 131
Quackle kibitz 124
Game 3) Maven sim 174
Quackle kibitz 157
Game 4) Maven sim 149
Quackle kibitz 118
Game 5) Maven sim 48 (so it wasn't all bad)
Quackle kibitz 20
Game 6) Maven sim 61
Quackle kibitz 70
The averages over the 6 games are Maven 114, Quackle 105.
What bothers me most is the discrepancy in some of the individual
move equity losses.
Here are a few examples:
Maven Quackle
43 55
21 34
16 9
41 30
8 20
14 28
1 10
9 0
0 15
19 0
The suggestion was made on UK-S that it is not necessary to sim
every move and that experience should be enough to tell you which 2
or 3 moves need simming. In response to that I posted the following
details of my fifth game at Shipley last Sunday, v Darryl Francis.
I asked, "Without reference to anything but the Quackle kibitzer,
which moves would you consider it necessary to sim?"
Darryl Pete
Change 5 - 0 0 ALRRUWZ LUZ G8a 24 24
QUIVER H7d 19 19 ARRWIED QUIVERIER H7d 63 87
TREACLY F11a 24 43 ADRWAOO AROW L12a 32 119
ICY J10d 10 53 ADGINOR ORGANDIE A14a 78 197
MOOI A12d 27 80 BEIMNTT WEMB O12d 33 230
SPOUSInG C7d 74 154 INTTOGN INGOT D8d 30 260
AXIL F13a 40 194 NTBANTE BANT E5d 17 277
aDVANCE B2d 88 282 ENTSANE NONE N11d 18 295
FaTED A1a 27 309 AESTPEE PEE A7d 20 315
VOGUE B3a 18 327 AESTJSW JEW F2d 29 344
HO F6d 28 355 AESSTAI SETA G3d 27 371
REF M12d 26 381 ADEIIRS DI H4d 16 387
KO A4d 24 405 AEIRS ARISE I1d 17 404
- HAULT -8 397 +8 412
The suggested response was the first move, the third and maybe one
or two more.
Here is the move-by-move comparison of a Maven sim and a Quackle
kibitz.
Maven sim Quackle kibitz
LUZ Best Best
QUIVERIER Best Best
AROW Best Best
ORGANDIE Best Best
WEMB Best Best
INGOT Best Best
BANT -5.82 -5.6
NONE -9.46 -3.8
PEE Best -2.6
JEW -6.26 Best
SETA -18.57 Best
Endgame moves (manual calculation)
DI -8.0
ARISE Best
My contention is that the kibitzers of both Maven and Quackle are
not reliable enough for us to accept their estimates without
simming. It would be interesting to see a comparison of a Maven sim
with a Quackle sim. Perhaps someone for whom Quackle runs fast could
do one to compare?
Pete
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/quackle/
<*> To unsubscribe from this group, send an email to:
[EMAIL PROTECTED]
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
