BGonline.org Forums

compare Rockwell-Kaz to G11 and different plies in RO: odd/even effect in XG?

Posted By: Klaus Evers
Date: Thursday, 3 June 2010, at 1:10 p.m.

In Response To: Rollout 3 ply (Adam Versaw)

Interesting: After I saw Adam's RO I wondered if the difference compard to my RO I posted earlier comes from the fact that he uses 3/3ply or from the different MET.

MET: Rockwell-Kazaross

	is Player 2 score: 8 pip: 108 11 point match pip: 97 score: 9 is Player 1
XGID=--aCCCBbC----------bccbb-A:1:1:1:21:9:8:0:11:10
to play 21

1.	Rollout1	Bar/24 4/2*	eq: -0,288
	Player: Opponent: 43,02% (G:13,56% B:0,73%) 56,98% (G:16,28% B:1,39%) Conf: ± 0,005 (-0,293...-0,283) Duration: 4 minutes 24 seconds
2.	Rollout1	Bar/24 8/6	eq: -0,310 (-0,022)
	Player: Opponent: 37,33% (G:14,54% B:1,28%) 62,67% (G:6,43% B:0,43%) Conf: ± 0,004 (-0,314...-0,306) Duration: 3 minutes 44 seconds
3.	Rollout2	Bar/24 3/1	eq: -0,362 (-0,074)
	Player: Opponent: 35,84% (G:13,36% B:1,12%) 64,16% (G:8,75% B:0,49%) Conf: ± 0,008 (-0,370...-0,354) Duration: 49,1 seconds
1 5184 Games rolled with Variance Reduction. Moves and cube decisions: 2 ply Search interval: Large 2 1296 Games rolled with Variance Reduction. Moves and cube decisions: 2 ply Search interval: Large

eXtreme Gammon Version: 1.14, MET: Rockwell-Kazaross

---

MET: GnuBG 11 point

	is Player 2 score: 8 pip: 108 11 point match pip: 97 score: 9 is Player 1
XGID=--aCCCBbC----------bccbb-A:1:1:1:21:9:8:0:11:10
to play 21

1.	Rollout1	Bar/24 4/2*	eq: -0,285
	Player: Opponent: 43,03% (G:13,54% B:0,74%) 56,97% (G:16,30% B:1,40%) Conf: ± 0,005 (-0,290...-0,280) Duration: 3 minutes 33 seconds
2.	Rollout1	Bar/24 8/6	eq: -0,309 (-0,023)
	Player: Opponent: 37,33% (G:14,54% B:1,28%) 62,67% (G:6,42% B:0,44%) Conf: ± 0,004 (-0,313...-0,305) Duration: 2 minutes 56 seconds
3.	Rollout2	Bar/24 3/1	eq: -0,361 (-0,075)
	Player: Opponent: 35,84% (G:13,34% B:1,11%) 64,16% (G:8,74% B:0,49%) Conf: ± 0,008 (-0,369...-0,353) Duration: 47,4 seconds
1 5184 Games rolled with Variance Reduction. Moves and cube decisions: 2 ply Search interval: Large 2 1296 Games rolled with Variance Reduction. Moves and cube decisions: 2 ply Search interval: Large

eXtreme Gammon Version: 1.14, MET: GnuBG 11 point

Well, not quite what I expected. I thought the difference comes from different MET. So lets see what my 3ply RO with G11 brings up:

---

MET: GnuBG 11 point

	is Player 2 score: 8 pip: 108 11 point match pip: 97 score: 9 is Player 1
XGID=--aCCCBbC----------bccbb-A:1:1:1:21:9:8:0:11:10
to play 21

1.	Rollout1	Bar/24 8/6	eq: -0,322
	Player: Opponent: 36,97% (G:14,55% B:1,10%) 63,03% (G:6,79% B:0,43%) Conf: ± 0,005 (-0,327...-0,317) Duration: 11 minutes 02 seconds
2.	Rollout1	Bar/24 4/2*	eq: -0,339 (-0,017)
	Player: Opponent: 42,46% (G:14,83% B:0,67%) 57,54% (G:20,21% B:1,62%) Conf: ± 0,007 (-0,346...-0,332) Duration: 12 minutes 49 seconds
3.	XG Roller+	Bar/24 3/1	eq: -0,357 (-0,035)
	Player: Opponent: 35,88% (G:12,95% B:0,81%) 64,12% (G:8,13% B:0,17%)
1 2592 Games rolled with Variance Reduction. Moves and cube decisions: 3 ply Search interval: Large

eXtreme Gammon Version: 1.14, MET: GnuBG 11 point

OK, looks like a confirmation that in this position it matters if you do the RO 2ply or 3ply. There is virtually no difference

in the equities if u compare R/K to G11 MET.

---

MET: GnuBG 11 point

Final Q: is there an odd/even effect in XG?

	is Player 2 score: 8 pip: 108 11 point match pip: 97 score: 9 is Player 1
XGID=--aCCCBbC----------bccbb-A:1:1:1:21:9:8:0:11:10
to play 21

1.	Rollout1	Bar/24 8/6	eq: -0,307
	Player: Opponent: 37,87% (G:14,45% B:1,26%) 62,13% (G:7,09% B:0,44%) Conf: ± 0,004 (-0,311...-0,303) Duration: 49,8 seconds
2.	Rollout1	Bar/24 4/2*	eq: -0,328 (-0,022)
	Player: Opponent: 42,85% (G:14,64% B:0,80%) 57,15% (G:19,94% B:1,66%) Conf: ± 0,005 (-0,333...-0,323) Duration: 58,3 seconds
3.	Rollout2	Bar/24 3/1	eq: -0,352 (-0,046)
	Player: Opponent: 36,60% (G:14,07% B:1,08%) 63,40% (G:9,30% B:0,51%) Conf: ± 0,006 (-0,358...-0,346) Duration: 20,9 seconds
1 5184 Games rolled with Variance Reduction. Moves and cube decisions: 1 ply 2 2592 Games rolled with Variance Reduction. Moves and cube decisions: 1 ply

eXtreme Gammon Version: 1.14, MET: GnuBG 11 point

OMG !

Well, if 3ply and 1ply RO agree and 2ply RO comes up with different results, that seems to be heavily indicating there is an odd/even effect in XG.

Would we need to do 3ply RO if 1ply RO comes up with the same result?

Isn't 3ply RO basically a waste of time, if you can get the same results from 1ply RO?

What does this effect tell you about 4ply rollouts?

Any comments appreciated!

Messages In This Thread

• 21 to play from the roof
  Adam Versaw -- Monday, 31 May 2010, at 1:53 p.m.
  • My quick thoughts OTB
    neilkaz -- Monday, 31 May 2010, at 2:58 p.m.
  • 21 to play from the roof
    Timothy Chow -- Monday, 31 May 2010, at 6:34 p.m.
  • some rollouts
    Klaus Evers -- Wednesday, 2 June 2010, at 8:55 p.m.
    • some rollouts
      Timothy Chow -- Wednesday, 2 June 2010, at 9:03 p.m.
  • Rollout 3 ply
    Adam Versaw -- Thursday, 3 June 2010, at 10:09 a.m.
    • compare Rockwell-Kaz to G11 and different plies in RO: odd/even effect in XG?
      Klaus Evers -- Thursday, 3 June 2010, at 1:10 p.m.
      • compare Rockwell-Kaz to G11 and different plies in RO: odd/even effect in XG?
        Timothy Chow -- Thursday, 3 June 2010, at 3:16 p.m.
        • compare Rockwell-Kaz to G11 and different plies in RO: odd/even effect in XG?
          Klaus Evers -- Thursday, 3 June 2010, at 3:56 p.m.
          • XG odd/even effect
            Matt Cohn-Geier -- Thursday, 3 June 2010, at 4:50 p.m.
            • XG odd/even effect
              Leonardo Jerkovic -- Thursday, 3 June 2010, at 6:26 p.m.
            • XG odd/even effect
              Nack Ballard -- Thursday, 3 June 2010, at 11:00 p.m.
              • XG odd/even effect
                misja -- Friday, 4 June 2010, at 3:15 p.m.
                • XG odd/even effect
                  Nack Ballard -- Friday, 4 June 2010, at 9:22 p.m.
      • rollout and g11
        Adam Versaw -- Thursday, 3 June 2010, at 11:33 p.m.