BGonline.org Forums

INTERPRETING SNOWIE ROLLOUTS

Posted By: Ray Kershaw
Date: Friday, 28 August 2009, at 3:59 a.m.

I want to know which of two plays is the better. I set Snowie rollout settings at "full rollout / 360 games / played 3-ply precise, cube 3-ply / settlement 0.550 at 4pts / random seed, with race database".

Play A: 95% confidence interval live cube: +0.200 +/-0.020

Play B: 95% confidence interval live cube: +0.230 +/-0.020

Since the confidence intervals overlap (A is +0.180 to +0.220; B is +0.210 to +0.250), am I correct to conclude that my sample size is too small to be able to say that Play B is better than Play A ?

I have a vague recollection that this is not the correct way to compare two sample means but I don't deeply understand what underlies Snowie output.

Messages In This Thread

• INTERPRETING SNOWIE ROLLOUTS
  Ray Kershaw -- Friday, 28 August 2009, at 3:59 a.m.
  • INTERPRETING SNOWIE ROLLOUTS
    Keene -- Friday, 28 August 2009, at 5:19 a.m.
  • INTERPRETING SNOWIE ROLLOUTS
    Tom Keith -- Friday, 28 August 2009, at 10:06 a.m.
    • INTERPRETING SNOWIE ROLLOUTS
      Ray Kershaw -- Friday, 28 August 2009, at 10:33 a.m.