More Decks = More House Adv. Why?

[DSchles]

Would have to study it, Eric. Just quoting Griffin to you. Surprised you ever thought CD strategy had anything to do with it. You can see from the differences in edges that this is a very minor effect.

Shack makes the dealer break rate at SD S17 BJ 29.8%, while at 6D, it is 29.6%. Doesn't seem huge to me, either, although with just a half percent difference in the two edges, 0.2% isn't insignificant. I have learned over the years to trust my intuition and to be extremely careful when we choose to say that Griffin is "wrong." It just doesn't happen very often.

My gut tells me the doubling is worth more. Splitting isn't, because splitting is much more frequent in multi-deck than in SD, so that offsets the effects of removal contribution, to a great extent.

I'll try to look at your work when I get a minute.

Don

[ericfarmer]

... I have learned over the years to trust my intuition and to be extremely careful when we choose to say that Griffin is "wrong." It just doesn't happen very often.

My gut tells me the doubling is worth more. Splitting isn't, because splitting is much more frequent in multi-deck than in SD, so that offsets the effects of removal contribution, to a great extent.

I don't trust my gut . I should clarify-- I did not intend to suggest that Griffin was wrong, just that his treatment is incomplete. I can see how a first glance at Chapter 8 would seem to corroborate your "gut" telling you that doubling is the dominant effect. But read more closely: "To begin with, almost half of the .69% difference in expectation between one deck and many decks can be traced to the fact that the favorability of doubling down is reduced from 1.59% for one deck to 1.34% for the infinite deck."

First, I wouldn't call 36% "almost half." Second, that favorability difference (for reference, compare with my computed contribution of 21%, which is unfortunately for a different set of rules) *combines* effects that I have tried to separate, namely the ability to vary strategy in a composition-dependent manner (7%).

But third and most importantly, Griffin seems to be speculating about the remaining unaccounted 53% of the effect (compare with my computed 57%) "presumably"-- his word, not mine-- being due to "judicious standing with stiff totals." We can make this presumption more precise, and then justify it, as described in the above analysis.

In short, all we are really doing is confirming-- and clarifying-- Griffin's essentially correct speculation that the dominant effect (where by "dominant" I mean over half of the effect by either his or my reckoning) is likelihood of dealer busts.