Add it up ...
the counter is only playing one hand, doesn't the house have the same, perhaps even more of an edge than the counter?
Most of the advantage in high counts comes from the increased chance of getting a blackjack. If you're playing one hand, you get paid 3:2 on each BJ, just as you do with two hands. Now, you COULD get BJ on both hands, if you spread to two. OTOH, the dealer could get BJ and wipe out two hands in one shot.
To work out which is best, you really need to consider how many good cards you use up in that plus count. Here's a replay of a formula I put up on Cover/Comps a couple months ago:
--------------------
Under my hypothesis, to maximize EV per hour, we want to maximize:
(Edge x total dollars bet)/(total number of hands including dealer and other players)
When playing alone, and "constrained" as per DD's POM, we get: (edge x 2B)/3 for two hands, and (edge x 2B)/2 for one hand. So if edge is negative, two hands is better. If not constrained, we get (edge x 2B)/3 for two hands, and (edge x B)/2 for one hand, so if edge is negative, one hand is better.
Under the formula, whenever we're constrained, we should play two hands into a negative edge. If we drop the constraint, we should always play one hand into a negative edge. Might be useful to point out here, that it never hurts DI to increase spread by DEcreasing your minimum. So the "constraint" we're discussing is purely a constraint for cover. Covariance and ROR should not change either conclusion. So the formula gives valid results, IMHO, for this situation of choosing between one hand or two in negative EV shoe.
Also interesting to note the formula gives good results for the classic +EV situation. The optimal amount to play for two hands is considered to be about 0.73 on each hand. So the object is to maximize (edge x 1.46B)/3 and (edge x B)/2. Since edge and B are positive, you can cancel them out of the inequality and compare 1.46/3 to 1/2. So when playing alone, it is marginally better to play one hand. When you have "company," the denominators increase (eg. 1.46/4 vs 1/3) and it's preferable to play two hands. If considering playing 3 or more hands, the 1.46 figure decreases and the formula prefers fewer hands.
One of the assumptions in the formula is that each hand takes a proportionate bite out of the remaining (positive or negative) shoe. When you know you're playing the last round, this assumption is wrong, and the formula isn't valid. Rule of 6, or rule of 5 also violates the assumption (moving cut card), so the formula isn't good for those rules.
Sims would put a finer point on it, by eliminating the constant cards per hour assumption, but I think the formula is useful.
ETF