I agree with both of the replies so far, but I'm most intersested in everyone's opinion of the following statement.
"All else equal, increasing statistical variance (whether counting or not) will have no effect on EV."
It will, however, increase swings (good if you enjoy swings, which I do) and increase your risk-of-ruin (I assume we'll all agree this is bad). But, each person will have a different opinion of an acceptable risk-of-ruin.
Here's what I'm getting at: Maybe a logical betting system could be developed that would focus on the effects of variance. For example, split your bankroll into 5 equal shares. With the first share use some betting scheme to maximize variance (maybe progression); forget RoR and go for the big score. If that doesn't work, reduce your variance as you get closer to your last share (maybe a more appropriate term would be "your last stand"). The final share's strategy would focus on minimizing RoR; hopefully it will keep you breathing through the last hours of your vacation.
Though betting cant change EV, maybe it can have some effect?
As always, this is just a thought; lemme know what you all think.
[Edit: This is based on my assumption that EV is calculated (or maybe can be calculated) with only one hand, that each hand is independant, and therefore that RoR doesn't affect EV. I may have this a bit twisted]