I would like to have a theoretical grounding for reducing my variance at the cost of value. Obviously I don't want to "fold anything but the nuts" and I do read the hands OK, but the problem is that my variance is too great, so even though I would love to take those odds a thousand times, the problem is I can take them ten times. I know the real answer is "get a bigger bankroll" or move down, but I'm asking theoretically whether there is another answer here.

So is there some kind of theoretical foundation for playing with reduced variance? I would think this is also relevant to playing with a shortstack in a tournament - since a single all-in loss removes you from the tournament - but this is just a guess. Thanks.

If you want to reduce variance, the typical method is "when you are not sure whether to call or fold, fold". If the decision is very close then the EV is close to 0, and the variance of calling will be larger. The earlier you fold, the better, so basically you can start by playing tighter.

If your bluffs are very low earners (including meta-game) then you can consider reducing bluffing considerably also. Whether or not this is feasible depends on how tough your competition is (so does the folding method, of course, because observant opponents will bluff you more)

Of course, you can also just get someone to buy part of your action.

thanks - this is useful.