HEM keeps track of standard deviation by BB/100. Hypothetically, you're playing a 100nl game, with a standard deviation of 30BB/100. Does this mean for every 100 hands, you're expected to be up or down 60 dollars if you're running within 1 standard deviation. If you're up 120 dollars, you'll be up 2 standard deviation. If you really run hot, 180 dollasrs would assume you're 3 standard deviations above the mean. For simplicity, let's assume that you're a breakeven player to avoid adding your winrate to the calculations.

If my above assumptions are correct, standard deviation isn't really a good measure of anything in poker. You swing multiple buyins within 100 hands even if you're a break even player... That would mean you're always running several standard deviations above the mean? I'm either not understanding how SD is working with HEM measurements or it's a poor measure of anything.

Well in the second part, I'm just trying to say that if my understanding of the first part is correct, then if you have a standard deviation of 30ptbb/100, you would expect to be up or down 60 dollars after 100 hands most of the time. If you're down 180 dollars, that's 3 standard deviations below the mean and that shouldn't happen that often, but 2 full buyin swings definitely happen in a matter of hands. Am I just being confused in that I play so many "hundred" hands a day that the variance isn't really that great and there are just so many samples for me to run "less than 3 standard deviations" below the mean eventually?

Yes, I've seen many of those variance simulators, but I don't understand how someone could eventually run so many buyins deviating from the mean if the variance is only 30ptbb/100 or a little more than half a buyin. I thought my misunderstanding of 30ptbb/100 or 60 dollars/100 in my example would mean that by just running at less than 1 buyin below the mean in 100 hands would mean a standard deviation of 2 almost and that really shouldn't happen. Eveything should fall within a standard deviation or two, so I'm thinking my understanding of the HEM standard deviation is flawed. Is the main flaw from thinking in terms of 100 hands and in a poker setting, you'll have many hundreds of hands so the variance is magnified because you have so many opportunities to deviate. I hope my thoughts are somewhat coherent.

If you have bigger swings than seem likely from standard deviation then you are probably underestimating your standard deviation. The other possibility is your results trend, most likely from tilting.

But I'm sure you're calculating correctly from your post. If you have a standard deviation of 30 $2 bb per 100 hands, then only rarely should you be up or down more than $120 over 100 hands. It won't be as infrequent as a Normal distribution would indicate (1 time in 40) because no-limit results have fatter tails than Normal over 100 hands. But over 1,000 or 10,000 hands, that effect should become negligible.

But remember, the size of your swings increases with the square root of the number of hands. If it's $120 over 100 hands it's $240 over 400 hands and $480 over 1,600 hands.

Thanks, this is what I was looking for, I think I understand it a lot better now.

NLHE hands aren't normally. For example, my standard deviation is typically about 120 bb/100. So that means my standard deviation is 12 big blinds in one hand (standard deviation is proportional to SQRT(n), so to go from 100 hands to one hand, i divide by 10). So, losing a buy-in in one hand is about 8 standard deviations from the mean. The calculator I normally use doesn't even deal with events that are so unlikely. So obviously the distribution is no where close to normal.

However, when you take the repeated sum of any random variable, the sum eventually starts to look normal. For 100 hold 'em hands, the normal approximation still is incredibly inaccurate, so standard deviation isn't an incredibly helpful stat. For like 10,000+ hands, the normal approximation is quite accuate, however.