Standard Deviation manipulation

OK, so I have some basic statistics knowledge, and I know how to calculate a standard deviation and what it means, but I either never learned this or am drawing a blank.

Pokertracker gives standard deviation per hour. How can you calculate the standard deviation per 2 hours? per n hours? I can't even figure out if it should be higher or lower. You expect the results to regress to the mean, but that's the average. The standard deviation per 2 hours would have to be less than double the per hour calculation but I can't figure out if it would even be higher. This is just really confusing me, can someone help me out here? Thanks.

wow, 40 views and 0 responses. Guess I should have found a math forum then

I have no idea what "standard deviation per hour" could possibly mean. That seems like going to the grocery store and seeing a sign that says "Watermellons, $2/hour". The time tacked onto standard deviation? Why? You need to ask PokerTracker, "wtf?"

I don't really know about SD over time, but it would probably be far more useful to you to analyse SD over x number of hands rather than strictly over time. It is of course possible that PT is providing you with this figure anyway (i.e. calculating your average h/ph and then using this figure to assign the number of data points that it tracks to provide the SD/h) but I don't know about the workings of this software. What is it that you are measuring the SD of?

Pure guess, because I have no idea what the Poker Tracker SD is measuring, nor do I have PT.

A single hour might not give you a decent sample, just like 100 hands is not as good as 10,000 hands at determining any particular measurement.

If it is measuring your play, you might think about taking random hours, perhaps over multiple days or sessions, say 10 of them, and recalculating those values to get a more valid idea. Once you determine a baseline with sufficient sampling, any particular hours SD will give you better info.

If it is measuring opponents play, that might be hard to do.

Well the way it is calculated I believe is it just takes each hour, and uses the profits for those hours as data points, where a standard deviation can be calculated. The mean would obviously be the profit/hour, but say you have a standard deviation per hour s and total profit/hour p, that would mean that every hour, your chance of ending the hour with p + or - s is 68%, or finishing with p + or - 2s is 95%.

That would not be standard deviations per hour but standard deviation of profit per hour. Perhaps their designation is just faulty that way.