I don't dare post this in the Probability forum, because I fear that running simulations to empirically demonstrate statistical theory is beneath them. But it's still fun! Especially when you're looking for ways to keep your brain engaged at work that aren't TOO obviously poker related.
So anyway, I wrote a little simulator in Python and ran it with the following tweakable assumptions:
<ul type="square">[*]mean = 1.1 BB/hr (Based on small sample from Foxwoods $2/4 and theoretical max win rate)[*]SD = 9.5 BB/hr (observed; consistent with theoretical 16 BB/100 hands)[*]session = 16 hr[/list]
Then I ran 10K sessions and collected the results by 10 BB interval.
Floor #
------ --
-130.0 2
-110.0 6
-100.0 19
-90.0 28
-80.0 41
-70.0 110
-60.0 153
-50.0 280
-40.0 418
-30.0 573
-20.0 716
-10.0 890
0.0 997
10.0 976
20.0 1032
30.0 959
40.0 839
50.0 647
60.0 469
70.0 336
80.0 209
90.0 141
100.0 79
110.0 41
120.0 27
130.0 7
140.0 3
150.0 1
160.0 1
So for example, two of the 10K simulated sessions were between -130 BBet and -120 BBet. Note that a substantial number (~10%) are worse than -30 big bets!
Anyway, I'd be happy to send the Python script if anyone wants to play with it. There's an Excel spreadsheet called swings.xls that someone posted a link to on one of the forums that does about the same thing, but prettier.
At any rate, it's fun to play around with. Hopefully it'll give me a better appreciation of how big variance is in poker.