After doing some research on bankroll and risk of ruin, I decided to do some empirical work on the subject.
I ran sets of 4000 hands on winpoker in 9/6 job. I ran 100 sets and found some interesting results:
The lowest result was a loss of 1845 coins. The best result without a royal was 915 coins. I used the variance from winpoker and calculated the standard deviation and found the results were -2.807 sigma on the low end and plus 1.6 sigma on the high end. If I include the sets with a royal, the sigma goes up to almost 8. There were NO sets with results between 1.75 and 4.25 sigma.
I realize that the size of the set might be a factor in these results. What it did show me is that the results you will see are a little more tightly clustered around a mid point with the 1 in 40,000 hand spike from a royal. I plan to repeat this a couple of times and see if anything else of interest comes out of it.