SD: 440 (this is an estimate but it's within 10%)
Hands: 200000 (2000*100 and we're trying to simulate 2000 hours)
# of Players: 100
The graph you get is the result of running an experiment where 100 players with the above true win rates and standard deviations play for 2000 hours. The results are pretty mind boggling if you're not ready for it. If you couple in playing multiple stake levels they get even worse; imagine what it could look like if each player took 400 of his hours off this graph and added in 400 hours at double the stakes, in a tougher and more aggressive game, where his win rate went up by only 60-80% (or not at all), but his standard deviation increased by say 125%. That's basically what I did in the second half of 2009, which turned out to be a complete disaster. I think I mentioned on here that at one point I had 200 hour stretch where I won exactly 1 big bet per hour but broke even, due to playing everywhere from a little 6/12 all the way up to 40/80.
I used to run these simulations using a true win rate of $50/hour, and admittedly they look a lot better when you do that. The constant tidal wave of positive expectation keeps the downswings and break even stretches much shorter; you're just absolutely crushing the game. But as you reduce that number towards the "accepted" maximum win rate, you start getting a lot of the artifacts you see here, with guys winning less than $20,000 for an entire year, or being under water for 1500 hours. Enjoy the toy.