What the poker variance calculator simulates
The simulator breaks the sample into 100-hand blocks. For each block it draws a result from a normal distribution with your entered win rate and standard deviation. It repeats that process thousands of times, then reports the distribution of possible outcomes.
The default win rate is 5 BB/100, the default standard deviation is 80 BB/100, and the default sample is 50,000 hands. That creates expected winnings of 2,500 big blinds, but the final result can still vary widely. With those inputs, the theoretical final standard deviation is about 1,789 big blinds.
Being a winning player does not mean you will win over every sample
A player beating a game for 5 BB/100 has a real edge. That edge still has to pass through all-in pots, coolers, bad runouts, table selection, fatigue, rake, and normal card distribution. Over a short sample, variance can be louder than skill.
This is why bankroll management matters. A bankroll is not proof that you will win. It is the cushion that lets a reasonable edge survive the stretches where results do not match decisions.
How to read the percentiles
The median is the middle simulated outcome. The 5th percentile is a rough bad-case result where only five percent of simulations finished worse. The 95th percentile is the opposite side of the distribution. Downswing probabilities look at peak-to-trough drops inside the sample, not only the final result.