You only run 500 samples in your simulation. You can see if you repeatedly hit the submit button that the resulting DPS estimation jumps around quite a bit. With the default values and hitting submit a dozen times I got values from 28 to 33 DPS. Something from 50,000 to 1 million would be more reasonable imo, depending on the impact to runtime.
It does nicely illustrate, though, that if you want to distinguish between two sets of weapons with DPS values that are within 2-3 DPS of each other, you need more than 500 samples to accurately tell which one is better.
|