Flame Sergeant's Spear VS Ifrit's Harpoon

[Silvano]

Seiken doesn't even count trials when he tests. We use a different method of data collection for MIN and MAX. Most other testers just find some arbitrary trial # and use the average. We only document the MIN and the MAX until:

(MAX - ((MAX+MIN)/2)) / ((MAX+MIN)/2) = 7.8% or greater

Because this gives the true average via calculation. The number of trials to achieve this true min/max depends on how high a damage you're dealing and luck. Could take 2 trials, could take 1000. The point is we're finding the true average instead of leaving it up to chance.

Is there a cap on this percentage where it is no longer effective for comparison? I've been trying to keep it around 8%, but for instance, Ifrit had an 86% deviation for my auto-attack lol.