I missed that it was per rank. I never used them. I always use 2 double rewards lenses and one to lock monolith. With the increased costs from lenses reroll cost is mostly irrelevant so I never needed to filter prophecy type.
In that case, the formula must be different. Otherwise, using 3 lenses of that type, you’d have more than 100% chance, meaning all of them would be guaranteed to hit, which I doubt is something that happens or even is possible.
Since you can’t have more than 100% chance of something happening in statistics (not talking about the %chance to apply bleed stacks which “overflow”, since that’s not statistics), and given how these things usually work, the likeliest thing that is happening is applying the bonus to the “reverse chance”.
Much like the way to calculate the chances of something with x% chance dropping y times in a row isn’t to apply x% y times but rather to apply 1-x/100 y times in a row, it’s likely that something similar is happening here.
The chance to not roll a sword is 90%. With the 58% bonus from the lens, you apply it to the rest of the prophecies. So rather than increasing the base 10% of the sword you reduce the 90% chance of getting anything else.
This would give us something like 52% to roll something else (which is then divided by the remaining 9 types) and we’d land at a 48% chance to roll a 1h sword.
Which is still a very high number and one that makes it unlikely we don’t get more swords.
So there’s either a different calculation being made (which is the most likely explanation) or there’s some actual bug there.
The chance of not getting a sword 20 times in a row if the sword chance is 48% is 0.002%, which is too unlikely to attribute to RNG.
I still don’t think it’s an actual bug. I think they’re doing some different math where the 480% increased chance has a low impact on the overall chances.
And even if it’s a bug, I’m guessing they’d rather deal with that after reworking CoF next season.
EDIT: reading back I realized I completely messed up the maths. I started using the 58% rather than the 480%. So ignore all that.
The gist is that the formula is likely decreasing the chance of other stuff first, so the overall increase to swords is actually much smaller. This is still what I think is happening.