DookiCape hitrate report anomaly
The highlighted comment was created in this revision.
I have been using DookiCape as the movement system for a bot I am working on while I test the targeting. It has been working fine until just a little while ago, when It seems to be calculating the normalized hit totally inaccurately causing the flattener to not be enabled. There may be other problems, Shadow and other top bots get around a 20% hitrate vs it , instead of around 11% before. I didn't change anything movement wise so I have no idea whats happening... Ideas anyone?
Well, first, how many matches are you running to calculate these hit rates? I could easily see that much variance across just a few matches.
Reviewing the code... The only thought I have is if you did something to screw up detection of the enemy firing bullets, their shots fired would be lower and their hit percentage would be higher.
But I also think the normalized hit rate may be functioning correctly and the result is just not what you expect. :-) It takes distance and max escape angle into account, so theoretically it's immune to things like different distances and bullet powers, but guns and movements still have strengths and weaknesses. To get within a percent or two, I'd probably want to run about 10-20 matches if I saw variance like that, if it is actually just variance.
