After writing that, I then kept firing like crazy at them, and missed almost every time …
But there’s something irritating about the hit chances and the number of multiple-repeat misses with chances in the 60s-80s. Some months ago, I decided to collect some data to prove that there was a bug. I recorded the hit chance of each attack, and sub-classified the attacks according to the type of attack (e.g. 1-tile melee, 2-tile melee or ranged), and recorded the outcome of each attack. Counting it up, there were definitely way more misses with high hit chances than there were hits with equivalent low hit chances, and yet (somehow) the average % of times hitting (for each sub-class of attack) matched up near-enough exactly with the average hit chance displayed when attacking …
Specifically, there were 15 total misses with a chance of 65% or higher, but there were 48 attempts, meaning that the successful hit % was actually about 68% …