Skip to main content
7 events
when toggle format what by license comment
S Mar 8, 2023 at 22:38 review First answers
Mar 8, 2023 at 23:10
S Mar 8, 2023 at 22:38 history edited Oly CC BY-SA 4.0
Update code slightly (previously pasted old code misusing `r` as `q`)
Mar 8, 2023 at 22:17 comment added Oly In the case I mentioned of a 'critical weighted' calculation, you'd get a single number counting the number of hits plus number of those which crit. Whether '8' means '8 hits, no crits', or '7 hits, one of which is crit', or something else, would not be specified but could be glossed or estimated if desired based on p_hit_given_crit.
Mar 8, 2023 at 22:14 comment added Oly A principled way to mix criticals into a single roll outcome could be to initially generate a distribution, not for the number of hits (a binomial), but for the number of critical-weighted hits (e.g. number of hits + number of crits if crits count as double). This would simply inflate the mean and variance of the distribution to be slightly larger. We'd add 1/20 * n_attackers to the mean, and 1/20 * (19/20) * n_attackers to the variance (the bit inside np.sqrt(...) in the calculation of std). Everything else as before, except the result would be critical-weighted hit count.
Mar 8, 2023 at 22:05 history edited Oly CC BY-SA 4.0
Add a bonus discussion on critical hits
S Mar 8, 2023 at 21:47 review First answers
Mar 8, 2023 at 22:21
S Mar 8, 2023 at 21:47 history answered Oly CC BY-SA 4.0