Hi folks, it's been a while since I posted here, but I thought I'd mention that I've done a new simulation of Pathfinder combat, this time to generate decision rules for creating fighters. I ran a million completely random simulations and applied a method called Classification and Regression Trees (CART) to identify groups of feats, ability scores and weapon choices that might work together to improve survival against our favourite orc, Gruumsh. Results are here. There's even a flowchart! (And a promise of more to come).
The results mostly support previous research on the issue, but it does appear that the people with the best survival chance are high strength, high constitution dwarves, possibly regardless of feat choice.
At this point I'm drowning in data ... ideas and suggestions would be welcome!
|Lamontius Marathon Voter 2014|
From your methods:
For example, 47% of those with improved initiative survived, compared to 51% of those without; and 46% of those with shield focus, compared to 52% of those without. This probably represents the opportunity cost of choosing these feats, or some unexpected confounding effect from some other variable.
Any speculation on what the confounders could be? What were the stats of the Orc?
I am also curious if you could model survivability as a function of combat rounds.
Interesting, will mull over your results some more... after my midterm. :P
Lamontius: these are the kinds of rare and unique insights that only statistics can provide!
Actually I'm wondering if I should do restricted analyses to, for example, a PC with high dex and low strength and con, to see what their best feat selection is to maximize survival. It might be more useful than just analyzing the entire set of data, which gives results dominated by the effect of ability scores.
I'm also pondering what difference the maximum HPs at first level rule makes to feat selection. Maybe choice of feat and fighting style is more influential (compared to "high strength wins!") when HPs are low. As it stands there is very little chance that a PC will die on the first blow from an orc, which makes e.g. improved initiative very low value. But if you have 1 hp, maybe getting the first strike in becomes more important...
CFet, I can't figure out what the confounders might be. It's a fully random, balanced data set, so there shouldn't be any undue effect of confounders in any particular variable. So I guess those high death rates represent the opportunity cost of choosing that particular feat.