Reverse Engineering Stat Formulas for an Old RPG

No.13798317 ViewReplyOriginalReport
What's a way I can derive a function if I have the inputs and the outputs across multiple samples?

I'm looking to figure how how to calculate formulas/equations for how various stats are calculated in an old RPG that barely anyone plays anymore called Shadowbane for my own interest.

The player is able to see stats such as "strength, int, con, etc.", their gear score, various other traits, etc. but the calculation for how defense is finally computed is done on the server-side of this game. Further, the game is currently maintained by a Chinese company (ChangYou) which has not responded to my e-mails about how these calculations are done.

So it seems like my only option at this point is to take a lot of data with inputs (stats/skills/traits) and output (final defense score, final attack rating score, etc.). I barely remember solving linear systems of equations in school and I don't really even know if that's the best way to do this.

My process for isolating for which stats are actual inputs to these functions is to isolate by removing all buffs, removing all gear, and one-by-one putting on gear that modifies one attribute at a time to see if that stat does/does-not contribute to the scores I'm interested in. Then when I have isolated all of that, I add differing quantities of that same stat to observe if the rate of change is linear or some kind of exponent (I literally arrange it like ax^2 + by + c = 0). Then I use a python script to save all of my inputs and outputs to a CSV and cross compare different samples to see if my function works for both sets.