search

UnownHash / gohbem

PvP Companion
Apache License 2.0
5 stars 2 forks source link

Make Tied Ranking Optional #9

Closed plinytheelder closed 1 year ago

plinytheelder commented 1 year ago

Add config to make tied ranking optional. Why?

lenisko commented 1 year ago

Hey! Thanks for idea. As mentioned on the other service, I don't plan to work on it for longer time. If anyone wants to pick it up, feel free, I'm open for customization with current defaults.

plinytheelder commented 1 year ago

maybe @Fabio1988?

lenisko commented 1 year ago

That's a great idea! Get him to stop working on a project for everyone to fulfill your desire. I'm sure the rest will be grateful as well! /s

plinytheelder commented 1 year ago

Just pointing out a bug in the pvp logic that feeds into the main project. Don't think anyone wants Golbat to be inferior to RDM in any aspect, which it currently is due to ohbem's pvp restrictions.

Mygod commented 1 year ago

Hey, would you like to also add an option of specifying how to rank tied mons? Maybe have an option of array of sorting things, like first by stat product descending, then by cp descending, then by atk IV descending, then by def IV descending, then by sta IV descending, then by level ascending? Maybe we should also allow people to sort stat product ascending because having fewer options would make it inferior? :)

plinytheelder commented 1 year ago

sorting by statproduct then CP to align to the most used PvP calculators seems sufficient :)

again, only an option that is up to the mapper. both are correct.

lenisko commented 1 year ago

Just pointing out a bug in the pvp logic that feeds into the main project. Don't think anyone wants Golbat to be inferior to RDM in any aspect, which it currently is due to ohbem's pvp restrictions.

Welcome to Open Source world! You are unhappy with current implementation, that happens :) People either create PR or Fork a project if owner is against merging. But! as I noted before, I don't mind extending API if it won't break current flow and is configurable. That still doesn't mean we are all going to jump into this, we all have better and much, much more important things to do. Feel free to stay on RDM if you are going to complain about PvP results, nobody will force you to change the software you use.

Mygod commented 1 year ago

Hold on a second. You said "have to settle", implying that there is an inherent difference between the two tied rankings. How is this true actually? (Especially if all their stats are equal, which is the most common case.)

On the flip side, I just realized that maybe we should sort by stat product descending, and in addition then by attack descending due to CMP (charged move priority) ties.

Mygod commented 1 year ago

Also from my understanding, the thing with PvP ranks (except for master league) is not that simple. Since your CP is capped, there could be many IV combinations that each have their pros and cons. Indeed, in the real world (both pro players and IV deep dive analyses from various websites), people sometimes prefer rank 2 over rank 1 due to winning CMP ties in the mirror. I can imagine that due to this, the Nash equilibrium of the PvP meta might involve using different IVs of the same Pokemon each with varying probabilities. At the end of the day, sorting by stat product is only a heuristics that often work but won't necessary give you the best of the best. (In fact, you probably need many different IV combinations and use one of them at random to be best of the best.)

With that said, some IV combinations are inherently worse than others (and thus would never be used in a Nash equilibrium state). There are also some IV combinations that would be outclassed by a possible level cap increase in the future. See also: https://github.com/Mygod/ohbem/issues/1, https://codepen.io/Mygod/full/eYgoGXO

Mygod commented 1 year ago

Anyway those comments above are only addressing your third point. As for the second point you raised, the confusion is caused by a disagreement between GO stadium and Ohbem(go), and for the reasons that I explained above, for PvP functionally rank 1 (by which I mean all 3 stats being equal to another IV combination) is really rank 1 since you are not going to use that Pokemon for anything else really (like powering it up to a different level), unless I am missing some use cases here? Maybe having a 67% IV would make one feel better than having a 64% IV, but then they are debatably not really just tracking a "perfect IV for great/ultra league" any more. (And as I alluded to earlier, there is not a simple formula for "perfect IVs" for great/ultra/little league.)