No animations. Input any pets that exist.
Adjust enemy team settings and then run simulations as many times you want.
Battles are text based so theoretically you can run 100 fights per second.
Why? Why not, for science!!
Seriously though, it would be really interesting to find out what pet is “the best”.
I would love to run this if it were to exist.
Set 3 of the same pets in a team against a similar enemy team: 3 of a kind.
All are healed before each stand alone round, so it’s not a “marathon”, but always a new instance. The same pet, but only the abilities can be changed, any possible combo Aa/Bb/Cc.
After about 10,000 (or 100,000) fights you might have tested 1 pet (team) with all possible skill combos, against every pet and their skill combos.
Now you would have some very cool stats to look at:
Highest win rate settings, in general. Meaning “this pet team, if put against any 3 pet team (non elite/unique) will have the most success and highest win chance, in the long run”. Ofc if you are unlucky you might find yourself picking a rock fight as -paper- scissors, but that’s the exception to the rule.
If anyone wants to do a 3rd party browser based pet battle turbo simulator like this, please do one. Would be so cool to see all the data.
I could easily build the simulator to do the actual battles. The problem is developing an AI for the pets themselves.
Would they just use their abilities at random? Pretty easy to code this but it wouldn’t really prove much a strong pet using its weakest abilities or even its strongest ones at the wrong time would get crushed.
Would you want them to do what they do in game? This would involve quite a lot of research in game noting down what each pet does against certain situations. Unless this information was either provided by Bliz or datamined.
Would you want them to train like how Machine Learning AI train? Perhaps an open source AI could be used to do this training (I’ve never used such an AI). Where it would learn what was the best ability to use in each situation when up against each pet family. This could tell what the strongest possible pet was (although it might not act the same as it does in game).
The AI would still need a lot of data, such as all the pets and their stats and abilities in the different breeds.
