in order to find the most stat-efficient combinations, the algorithm needs to sort through
every possible combination (up to 6 items!) while also applying factors like bonus stats from passives
(for instance the 3% power -> mana bonus from Transcendance). A faster "greedy" algorithm that only
tries to find the best stat items one at a time may not actually produce the most optimal result due to
this. i wrote this generator specifically to experiment with the new ratatoskr acorns synergy with other
tanky items
if you want to speed up your search times, try adding in a few items manually before starting. Each item
multiplicatively speeds up the search. the search sorts its results as it operates, only holding onto
the amount of results you specify while it works - so it doesn't eat up inane amounts of ram generating
every possibility. you can safely set the value to some really high values without a speed impact
if you want you can , and
the site will ping you when the search is done
no not really. there are a few notable things about how this generates results:
your effective hp is a figure representing how much damage it would take to kill you,
accounting for both your hp and your armors. when sorting for high defense stats, the system uses these
rather than the raw protection stats of the items. for example, at 100 armor, you negate 50% of damage
taken (of that type) - so if you have 1500 hp, your effective health is 3000. this is why there are two
numbers: red(first) is physical, blue(second) is magical. you can read more about
damage scaling in the wondrously comprehensive word of thoth for
which the formula was derived from
the 'relative score' of a build does not mean much actually. its simply a value generated for the build
while sorting based off the parameters of the search. its not a description of how 'objectively good' a
build is or isn't - rather how well it matches the parameters of the search. the same build can have
drastically different scores if you provided different search criteria
in a sense the framework for it is already in place - so it could be likely. my plan is to do more work
for now on ratatoskr specific models, such as estimating damage from auto-attack chains and abilities,
to give a more realistic view of the damage output for the character. it would be way too much on my own
to try and model that for even a few other characters, let alone all the others, especially considering
the fact that i (personally) dont have an interest in playing any other character. ratatoskrs new acorns
also have the benefit of allowing him to have a faster overall search time, since one item slot only has
4 options. if
you are currently using this site to test builds for other physical dmg characters, please note that the
stat totals listed include the lvl.20 base stats for ratatoskr (def: 72,
mdf: 48,
mit: 0,
hp: 1880,
mana: 1060)
right now the app isnt connected to the smite api database to retrieve item info since the way data is
stored is 'opinionated'. so adding support for other characters / classes is largely a matter of the
data formatting necessary to get them usable by the algorithm. if you wanna help add more data, or look
at my terrible, strenuous sorting system, feel free to reach out on the project github