Hi, hi, i think is possible to get a pretty consistent algorithm that predict pokes price based on ivs and some other variables such as is S or not, needed ivs are optimal or not, is 20+, 25+, 28+ and probably others.
Is been like 3 years now that i'm trying to find good algorithm that can do this task but the main problem is lake of data, if is possible to create a database that contain some real pokes well labeled i think i can manage to adapt a perceptron algorithm in to some thing that can predict a pretty close price for a given pokemon ivs and national id.
But a major problem is the time evolution variat that will be hard to estimate essentially caused it will need to track all sales and predict how things are evolving, without a time track evolution is like being trapped in a specific moment in time, so prices will always be related to this specific moment.
But is possible to deal with this problem just by constantly adding new pokes into the data base and removing the old ones, or just update them all.
It can be a challenging project but still possible if done progressively.