As a start we tried calculating price multipliers of px(size x) / px(size 9) for each model.
It appears there is a large variance multipliers itself: we see flat rate across present sizes, tail sizes being cheaper or more expensive than benchmark 9 prices.
Presumably flightclub functions in how many pairs remaining when deciding the price of a size. Without scraping such it'd be dangerous to just make inference with a fixed multiplier or assume sold-out sizes can be sold as the most expensive size right now.
Data analysis in arbitrage_locator/fc_price_inference.py.
This, albeit promising, would be blocked until further volume / remaining_cnt collection from flightclub.
As a start we tried calculating price multipliers of
px(size x) / px(size 9)for each model. It appears there is a large variance multipliers itself: we see flat rate across present sizes, tail sizes being cheaper or more expensive than benchmark 9 prices.Presumably flightclub functions in how many pairs remaining when deciding the price of a size. Without scraping such it'd be dangerous to just make inference with a fixed multiplier or assume sold-out sizes can be sold as the most expensive size right now.
Data analysis in
arbitrage_locator/fc_price_inference.py.This, albeit promising, would be blocked until further volume / remaining_cnt collection from flightclub.