Optimal Position

For supervised training where we would like our neural network to learn when to long or short, we would create the optimal position at any given point in time.

There is a class ml.optimal_position.OptimalPositionGenerator that allows us to do just that.

Usage:

import kydb
from datetime import date
from ml.optimal_position import OptimalPositionGenerator

db = kydb.connect('dynamodb://epython/timeseries')
ts = db['symbols/bitflyer/minutely/FX_BTC_JPY']
opg = OptimalPositionGenerator(ts, start_date, end_date)
optimal = opg.generate()

Picking some random 3 days windows we would see that if we manage to trade like that, we’d be laughing.

Green is buy and red is sell.

../_images/optimal_strategy.png

See Optimal Position Notebook for more details.