qwakes project setup on data from polygon and get something basic working.
Use this also as a way of validating the stochadex analysis package is fit for purpose.
Goal: Build a stock portfolio management system which can do backtesting by looking at historical actions/shifts in portfolio positions as training data for price impacts.
Project todos:
Use the Polygon Go client (https://github.com/polygon-io/client-go) to create functions for reading stock market data and loading it into simulator.StateTimeStorage objects
Create a custom Q-Hawkes simulation iterator and some convenience functions for running them with the market data
Develop a custom online inference methodology and mechanism for cross-validation against the data, then create some custom functions for this
Create a portfolio agent iterator which acquires positions in the market and affects the market liquidity (and possibly price) through its actions via the Hawkes model
Develop some more convenience functions for backtesting portfolio configurations - Obviously need to reconcile simulation disparity with historical data…
So is it possible to use the historical trade data as possible actions/shifts in portfolio positions?