We would like to allow infertrade package users to be able to evaluate large number of time series for prediction.
To do this we could create a script that sequentially calculates the performance of trading strategy using pairwise combinations (A versus B) for a set of time series supplied by the user. E.g. 10 cryptocurrency price series.
The output could be a Python object, ideally with an export method to CSV or similar.
The initial version could use only the infertrade package, but a later upgrade could be to integrate the API functionality.
This functionality is analogous to what is already available on InferTrade.com but will allow the users to do the screening locally.
To do list
[ ] Create tests for an interface similar to the below:
def evaluate_cross_prediction(list_of_dfs_of_asset_prices: pd.DataFrame, export_as_csv: bool = True):
"""A function to evaluate any predictive relationships between the supplied asset time series, with rankings exported to CSV."""
[ ] Add documentation to readme explaining how the function can be used
Explanation
We would like to allow infertrade package users to be able to evaluate large number of time series for prediction.
To do this we could create a script that sequentially calculates the performance of trading strategy using pairwise combinations (A versus B) for a set of time series supplied by the user. E.g. 10 cryptocurrency price series.
The output could be a Python object, ideally with an export method to CSV or similar.
The initial version could use only the infertrade package, but a later upgrade could be to integrate the API functionality.
This functionality is analogous to what is already available on InferTrade.com but will allow the users to do the screening locally.
To do list