jamesflynn1 / finitemarketmodel Goto Github PK

Expose the best models for a final evaluation stage. Compare models against the binomial model.

Add more theory and mathematics behind theory

Similar to the binomial part

Select best hyperparameter on the validation set (use Wasserstein distance), for each stock and use this for evaluation stage.

Wasserstein distance might not be the most suitable to compare between the empirical distribution.

Ensure adequate comments and layout parts again

Describe the binomial model in some detail, possibly omit details surround the trading strategies.

Multiply instead of add.

• Autocorrelation (line graph).
• Models selected.
• Visualising t-test performance.
• Visualising selected variance test performance.
• Visualising Wasserstein performance (test)
• Visualising Wasserstein performance (train vs validation)
• A few example distributions
• Comparisons to binomial model

https://eodmetrics.com/query perhaps, match the end data of the validation period.

Likely rolling cross validation.

Within the validation phase, rank order the models in terms of their projected accuracy on the testing data.

Explore the process with one chosen stock