Stats for Pi Gamma

[Anton-Yakovenko]

Add collection statistics to the trader and integrate them with Pi Gamma Bot for display purposes.

List of stats to include:

• Current account balance
• Difference from the initial state
• Total losses and earnings

[!NOTE] The list may be expanded in the future with additional statistics.

[Anton-Yakovenko]

Potential aspects for statistics:

1. Prediction Accuracy Metrics:

   • Overall accuracy of the neural network model.
   • Precision, recall, and F1-score to assess model performance.
   • Mean squared error (MSE) or root mean squared error (RMSE) for regression-based predictions.

2. Model Performance Over Time:

   • Graph showing how prediction accuracy has evolved over time.
   • Comparison of accuracy between different versions of the model.

3. Market Analysis:

   • Historical price data for the cryptocurrency being predicted.
   • Correlation between predicted prices and actual market prices.
   • Frequency distribution of prediction errors.

4. Feature Importance:

   • Ranking of the most influential features in the neural network's predictions.
   • Visualizations of feature importance to understand their impact.

5. Model Health:

   • Training and Validation Loss: Display the loss curves during the training and validation phases. This gives users insights into how well the model is learning.
   • Overfitting Analysis: Show whether the model is overfitting or underfitting the training data, as this can affect its generalization ability.

6. Confidence Intervals:

   • Prediction Confidence: Provide confidence intervals around the predicted prices to indicate the level of uncertainty.

7. Backtesting:

   • Backtest Results: Show the results of backtesting the neural network's predictions against historical data. This helps users understand how the model would have performed in the past.