{}
Model Card for Model ID
The Hybrid Image-Numerical Stock Prediction Model is a multi-input neural network designed to predict binary stock price movements (up/down) based on historical chart images and technical indicators. It processes both image data (stock charts) and numerical data (technical indicators like RSI, MACD, Bollinger Bands, ATR, and OBV) to make predictions.
This modelcard aims to be a base template for new models. It has been generated using this raw template.
Uses
Direct Use
The model can be used for predicting short-term stock price movements based on historical data. It is intended for use in financial analysis when combined with other tools and strategies.
Bias, Risks, and Limitations
The model may exhibit biases based on the training dataset's characteristics. It may not perform well during sudden market shifts caused by external factors. Users should be cautious about over-reliance on the model's predictions without additional analysis.
Recommendations
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
How to Get Started with the Model
Training Details
Training Data
The model was trained on the Kaylah072001/Stock_Graphs_Dataset, which includes stock chart images and corresponding price data. Technical indicators were extracted using the ta library.
Training Procedure
Preprocessing [optional]
Image data was rescaled to a range of 0–1 and augmented using rotation, shift, zoom, and flip techniques. Numerical features were extracted dynamically during training using technical indicator calculations.
Training Hyperparameters
- Training regime: Training regime: Mixed precision (fp32) Optimizer: Adam Loss function: Binary crossentropy Metrics: Accuracy, AUC-ROC Cross-validation: 5-fold stratified k-fold
Testing Data, Factors & Metrics
Testing Data
[More Information Needed]
Factors
Performance was evaluated across different market conditions and stock types.
[More Information Needed]
Metrics
Accuracy