Model Information
Model Type:
(Waiting For Data)
Test Loss:
(Waiting For Data)
RMSE:
(Waiting For Data)
r2 (Naive Model):
(Waiting For Data)
r2:
(Waiting For Data)
RMSE (Naive Model):
(Waiting For Data)
Train Loss:
(Waiting For Data)
Beat Naive Model?
(Waiting For Data)
Features Used For Training:
(Waiting For Data)
Next Trading Day Predicted Closed Price:
(Waiting For Data)
Recent News
Positive Sentiment
Neutral Sentiment
Negative Sentiment
*Sentiment can sometimes be classifed incorrectly (≈ 91% accuracy)*
Date
(Waiting For Data)
Date
(Waiting For Data)
Date
(Waiting For Data)
Date
(Waiting For Data)
Date
(Waiting For Data)
Date
(Waiting For Data)
Date
(Waiting For Data)
Date
(Waiting For Data)
How to Use
Use the dashboard to predict stock prices by:
- Choosing a stock ticker.
- Selecting a timeframe (at least one month).
- Picking features for the model.
- Pressing "Load Data".
Your queue position is shown in the navigation bar if another user is currently loading data.
Insights
Our analysis shows:
- LSTM models are not more accurate than naive models (previous day's closing price).li>
- Price-related data alone is insufficient for accurate predictions, supporting the Efficient Market Hypothesis. (Sentiment isn't used in the LSTM due to resource issues and is shown to display my entity sentiment model.)
- Many correlated features do not enhance prediction accuracy.
More information, along with further insights about this project can be found at Stock Market Predictor
Future Plans
Planned enhancements include:
- Integrating unrestricted historical news data.
- Improving headline scraping efficiency.
- Refining the sentiment analysis model, currently at 92% accuracy with a fine-tuned BERT model.