Model Card for SentimentBasedOnPriceVariation

This model was created by fine-tuning the base model on a dataset containing summaries of stock market news along with the corresponding price variations. It is capable of extracting price movements from the news, making it a useful tool for detecting reactions in less closely monitored markets. I also compared it with FinBERT, and the results are quite similar. Interestingly, it can detect the sentiment of financial news, even though it was trained using a different methodology than the one used with FinBERT. Also, as we know, not all stocks react the same way to a specific event. That’s why I created additional models tailored for individual stocks. You can find them on my profile, along with the datasets used. (SelmaNajih001/PricePredictionForTesla and SelmaNajih001/PricePredictionForMicrosoft) This particular model is designed for general news — try it with phrases like “The stock market crashed” or “Tesla’s price dropped.” When evaluating news related to investment opportunities, this general model might provide a neutral score, whereas the stock-specific models can estimate potential price variations that could occur after such events. The predictions are tailored to each individual stock. This model is effective at extracting price movements directly from news. However, if you are interested in predicting price changes for specific events, I recommend using the stock-specific models available on my profile

Model Description

• Developed by: Salma Najih
• Model type: Text-Classification
• Language(s) (NLP): EN

Uses

This model is designed to extract price movements from financial news.

Direct Use

The model can be used directly to estimate price movement signals from news headlines or summaries. Users can input general market news or specific company news, and the model will return a predicted price movement direction or sentiment.

Results

The accuracy was about 97%

How to Get Started with the Model

Use the code below to get started with the model.

from transformers import pipeline

pipe = pipeline("text-classification", model="SelmaNajih001/SentimentBasedOnPriceVariation")
pipe("Apple Stock Rises On New U.S. Investment Commitment")