market-sentiment-roberta-pro

Overview

Market-Sentiment-RoBERTa-Pro is a high-precision sentiment classifier for financial news, social media posts, and earnings call transcripts. It distinguishes between Bullish, Neutral, and Bearish market sentiments.

Model Architecture

Base Model: roberta-base
Data: Trained on the Financial PhraseBank and 5M+ scraped financial tweets.
Optimization: AdamW optimizer with a linear learning rate decay.

Intended Use

Real-time algorithmic trading signals.
Sentiment monitoring for hedge funds and retail investors.
Correlation analysis between social sentiment and price action.

Limitations

May misinterpret high-frequency trading sarcasm.
Context window is limited to 512 tokens; very long earnings reports should be chunked.

Example Code

from transformers import pipeline

sentiment_analyser = pipeline("sentiment-analysis", model="market-sentiment-roberta-pro")
tweet = "The quarterly revenue for $AAPL exceeded expectations, suggesting a strong growth trajectory."
print(sentiment_analyser(tweet))