You can deploy this model and start earning money today!

jialeCharlotte/
finbot

Name: finbot
Availability: InStock

deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B

finetune

mit

Views

No views yet

FinBot - Financial Sentiment Analyzer

This model fine-tunes the DeepSeek-R1-Distill-Qwen-1.5B model to analyze sentiment in financial news and reports. It classifies financial news into Bullish, Bearish, or Neutral sentiment categories based on the implied impact on specific stocks.

Model Description

Base Model: DeepSeek-R1-Distill-Qwen-1.5B
Task: Financial sentiment analysis
Training Technique: LoRA fine-tuning
Target Domain: Financial news, stock market reports, and company announcements

Intended Uses

This model is designed to help investors, traders, and financial analysts quickly assess the sentiment implications of financial news for specific stocks. It can be used for:

Analyzing news sentiment for portfolio management
Screening large volumes of financial news
Generating sentiment signals for trading strategies
Research on market sentiment and stock price correlations

Usage

You can use this model to analyze the sentiment of financial news by providing a news title, summary, and the stock ticker:

python

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
import json

# Load model and tokenizer
model_name = "jialeCharlotte/finbot"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, device_map="auto")

def analyze_sentiment(news_title, news_summary, ticker):
    prompt = f"""You are a financial analyst in a leading hedge fund. 
Analyze the sentiment of the following financial news for the given stock ticker step by step.

Title: "{news_title}"
Summary: "{news_summary}"
Stock Ticker: {ticker}

Step 1: Identify key financial terms and their implications.
Step 2: Determine whether the news suggests market optimism, pessimism, or neutrality for this specific stock.
Step 3: Based on your analysis, classify the sentiment into one of the following categories:
- "Bullish": If the news suggests confidence, growth, or positive impact on this stock.
- "Bearish": If the news suggests decline, risks, or negative impact on this stock.
- "Neutral": If the news is ambiguous or does not convey strong sentiment.

Finally, **return only** the final result in valid JSON format, with the structure:
{{
  "ticker": "{ticker}",
  "sentiment": "Bullish" | "Bearish" | "Neutral",
  "sentiment_reasoning": "Provide a brief explanation of the sentiment analysis."
}}

Do not include any extra text or explanations outside the JSON.
### Response:
"""

    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_new_tokens=200,
            do_sample=True,
            temperature=0.7,
            top_p=0.9,
            pad_token_id=tokenizer.pad_token_id
        )
    
    response = tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True)
    
    try:
        # Parse the JSON response
        result = json.loads(response)
        return result
    except json.JSONDecodeError:
        # If the response isn't valid JSON, return the raw text
        return {"error": "Failed to parse response", "raw_response": response}

# Example usage
news_title = "Apple Reports Record Q1 Revenue"
news_summary = "Apple Inc. announced today that they have achieved record-breaking revenue in Q1 2025, exceeding analyst expectations by 15%."
ticker = "AAPL"

result = analyze_sentiment(news_title, news_summary, ticker)
print(result)

Training Data

The model was trained on a curated dataset of financial news articles and reports, each labeled with sentiment classifications (Bullish, Bearish, or Neutral) specific to the mentioned stock tickers. The training data includes diverse sources of financial information covering various market sectors and company types.

Limitations

The model is focused on English-language financial content
Performance may vary for highly technical financial documents or specialized industry sectors
The model analyzes sentiment specifically for the mentioned ticker, not overall market sentiment
As with all language models, outputs should be reviewed by human experts for critical financial decisions

Ethical Considerations

This model is intended to assist with financial analysis but should not be the sole basis for investment decisions. Users should:

Always verify model outputs against other sources
Be aware that financial markets are influenced by many factors beyond sentiment
Consider the model's outputs as one of many inputs in a comprehensive analysis process
Not use the model for market manipulation or other unethical financial activities

Citation

If you use this model in your research or application, please cite:

@misc{finbot2025,
  author = {Charlotte Zhou, Zhilin Zhu},
  title = {FinBot - Financial Sentiment Analyzer},
  year = {2025},
  publisher = {HuggingFace},
  howpublished = {\url{https://huggingface.co/jialeCharlotte/finbot}}
}