Introducing
Sep 6, 2024
In this tutorial, we explore the Ticker News API, enhanced through our internal research and the insights from our recently published paper on sentiment analysis using Large Language Models (LLMs). This API now captures structured data from unstructured financial news, enabling precise sentiment analysis tagging directly tied to specific tickers.
We'll demonstrate the practical application of these capabilities by examining sentiment trends for tickers such as
The research from Polygon.io introduces a groundbreaking approach using Large Language Models (LLMs) to extract structured data from unstructured financial news. By combining the generative capabilities of LLMs with advanced prompting techniques and a robust validation framework, the system can accurately identify relevant company tickers and perform sentiment analysis at the company level from raw news content.
The evaluation showed high effectiveness, with the new methodology surpassing traditional data providers in both coverage and relevance, providing a granular, real-time analysis via an accessible API.
The Ticker News API has long been a cornerstone for accessing financial news associated with specific tickers. However, we have now enhanced this API by integrating insights that provide structured data from this unstructured news content. This feature adds valuable layers of sentiment analysis tagging, ticker identification, and summarization directly into the news data, enriching the information available to investors and analysts.
Before you start, make sure to obtain an API key by signing up at Polygon.io. In the following example, you can retrieve news articles for a specific ticker using the Polygon.io Python client library:
from polygon import RESTClient client = RESTClient() # Fetch news articles for 'AAPL' with insights news_articles = client.list_ticker_news( "AAPL", params={"published_utc.gte": "2024-09-03"}, order="desc", limit=1000 ) # Display the title and insights for each article for article in news_articles: print(f"{article.title} [Insights: {article.insights}]")
The response from the Ticker News API now includes detailed
Title: Apple, Nvidia Key Supplier TSMC Leads Taiwanese Chip Giants To Localize Neon Gas Production By 2025 Insights = [ Insight( sentiment="positive", sentiment_reasoning="TSMC is leading the initiative to localize neon gas production, which is a significant step towards enhancing supply chain reliability for the Taiwanese semiconductor industry.", ticker="TSM", ), Insight( sentiment="positive", sentiment_reasoning="UMC is considering purchasing locally produced neon gas from Linde LienHwa, further supporting the initiative to localize neon gas production.", ticker="UMC", ), Insight( sentiment="positive", sentiment_reasoning="TSMC, a key supplier for Nvidia, is leading the effort to localize neon gas production, which will benefit Nvidia's supply chain.", ticker="NVDA", ), Insight( sentiment="positive", sentiment_reasoning="TSMC, a key supplier for Apple, is leading the effort to localize neon gas production, which will benefit Apple's supply chain.", ticker="AAPL", ), ]
This updated API functionality not only provides traditional news retrieval but now offers enriched data that can significantly enhance financial analysis and decision-making processes.
Now, let's dive into a hands-on example of actually using the Ticker News API. The script we'll explore is designed to systematically analyze market sentiments by focusing on specific tickers, such as
The script begins by fetching news with insights for a specific ticker using the Polygon.io Python client, spanning a chosen date range. It then organizes this data using the
from polygon import RESTClient import pandas as pd import matplotlib.pyplot as plt import matplotlib.dates as mdates client = RESTClient() # Set the date range start_date = "2024-06-01" end_date = "2024-09-04" # Fetch news and extract sentiment for 'CRWD' sentiment_count = [] for day in pd.date_range(start=start_date, end=end_date): daily_news = list(client.list_ticker_news("CRWD", published_utc=day.strftime("%Y-%m-%d"), limit=100)) daily_sentiment = { 'date': day.strftime("%Y-%m-%d"), 'positive': 0, 'negative': 0, 'neutral': 0 } for article in daily_news: if hasattr(article, 'insights') and article.insights: for insight in article.insights: if insight.sentiment == 'positive': daily_sentiment['positive'] += 1 elif insight.sentiment == 'negative': daily_sentiment['negative'] += 1 elif insight.sentiment == 'neutral': daily_sentiment['neutral'] += 1 sentiment_count.append(daily_sentiment) # Convert to DataFramesentiment_count df_sentiment = pd.DataFrame(sentiment_count) # Convert 'date' column to datetime df_sentiment['date'] = pd.to_datetime(df_sentiment['date']) # Set the date as the index df_sentiment.set_index('date', inplace=True) # Plotting the data plt.figure(figsize=(20, 10)) plt.plot(df_sentiment['positive'], label='Positive', color='green') plt.plot(df_sentiment['negative'], label='Negative', color='red') plt.plot(df_sentiment['neutral'], label='Neutral', color='grey', linestyle='--') plt.title('Sentiment Over Time') plt.xlabel('Date') plt.ylabel('Count') plt.legend() plt.grid(True) # Format the x-axis to display dates better plt.gca().xaxis.set_major_locator(mdates.WeekdayLocator(interval=1)) # Adjust interval as needed plt.gca().xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d')) plt.gcf().autofmt_xdate() # Rotation # Saving the plot as an image file plt.savefig('sentiment_over_time.png') plt.show()
I've executed the script twice, selecting both
For CrowdStrike, the graph displays a significant spike in negative sentiment following the global outage caused by a software update. This visual highlights the immediate and substantial increase in negative press, reflecting the severity of the impact across multiple industries.
In contrast, the graph for NVIDIA illustrates a sustained increase in positive sentiment over an extended period, coinciding with the company's advancements in AI technology. This trend underscores the favorable media coverage that NVIDIA has received, bolstered by its pivotal role in driving AI innovations.
These visualizations underscore the distinct sentiment dynamics that can emerge from different industry events, providing a factual, data-driven basis for strategic decision-making in financial markets.
In this tutorial, we explored the enhanced Ticker News API, now with insights, that offers advanced sentiment tagging capabilities. We also explored how sentiment analysis works using the
Happy exploring!
Team Polygon
editor
See what's happening at polygon.io
This tutorial demonstrates how to detect short-lived statistical anomalies in historical US stock market data by building tools to identify unusual trading patterns and visualize them through a user-friendly web interface.
editor
Explore how Polygon.io's enhanced Ticker News API leverages advanced sentiment analysis for strategic market insights, featuring practical applications with real-world data.
editor
In this tutorial, we'll explore the capabilities of Polygon.io’s Related Companies API, where we learn how to identify and visualize intricate corporate relationships using Python and vis.js.
editor