Complete list
Trading based on News Headlines using NLP
What is NLP?
Natural Language Processing (NLP), is a branch of artificial intelligence that deals with the interaction between computers and human languages.
It involves analysing and processing natural language data, such as text, speech, and conversation, to derive meaningful insights and automate tasks that would normally require human understanding.
VADER is a popular NLP tool for sentiment analysis.
What is VADER?
VADER (Valence Aware Dictionary and sEntiment Reasoner) is a tool used to generate sentiment scores from the text. It uses a set of rules and a lexicon of words to determine the positivity, negativity, and neutrality of a given piece of text.
Approach
Here's a general approach to trading based on news headlines.
The result of this general approach to trading based on news headlines on Apple stock from Jan 2014 to May 2016 is shown below.
The Python code for the strategy is available here. It is important to note that backtesting results do not guarantee future performance. The presented strategy results are intended solely for educational purposes and should not be interpreted as investment advice. A comprehensive evaluation of the strategy across multiple parameters is necessary to assess its effectiveness.
In this tutorial, we will discuss all the steps and methods involved in creating a trading strategy based on news headlines with the help of NLP.
Step 1: Collect News Headlines
Step 2: Generate the Sentiment Score
Step 3: Generate Trading Signals Based on Sentiment Score
Step 1: Collect News Headlines
To collect news headlines, you could use Python APIs like :
- Webhose,
- NewsAPI,
- News Fetch,
- GoogleNews, etc.
These Python APIs aggregate news headlines from various media sources.
The Python code to fetch news headlines from GoogleNews is available here.
Step 2: Generate the Sentiment Score
Once you have collected the headlines, you can use VADER, a tool to perform sentiment analysis on each headline.
How does VADER calculate the Sentiment Score?
The way VADER works is by assigning a score to each word in the text based on how positive or negative it is. It then combines these scores to give an overall sentiment score for the text.
For example, if a sentence contains the word "happy," which has a positive score, and the word "sad," which has a negative score, VADER will assign a higher score to the positive word and a lower score to the negative word. VADER takes into account the intensity of the words and their context to determine the overall sentiment score.
Overall, VADER is a useful tool for quickly analysing the sentiment of large amounts of text, such as news data on financial assets.
How to generate sentiment scores using Python?
NLTK (Natural Language Toolkit) is a popular Python library for NLP tasks. VADER is a sentiment analysis tool that is included in the NLTK library.
You can refer to this blog for a guide to performing sentiment analysis of news headlines using VADER in Python.
Step 3: Generate Trading Signals Based on Sentiment Score
After generating the sentiment scores of the news headlines, the final trading signal can be generated by considering the sentiment score of all news headlines collected over the data.
To generate trading signals, you can define a threshold for the sentiment score. For example, you can define a threshold for the sentiment score as 0.25, and generate a buy signal if the average sentiment score of news headlines collected over the day is greater than or equal to 0.25.
Similarly, a sell signal is generated if the average sentiment score of news headlines collected over the day is less than or equal to -0.25. No trade signal will be generated if the score is between -0.25 and 0.25.
The following graph represents the cumulative returns of the sentiment score-based strategy on AAPL stock between ‘2014-01-28’ and ‘2016-05-30’ using the trading signals mentioned above.
The Python code for the strategy is available here.
It is important to note that backtesting results do not guarantee future performance. The presented strategy results are intended solely for educational purposes and should not be interpreted as investment advice. A comprehensive evaluation of the strategy across multiple parameters is necessary to assess its effectiveness.
To review, the following is the workflow to generate daily trading signals based on news headlines.
Why limit the trading strategy to just the news headlines?
Can we generate the trading signals based on complete news articles instead of just the news headlines?
The answer is YES!!
The VADER is limited to generating sentiment scores just short text like news headlines. For longer text like a news article, the article should be broken into smaller parts. This can be done using the ‘sent_tokenize()’ method from NLTK.
Once the news article is divided into multiple sentences, the ‘polarity_scores’ method of the ‘SentimentIntensityAnalyzer’ class in the NLTK library is used to generate the sentiment score of each sentence. The average sentiment scores of all these sentences of a news article will be the final sentiment score of each news article.
Once the final sentiment score of each news article is calculated, define the sentiment class of each news article.
- The sentiment class of a news article is 1 if the final sentiment score is greater than or equal to 0.25.
- The sentiment class of a news article is -1 if the final sentiment score is less than or equal to -0.25.
- Similarly, the sentiment class of a news article is 0 if the final sentiment score is less than 0.25 and greater than -0.25.
Once the sentiment classes of each news article have been classified, find the average sentiment class of news articles collected over the day.
- A buy signal is generated if the average sentiment class is greater than or equal to 0.25
- A sell signal is generated if the average sentiment class is less than or equal to -0.25.
- No trade signal is generated if the average sentiment class is less than 0.25 and greater than -0.25.
The Python code to generate sentiment scores for an entire news article is available here.
To review, the following is the workflow to generate daily trading signals based on news articles collected over the day.
So, we discussed how to enter a trade based on the sentiment score generated on the news data.
When to exit the trade?
There are several possible exit strategies that you could consider for your trading strategy based on sentiment scores of news data generated by NLP and VADER. Here are a few options:
- Time-based exit: Set a fixed holding period for your trades to capture short-term movements in the market, such as a few days or a week. This would allow you to capture short-term movements in the market based on changes in sentiment.
- Profit-based exit: Set a profit target for each trade, based on your expected returns from the sentiment analysis. Once the price reaches this target, you would exit the trade and take your profits.
- Stop-loss exit: Set a stop-loss level for each trade, based on your risk tolerance. This would limit the losses if the sentiment reverses.
- Reversal-based exit: Use a reversal signal to exit your trades when sentiment changes from positive to negative or vice versa, to capture trends in the market and avoid losses due to sudden sentiment changes.
Learn more about different methods to analyse the sentiment of news data and design trading strategies by following the steps below.
What to do next?
- Go to courses Natural Language Processing in Trading and Trading Strategies with News and Tweets.
- Click on Free Preview
- Go through 10-15% of the course content
- Drop us your comments, and queries on the Quantra community.
IMPORTANT DISCLAIMER: This email is for educational purposes only and is not a solicitation or recommendation to buy or sell any securities. Investing in financial markets involves risks and you should seek the advice of a licensed financial advisor before making any investment decisions. Your investment decisions are solely your responsibility. The information provided is based on publicly available data and our own analysis, and we do not guarantee its accuracy or completeness. By no means is this communication sent as the licensed equity analysts or financial advisors and it should not be construed as professional advice or a recommendation to buy or sell any securities or any other kind of asset.
