TL;DR — In this article we introduce a few tools and techniques for studying relationships between the stock market and the news. We explore time se

Exploring relationships between news and market data using time series analysis

submited by

Style Pass

2022-07-05 19:30:05

TL;DR — In this article we introduce a few tools and techniques for studying relationships between the stock market and the news. We explore time series processing, anomaly detection, and an event-based view of the news. We also generate intuitive charts to demonstrate some of these concepts, and share the code behind all of this in a notebook.

The impact of news on markets has been an area of strong interest for many years, and with the advent of quantitative analysis, many from academics to quants have used statistical methods to better understand the relationship between news and market data, such as stock prices or trade volumes.

In this article we aim to provide a simple practical guide for exploring statistical relationships between news data and the market, as a foundation for deeper and more targeted analyses. Our goal is to provide a set of tools to the reader that will assist them in exploring and cross referencing financial and news data. Our goal is NOT to predict the stock market, or discover specific correlations or causal relationships between the news and the market.

In order to study the relationship between news and market data, we first leverage Aylien’s News API which allows users to filter the world’s news data, both real-time and historical, using a comprehensive set of filters. We use these filters to create news feeds that correspond to our securities of interest (e.g. stocks, ETFs, cryptos, etc). We then use Yahoo! Finance to retrieve the market data (price and volume) for these securities. Putting the two together, we can construct a joint dataset consisting of both market and news-related data for any given security, in a timeframe of choice.