Stock Price Signals from Online News
The impact of news articles on stock prices is under investigation in a lot of publications like “Analysing News for Stock Market Prediction” or “A quantitative stock prediction system based on financial news”. In a post some month ago I wrote about “Newstrace”, a new method to analyze how news may influence future prices.
In the articles “News history for stock prediction” and “A “Newstrace” analysis example with pinalyze.com” I used the service Pinalyze to analyze the possible impact of online news on stock prices.
In the second article, I showed an example of some news about digital outdoor advertising and its possible impact on Facebook stock prices. Pinalyze found some interesting patterns in the data with a significance level of about 5% error probability.
In this article, I look at another example with a much higher significance, and how we can visualize the signal with a density estimation plot.
We look at the following news:
Pinalyze first builds as we have seen in my older posts the “Newstrace” of similar past news like:
on 24 different days back until 2015–12–28.
The mean daily changes of Westpac Banking Corporation (WBK) on the “Newstrace” (days with similar news) are -0.157 lower compared with the mean of all days in the same time span. The significance level of this difference is 0.7%, therefore with high probability, the difference comes not from random.
The plot of the density estimation of the daily changes of this stock looks like:
The distribution and its mean for days with similar news to the inspected is shifted to the left and could, therefore, indicate a movement down of the stock price of WBK.
Other examples of news articles with positive or negative signals on some stocks can be found by using the www.pinalyze.com website. “Pinalyze” tracks the published news from some interesting sources and scans for signals.
