Researchers in the USA have found a way to extract information from the well-known internet search engine, Google, that can be used to assist with understanding trading on the stock market. The approach follows, what the team refers to as “a long short-term memory approach”.
Writing in the International Journal of Financial Engineering and Risk Management, Joseph St. Pierre, Mateusz Klimkiewicz, Adonay Resom and Nikolaos Kalampalikis of the Worcester Polytechnic Institute, in Worcester, Massachusetts, explain how they have extracted Google search indices from a Google trends tracking website. This allows them to study the putative investor interest in stocks listed on the Dow Jones index (Dow 30). Essentially, they accomplish this task by using a long short-term memory network that finds correlations between changes in the search volume for a given asset with changes in the actual trade volume for that asset.
“By using these predictions, we formulate a concise trading strategy in the hopes of being able to outperform the market and analyse the results of this new strategy by backtesting across weekly closing price data for the last six months of 2016,” the team reports. In that proof of principle based on historical data they demonstrated a success rate of 43% and suggest that their algorithm would be scalable beyond the narrow scope of their study and so might be applicable to numerous other assets on the market.
The study begins by citing the received wisdom that the “market” cannot be outperformed and that any attempt to predict stock market rises and falls is effectively doomed to failure. However, given that some investors do regularly “beat” the stock market and make a profit, this conventional theory perhaps does not hold universally and there might be algorithmic methods that look at live data that might allow some investments to predictably outperform the market. They say that their 43% success rate is significant and worth exploring further.
St. Pierre, J., Klimkiewicz, M., Resom, A.and Kalampalikis, N. (2019) ‘Trading the stock market using Google search volumes: a long short-term memory approach‘, Int. J. Financial Engineering and Risk Management, Vol. 3, No. 1, pp.3-18.