Equity-data triumvirate: price, volume and news sentiment!

Historically, stock analysts and portfolio managers have relied on two kinds of information: a) high-velocity data, i.e., price and volume and b) company (earnings) reports and SEC filings, the bulk of which has a quarterly frequency. With the new, emerging frontier of computer-driven news analysis – using AI – high-velocity data can be broadened to include news quantity, quality and sentiment. Analysts can use price, volume and news summary data (e.g., number of snippets, number of positive snippets, positive snippet-change in the last week) as a three-pronged platform. The following recent examples – INTC and WMT illustrate the interplay among the three data components:

Trading volume for INTC spiked on 10/27/2017 driving the stock price to $44.14 (which eventually peaked at $46.82 on 11/2/2017). Note that the positive (news) snippet count increased to 68 on 10/26/2017 and peaked to 82 on 10/27/2017.

Trading volume for WMT peaked on 11/16/2017 with the stock price surging to $99.62 (following a great earnings report and optimistic business update). Positive (news) snippet count rose sharply to 58 on 11/14/2017 and peaked to 155 on 11/16/2017.

In our view, automated (near) real-time news analysis further expands the universe of quantified data available to security analysts and enables a richer set of analysis for rapid and actionable changes to portfolios or trading strategies. Much like analysts and quantitative modelers use price and volume in many creative ways, we believe a number of new models will emerge based on the quantified news data provided by Stocksnips!

Examples of use-cases / alerts that can be tested with this triumvirate include:

Negative price momentum, tapering volume and news-sentiment bottoming (and starting to uptick) as a potential signal for going long or covering a short position.
Positive price momentum, tapering volume and news-sentiment peaking (and starting to downtick) as a potential signal for selling a holding or initiating a short position.
Sideways price action, increasing volume and divided news sentiment (i.e., almost equal number of bearish and bullish snippets). This could indicate a battleground stock, for which volatility is expected to increase.

Multi-factor models can be built and backtested using high-velocity data (price, volume and quantified news), providing a “richer and more-complete technical analysis” paradigm like the alert examples above. It can also be suitably combined with fundamental factors such as price-to-book, firm size and cash flow growth. In summary, additional, alternate data (especially those derived from machine-read news) has the potential to make the alpha quest more facile and robust.