How I Use Alternative Data: An Institutional Trader’s Perspective

February 15, 2018 09:00 AM
Every tool has a specific use and it is important to apply the proper alternative dataset to its best use.

The goal in trading is to get ahead of the herd. It’s best to anticipate what’s going to happen in the next five to 60 days and focus on major events. Is the top analyst going to upgrade a stock before earnings? Will global macro traders try to position hawkish or dovish around the next FOMC statement? Will the next crop report show more supply or demand? These are the questions; the answers to which will move markets.  

This is never easy, but alternative data supplements solid research and gives insights and clues of an event outcome. Earnings are released quarterly, and traders need to anticipate earnings before the event. There are many questions that need answers: What data points can we use to help form a prediction? Do we go long or short? What’s our conviction? What data do we use?

In December 2017, I was researching stock earnings plays and looking for trades. My typical process is to look at the public financials and technical profile to get a view of the widely available information. If the stock looks interesting and passes some general filters, I dig deeper and start looking for alternative data to help predict the event outcome. In this case, I am searching for clues about earnings momentum.

There are many alternative data indicators and tools to choose from. My goal is to complement my fundamental and technical research with the most predictive alternative data. It is important to match the appropriate alternative datasets to the fundamental analysis you are researching. For example, never use App downloads to predict auto sales.  However, it would be appropriate to use data on “new insurance VIN auto sales” or social media indexes from Twitter (TWTR) or “product trend shifts” from mobile geolocation services. There is a lot of noise in alternative data. The only way to trust the data and confirm predictive power is through data science analysis. You have to do the work.

Here’s a good place to start. Every stock or commodity or sector has four to five leading factors. Look for data and clues around those leading indicators. My favorite alternative data tends to look for “tells” or statistically significant predictive power of stock returns lagged five to 60 days. There is not a one-size-fits-all approach to analysis. Adapt the research process to your trading strategy goal. 

We call it the quest for alpha or edge research for a reason. Portfolio managers at quantitative hedge fund are constantly researching data.

Most of the research conducted, 95%, is never used in live trading. With research, one should expect a lot of failed projects. While that may seem discouraging, it only takes one or two great indicators to produce millions in profits and outsized returns. 

Expect to be discouraged frequently. Always remember, it only takes two to three great trades in a career to be considered a legend.   

About the Author

Chris Randle is a proprietary global futures and commodity trader. He focuses daily on short term relative value and momentum trading. He also has investment projects in FinTech, Crypto, Alternative Data and AI. Twitter @crandl