Bucketing Alternative Data

February 15, 2018 08:00 AM
The variety of alternative data is huge and growing, but can be placed in five general buckets.

London-based alternative data broker Eagle Alpha has listed 24 primary categories of alternative data. But, as you can see, the types of datasets are increasing all of the time. We have defined five types of datasets these various categories generally fit into along with how they are used. 

1) Text Analytics/Social/Sentiment Data

This category has the most alternative data providers and is the most competitive. Alt data providers are structuring the firehose of news, social posts and reviews and quantifying the text data by identifying and tagging the data. Typically, hedge funds and asset managers are provided the data as normalized scores and then use machine learning and backtests to determine how changes in the data affect asset prices and volatility.

Investment managers use sentiment data to make a quick assessment of news. Machine learning techniques instantly assign metadata to each story and then gives a sentiment score of positive, neutral or negative. If a story scores negative, a trader can quickly enter a short-sale trade and profit from the drop in share price. The days of watching news feeds and reading the morning paper for updates are long gone. Traders can’t afford to wait for stories to break on traditional news sources. Information found on websites and social networks can provide unique and valuable insight into real-time news events breaking around the world. 

2) Consumer Transactions

From our sources, this data has the best return-on-investment, and is the most requested form of alternative data from buy-side firms. Credit card transactions are real-time analysis of what’s happening. This gives portfolio managers a snapshot of the health of company sales before earnings announcements. Anticipating is the name of the game; being one step ahead of analyst upgrades/downgrades or major media news stories. 

Some consumer transaction data is mined from mobile apps that help consumers manage their finances – such apps track spending patterns and provide data for investment process. These apps typically gain access to bank/investment/retirement accounts, loan/ insurance details, bills/rewards data and even payment transactions.

3) Satellite & Weather

Through government satellites and privately launched nanosatellites, a lot of big data can be analyzed. There are companies in the category that launch the satellites and other that analyze the data. We are most interested in the analysis of data and predictive alpha. Think of stringing together the data of hourly overhead images of a specified location. This data allows asset managers to monitor 1,200 Walmart parking lots, or monitor the slowdown in shipping of oil tankers through the Strait of Malacca (friction point between Indian Ocean and South China Sea where about 40% of world cargo moves), and many other use cases. All this can be monitored while sitting in an office in New York City.

4) Web Data/Online Search

Equity fundamental portfolio managers tend to favor web data to supplement their traditional investment thesis. Sources say that return on investment for web data is high and demand is robust. Data is typically provided in weekly datasets and tends to have the most price predictability on stocks in a specific sector. The best alpha is being found in airline, auto, internet, travel, and consumer discretionary sectors. Investment managers collect, measure, and analyze web data for purposes of understanding and predicting key drivers of companies, brands and products.

5) Geolocation

Geolocation data is generated by individual mobile phones location sensors, and is a good indication of foot traffic to a certain store or location. Mobile apps allow users to check-in to tell their friends what they are up to. This data is intriguing to portfolio managers to monitor sales at companies such as Starbucks (SBUX), Chipotle (CMG) and Apple (AAPL); to monitoring product shifts and demographic preferences, think millennial women buying shoes at Nordstrom’s (JWM) versus DSW. Some data providers have millions of users sharing this data, so it’s a good crowd proxy. Portfolio managers have been struggling to find consistent predictability with this dataset, but have found a few outsized returns, so they keep researching and data mining. We are hearing more alpha is being found in predicting product trend and demographic shifts than company sales.  

About the Author

C2 Capital Management Research is a technology forward proprietary investment firm that backs and guides hedge funds and traders in exchange-traded stocks, futures and options.