How Alternative Data Was Used to Beat the S&P 500

Can alternative data help you beat the S&P 500? Yes, and by a wide margin. 

Alternative data refers to data used by investors to evaluate a company or investment that is not within their traditional data sources (financial statements, SEC filings, management presentations, press releases, etc.).*

Typical alternative data sets include employee & job data, footwork data, web traffic, satellite information, social media sentiment, news analysis, review sites, crowd scoring and more. 

While the value of alternative data is easy to imagine, putting alternative data into use is much more difficult. It is often unstructured, requires huge efforts to process and clean and in many cases difficult to model into a consistent investment strategy. Effective use requires grouping various skill sets, from data processing through quantitative modeling all the way to natural language processing and artificial intelligence. 

This is why alternative data, for the most part, has been used by a select group of hedge funds and tier 1 investment firms, and has yet to penetrate the mass market of asset managers, family offices and individual investors, despite its tremendous potential. 

For the past 6 years, we have built a platform that collects alternative data from multiple sources and translates them into investment tools. Our workflow incorporates collection from versatile sources, transformation to clean time series, quantitative model of each time series into investment signals and then portfolio selection using these signals. With this model, we have been able to score all stocks in universes such as the S&P 500, and present investors with an alternative data approach to stock picking and basket construction. 

In order to achieve this, we have built and trained a unique AI model, specifically for this purpose, using a variety of machine learning algorithms and trained NLP and AI code.

For the past 6 months, we have run long-only and market neutral strategies with a top tier asset management firm. The strategy included picking the top 15 best and worst scored stocks from the S&P 500, using score based weights, and rebalancing the baskets each month. . The results, as you can see below, have been phenomenal: 36% overperformance for a market-neutral strategy, and 19% over performance for a long-only strategy. All this vs. the S&P 500 as a benchmark and during a 6 month period only

And yes, this happened during the COVID-19 crisis, and with a Sharpe ratio of over 2.6.

This is just one example of how alternative data can be put into investment use today without going through the hassle of collecting, cleaning, modeling and scoring. 

If you’re interested  to see how Zirra can harness alternative data for your own strategy, eel free to contact me personally at [email protected]

Zirra utilizes thousands of data sources to collect data across hundreds of metrics on any company. These are then organized into over 100 distinct time-series signals providing deep insights for scoring and benchmarking, quantitative modeling, compliance and risk management.

Based on curated alternative signals, Zirra’s proprietary scoring model offers a unique system for relative comparison between companies within a given basket. Each company receives a score per category, based on carefully weighted signals, where rankings are then applied to the set according to their performance.

* From AlternativeData.org