Outperforming The S&P 500 With Consumer Survey Data

[view original post]

About half of US allocators are planning on increasing their investments in hedge funds for 2022 according to a recent With Intelligence survey of 200 allocators. One reason may be investor concerns about the 40-year inflation rate the US is experiencing. Although last year was a good year for funds historically the Credit Suisse Equity Market Neutral Index that evaluates the Alpha generation of hedge funds stands at 0% over the past 17 years since it began in 2004.

One firm working to overcome the challenge of generating alpha is a company named CloudQuant. They enable financial services and corporate clients to realize the power of alternative data by providing rapid access to thousands of high-value datasets, research, and technology. By using a vendor partner’s high-quality, proprietary consumer survey data, they are creating accurate predictive analytics and signals for portfolio allocation, over or underperformance of industries within a sector, and forecasts or nowcasts of government economic factors 3-4 weeks in advance of release.

I spoke with Morgan Slade to discuss how these novel consumer survey datasets are powering new analytic signals to generate alpha.

Gary Drenik: Why should investors be taking a look at consumer survey data?

Morgan Slade: Consumer privacy advocates have notched a long series of wins in the war to establish stricter global privacy regulations. The European Union’s 2018 GDPR was a shot across the bow of data-hungry tech companies and consumer businesses that rely on digital footprints to target ads and drive revenue growth. A recent EU court ruling restricted big tech companies from transferring personal European consumer data to inexpensive American data centers for analysis and ad serving. The European court feared that unwitting European consumers could have their private details caught up in the US government surveillance dragnet set up and expanded since 9/11. If you believe it is going to be easy to get your hands on consumer digital breadcrumbs in the future, think again. Look at the recent SEC AppAnnie fine for violating the permitted use and format of the user privacy data. These shifting sands in the digital tracking ecosystem are why we foresee consumer survey data as the only predictable method of collecting and curating scientifically consistent consumer trend insights. Consumer survey data harnesses private consumer data from willing contributors and enables direct inquiry about questions that matter. If you want to know what Millennials in Atlanta, GA in the 50k to 100k income bracket are going to spend on travel in the next six months, you can ask them directly using a survey or you can attempt to infer that from transaction data or digital engagement after the fact.

Drenik: What did your research team find that was unique?

Slade: When we begin analyzing a survey dataset like Prosper, we first look at the sampling methodology, the geographic coverage, and the demographic balance of the data. Prosper goes to great pains to mimic the demographic make-up of the US and Chinese markets. In addition, they have curated a history of data going back over 20 years for the US. In addition to providing well-designed segmentation of the participants based on key demographic features, we liked the fact that many of the questions were phrased to elicit forward-looking perceptions and expectations.


We utilized our machine learning research stack to delineate the key predictive features hidden within the data from 2015 to mid-2021. Within that time period, we identified particular questions and responses that were predictive of over or underperformance of certain sub-industries within a sector. With that predictive model, our research team was able to select a portfolio of subindustries that were predicted to outperform/underperform the broader market.

We have also presented results in a recent JP Morgan Institutional Investor conference showing the use of the same dataset to forecast or nowcast government economic numbers 3-4 weeks in advance of their actual release.

Drenik: How would you recommend incorporating survey data into your investment strategy?

Slade: We view this type of survey data as being adaptive to market regimes and able to peek into the future to gain a slight advantage over transactional or engagement-based datasets. We recommend utilizing the data to identify which segments of the stock market or economy are likely to be a target-rich environment with strong revenue and earnings momentum. Ultimately this data is getting at “how confident are you?” and “where do you plan to spend your money in the near future?” for US and Chinese consumers. Our research shows that regardless of whether you are investing in traditional growth or value market segments, the data and signals developed from the dataset were deterministic in their ability to forecast relative over or underperformance. This dataset definitely gives you the ability to skate to where the puck is going rather than looking at where it was and trying to predict the next move.

Drenik: How have clients used this data to navigate the markets?

Slade: Institutional investors seem to have rediscovered this Prosper data as the allure of digital trend data has dimmed. I attribute that to two factors: (1) the regulatory environment for data privacy is trending strongly in the direction of restrictions and (2) backward-looking transaction data as it turns out simply isn’t as useful for prediction of the future. This can be seen in the investment returns we measured in the investment processes we developed utilizing the Prosper data.

Drenik: What is unique about this signal?

Slade: The breadth of questions covered by this survey can set any research analyst onto a dozen different investment theses. It does an excellent job of capturing the current state and future outlook of the economy, the job market, investment returns, consumer products, government, travel plans, savings plans, spending plans, eating out, major purchases, consumer discretionary spending, family and even your love life. Having a long history of consistent measurements of these features is nearly impossible to find. With monthly US data going back over twenty years, and quarterly Chinese data going back over ten years, there is coverage of enough economic shocks to show this is an all-weather dataset. Its ability to detect and forecast trends throughout the pandemic powered impressive investment performance over the last few years.

Drenik: The Covid-19 Pandemic has caused markets to be unstable over the course of the last 2 years or so. How did the signal perform in the markets throughout these uncertain times?

Slade: As I alluded to earlier, Prosper’s data completely nailed it during Covid-19 and essentially gave our investment model an early preview of the future to make superior investment decisions. It has definitely stood the test of time and circumstance for us. One investment strategy developed by CloudQuant outperformed the market by 23% during the last 24 months during Covid-19. What’s more, is that the strategy did so by identifying some subindustries that should be sold short and some that should be bought, something that requires more skill.

More details & additional charts are available on our use case research.

Drenik: Morgan, thank you for demonstrating your innovative predictive analytics. With consumer spending contributing almost 70% of GDP, the novel analytics derived from accurate research data from consumers stated behaviors and future purchase intentions makes sense. Corporates and investors now have an exciting new way to see the future of markets and make better decisions.