Testing Multi-Factor Models in QuantDesk

  • By Paul Wilcox
  • April, 23
Blog Testing Multi-Factor Models in QuantDesk

Jonathan Moreland, Director of Research, InsiderInsights.com

Last December’s extreme market action reminded investors yet again that—despite the time, money, and technology invested to develop fancy schmancy valuation metrics and models—the golden rule in equity analysis is this: A stock is only worth what someone else will pay for it.

I never saw this lesson presented in any of the MBA and CFA texts I studied, but as far as I’m concerned, the first page—even chapter—of any investment text should be devoted to that irrefutable reality. An acknowledgement that, if you’re in the investment analysis business long enough, at some point the system and data inputs you use to form research conclusions will just not work for some period of time. 

And that’s…ok. Because—fortunately—the market spends most of its time in a rather more rational state than it was in last December. Common sense research considerations like company earnings, cash flows, balance sheet strength, stock technicals…and insider sentiment, are profitable to analyze.


Finding the alpha-generation potential

I teamed up with Lucena years ago, and QuantDesk proved an invaluable tool to test the alpha-generation potential of our (then) new expert system built to analyze our value-added insider data feed, and generate Insider Company Ratings for all U.S.-listed stocks in real time.

The positive results were published in a 2013 White Paper, and we subsequently undertook numerous other QuantDesk Event Studies to analyze our Ratings more granularly. With insider data still surprisingly underutilized (or even dismissed) by some professional investors, it was satisfying to successfully quantify all the factual and behavioral investment intelligence from this noisy and less-than-fully-structured data stream—into a single number that can be profitably integrated into research models.

But no investment factor is perfect. 2018 was not our best year, and I can definitely confirm that a “Significantly Bullish” Insider Company Rating on a stock last November was not enough to keep it from succumbing to the bearish secular price action leading into Christmas Eve. Insider-bought stocks tended to rebound better than most from that nadir, however, and the “insider factor” has certainly returned to profitable form this year based on our results.


Cue up QuantDesk

More importantly, new event studies recently run on QuantDesk independently confirm that—though we underperformed during several months last year—those periods were minorities among the many more profitable months and years our Insider Company Ratings have been able to generate. It’s comforting to confirm that a few periods of underperformance aren’t the same as our Ratings becoming commoditized to the point that their predictive value has degraded.

I’m excited to continue using QuantDesk to find new ways to use our InsiderInsights Company Ratings in the real world. Even more interesting is to test our Insider Ratings in combination with other data series ingested into QuantDesk, to find out what companion factors can make our Ratings even more consistently profitable.

The earnings and event data from Wall Street Horizon looks particularly intriguing to combine with our Insider Ratings. And with Lucena constantly adding more data partners, the range of multi-factor models that can be tested with our Insider Ratings is only getting larger.

If any other QuantDesk users want to discuss using our Insider Ratings as a primary or tertiary input into multi-factor modules you’re testing, give me a call or email.

Stay curious!

Jonathan Moreland

Director of Research, InsiderInsights.com



For more on utilizing and adopting Alt Data watch: “Constructing Unique Data Feeds for KPI and Price Forecasting”.

Have a media inquiry or a topic you’d like to contribute to our blog?