Guest Opinion: Why Robo-Analysts, Not Robo-Advisors, Will Transform Investing

David Trainer, New Constructs

Robo-advisors and robo-analysts are both important to enabling wealth management firms to cut costs without sacrificing quality of advice, but the importance of a robo-analyst to enhance the quality of investment advice shouldn’t be underestimated.

Today, many of the tasks performed by robo-advisors are low value-added services such as determining and communicating asset allocation strategies (e.g., 60% equities, 30% fixed income and 10% cash). In fact, these services are so low value-added that advisors cannot make money doing them unless they are bundled with higher value-added services.The value proposition of a robo-analyst is very different. That’s because robo-analyst technology empowers wealth managers to provide the kind of long-term, value-oriented advice that not only meets the fiduciary duty of care, but also provides peace of mind for both advisor and client. Specifically, by shining an analytical light in the dark corners of financial filings, robo-analyst technology can identify many critical data points overlooked by most research analysts today. No longer must investors rely on the headlines or management-manipulated earnings. With new technologies, investors can receive a much fuller, more comprehensive analysis of financial filings, company profits and valuation so as to make better informed decisions than ever before. As a result, robo-analyst tech raises the analytical bar universally, enabling investors to transcend the short-sighted and high turnover trading mentality that, in the long run, does more damage to investors than good.

Robo-Analyst Technology Brings Analytical Rigor Back to the Advice Business

It’s no surprise that the market has become more focused on headlines and management-manipulated earnings than economic earnings over the past 20 years. Management-manipulated (a.k.a. accounting) earnings are widely available for free. They maintain a large presence in the media and are the focus of the always attention-grabbing quarterly releases. On the other hand, economic earnings, the truly comprehensive view of corporate profits, are hard to find and rarely free, because they have historically required considerable human effort to calculate. To be derived with integrity, economic earnings require rigorous analysis of every annual and quarterly financial filing, cover to cover, that a company has published for several years. For those not familiar with these documents, they average over 200 pages, can be as long as 1,900 pages and are filled with complex accounting and legalese.

Who has the desire to do this much work on one stock and/or an entire portfolio of stocks, ETFs and mutual funds? How does one have time, with all the fast-moving trades and constant news? You get the idea. In fact, Wharton University’s Brian Bushee published a survey of institutional investors in the Journal of Applied Corporate Finance showing that only eight percent of investors performed this kind of rigorous work in 2004. That percentage is almost certainly much lower today.

Natural Language Processing Techniques

Given the disconnect between the importance of analyzing the filings and the number of people doing that diligence, it was only a matter of time before someone figured out how to get machines to do much of this work for us. There is still plenty of room to improve the natural language processing (NLP) technology utilized, and I’m not sure it will ever be perfect, but we have come a long way. The technology is not fancy artificial intelligence software. To the contrary, it can be built by pairing forensic accountants with programmers who can create software to track the parsing decisions of analysts as they analyze 10Ks and 10Qs. In our case, these decisions form the foundation of a library of human-validated parsing instructions that NLP techniques leverage to teach machines how to parse automatically. The automated parsing tools are further enhanced by our ontology for translating raw accounting data into value-added investment advice. In other words, we model the raw data and derive investment ratings to give further context and meaning when analyzing filings ─ a powerful intelligence feedback loop.

Humans continue to guide the parsing and model-building process, as they have for the last 130,000 filings we’ve analyzed. But their role is altered dramatically, as they can now spend more time scouring footnotes and the MD&As for usual items and far less time on mind-numbing repetitive tasks like parsing financial statements. In case you’re wondering whether this extra high-value time pays off, we can illustrate that it does.

Hasbro (HAS) is a great example. The stock has more than doubled since our initial call in January 2015. During that time, HAS has grown economic earnings by 9% compounded annually over the past decade, while accounting earnings have grown at an annual rate of just 6%. As such, and unlike most of the S&P 500, Hasbro’s underlying drivers of valuation are better than its top-line metrics. This disconnect is driven in part by items such as a $33-million impairment charge the company took last year that artificially decreased its reported earnings.

The same analytical rigor applies to funds, too. For example, Rydex Energy Services Fund (RYESX) is down 28% vs. its benchmark, XLE, which is down just 14% since we highlighted it on January 30, 2017. The reasons for our warning emanated from two key issues our technology discovered:

  1. The fund’s annual turnover was an astounding 1241%, which translates into an annual cost of 4.73%. No matter the investment strategy, such high trading costs can erode any potential returns for investors.

  2. Financial filings analyzed and models built by our roboanalyst revealed that the fund’s managers had allocated more capital to stocks that were simultaneously less profitable and more expensive than their benchmark (XLE).  

Heightened awareness of fiduciary duties by both investors and advisors is ushering in a return to the more diligent habits that underpin value investing (e.g., focusing on economic earnings) and are more important than ever before. Moreover, now that interest rates are no longer falling, investing is getting harder. To fulfill fiduciary duties, it’s imperative for investors to roll up their sleeves and do the extra work necessary to understand true, economic earnings and not rely on unscrubbed accounting results. Technologies like ours support this trend, because they make it much more feasible for investors to apply the value investing principles that play a critical role in providing high-quality investment advice.

David Trainer is CEO of New Constructs (www.newconstructs.com), an independent research firm that provides forensic-accounting-driven investment ratings on stocks, ETFs and mutual funds. He is also a member of FASB's Investors Advisory Committee and author of the Chapter “Modern Tools for Valuation” in The Valuation Handbook.