Clever? Contrarian? For Laplace’s predictive market AI, it’s just a matter of clarity

For Jean-Marc Patenaude, designing machine learning and AI systems to understand stock markets is a lot about time: timing, time frames, trading frequency, market history, reaction time, processing time. Patenaude and the Laplace Insights technology team spent the last 6 years building PSIMON, the predictive market AI platform that powers everything we do at Laplace Insights.

The month-to-month reality of PSIMON’s investment calls reveal that many investment themes are relatively short-lived: one sector or another outperforming for a few weeks, even a few months before settling back. None of this bothers Patenaude; our predictive insights platform is designed to make the most of these shifts, many of which are common sense, even obvious to industry veterans; still, many more are clear to PSIMON long before they become evident to other market players. And on some occasions, PSIMON’s algorithmic foresight identifies a major market shift as it emerges, helping investors position early for maximum returns.

Looking back at the last 9 months, PSIMON consistently generated well-timed, insightful suggestions about what to buy and when, what to sell and when, resulting in solid performance within PSIMON’s model portfolio, Columbus. Its decisions – proven to be savvy in hindsight –weren’t necessarily recognized by the street at the time, i.e., they were ahead of the curve. Now, Columbus wasn’t 100 per cent right 100 per cent of the time. No person or system is. In fact, we should be wary of anyone claiming ‘selection perfection’ and returns that always exceed expected probabilities. The often cited mythical investing crystal ball is just that: a myth.

However, intelligent systems drawing on the power of machine learning, artificial intelligence, advanced statistics and predictive analytics are real and can deliver results – our Columbus portfolio effectively tells us it’s possible. Since last October, PSIMON slightly lagged in identifying one single opportunity (discussed below). But that just means the platform is still learning. We continually improve the underlying code to support its ability to discern and comprehend actionable patterns and increase its precision.

Let’s take a look back at what PSIMON recommended and what happened starting in October 2020 and working forward. For clarity and credibility, we’ll examine a few key themes that we believe capture the bigger story. Where possible, we’ve also referenced opinions shared by influential market media sources that indicate our predictive analytics detected opportunities before the broader marketplace fully recognized them.

Appreciating Small Caps Early On

Last October, the U.S. presidential race was in full swing (creating market jitters) and COVID-19 was taking a terrible toll around the globe physically, psychologically and on confidence generally. You might have thought the marketplace would stick to safe bets like the debt-free “tech titans.”

Yet PSIMON’s analysis (as presented in our November allocation report published at the end of October) foresaw real opportunity in small caps based on the ultra-low interest environment and their lower volatility: we declared them to be, “a more compelling investment than the tech heavy large caps at this time.” It was the right call at the right time. And it took a bit before others caught up.

To be fair, not everyone missed it. Sanghamitra Saha of Zacks writing for Nasdaq (Are Smaller-Cap ETFs Good Bet for September?) promoted the segment around the same time we did. However, for the most part, the small cap recommendations were not on pace with what was actually happening, and endorsements were made retrospectively. These include the Motley Fool (December 2020), U.S. News & World Report and CNBC (January 2021).

Reaping the Rewards of Small Caps at the Right Time

It’s not just about what you buy and when. It’s also about what you sell and when.

With investing, rewards come from making the right decisions at the right time, and PSIMON helped us to reap returns from the small cap segment, beating the S&P 500 handily. In February (while some sources were still touting them, e.g., Kiplinger, April 2021), we sold our entire position, following PSIMON’s recommendation.

To quote our report at that time: “This proved to be a very profitable trade as small caps…produced a return of around 45%… . Since then, small caps have been correcting, in sync with growth stocks.” Those are results that are hard to argue with.

Getting Solid Returns from Well-Timed Commodity Investments

The arc of our investment in commodities also affirms PSIMON’s predictive insights to pursue superior returns.

In our December 2020 report, we announced the Columbus portfolio had taken a small position in commodities (five per cent). This increased dramatically to 26 per cent as inflation lifted the segment and PSIMON predicted it would “also benefit from a broad-based, demand-driven recovery.” We then trimmed the sails to 10 per cent before increasing to 28 per cent in March and selling out our position for a healthy gain at end of month. We bought back in during April and increased our percentage holdings slightly in May because PSIMON perceived commodities as an “inflation-friendly” asset class and they’re “poised to regain upside momentum driven by the economic recovery.”

The commodities story continues, as does our confidence in PSIMON’s recommendations.

Getting Great Value out of an Early Switch to Value Alternatives

Even while it was acknowledging the emerging opportunity of small cap investments, Columbus also began to highlight the merits of making a switch from growth to value stocks.

Starting with our December report, we began to make a recommendation about the high prospects of value stocks in contrast to technology-focused growth stocks that were still the market darlings. That conviction continued through our January report, was sustained in February and was validated in March: “We can see a clear rotation away from growth stocks and into value stocks happening in the market.” The April report continued to point to the attractiveness, momentum and lower volatility strengths of value versus growth.

Once again, there’s an echo in marketplace coverage for what we discovered before others also did as is seen in published pieces from Nasdaq, global money manager Schroders, Markets Insider and Forbes. We don’t dispute that intelligent minds will observe and describe investing opportunities with potential. But as the old adage states, “the early bird gets the worm”, and with an AI-driven platform to objectively process and interpret the shifting data, calculating complex probabilities, it is clear PSIMON can attain clarity faster, better and more consistently.

Notably, as others succumb to confirmation bias and grow more confident in this value theme, PSIMON has already recognized that it has—at least at the moment— lost momentum; as of our June report, Columbus is now overweight growth stocks, once again.

Getting Healthy Gains from Real Estate; Wisely Avoiding Gold

Real estate has performed exceptionally well, especially when compared with the S&P 500. Admittedly, PSIMON was slightly late to the party: while real estate began its ascent in March, our Columbus portfolio only went overweight real estate at the end of April; the outcome has nonetheless been profitable.

It is worth noting that over the same period, PSIMON avoided the shiny and traditional appeal of gold that others have fallen for. Many respected commentators pointed to gold’s traditional inflation hedge “superpowers,” but returns show that gold has been underwhelming in the face of recent inflationary worries. Through the period, the Columbus portfolio has been overweight real estate and underweight gold through the period, a stance that has resulted in measurable outperformance. So while PSIMON was uncharacteristically slow (by one monthly report) to embrace real estate, the longer term outcome has been solid.

In an article pointing out PSIMON’s prescient calls, it is worth noting, as we did above, that the platform appeared to be a month late to recommend real estate. Our AI is probabilistic, not a crystal ball, and caution vs confidence is obviously an ongoing calculation. Sometimes things go up that could easily have not; PSIMON does not recommend allocation changes flippantly, even if that sometimes means not being early to a still-forming theme.

An Ongoing Quest to Build on Evident Strengths

As a research-focused organization with deep expertise in financial machine learning, we strongly believe in learning from the sum of our experiences and embedding them in our organization and our tools.

We will continue to optimize PSIMON (as we do continuously) to further enhance its ability to discern meaningful investing patterns we can capitalize on. Marketplace noise will never distract us from making real-time, emotion-free recommendations grounded in analysing thousands of extreme market scenarios – not via back-tested models.

An investor can get lucky by following a certain formula once, or occasionally, but not repeatedly and consistently in all market scenarios and situations when both buying and selling regularly.

PSIMON’s predictive insights, especially over the last 9 months, has demonstrated that our approach to AI excels at finding opportunities – perhaps with a decidedly contrarian orientation, as seems the case with small caps. It’s certainly about more than simple data analysis of select aspects of the market: Columbus is interpreting the market movers more subtly and with greater sophistication, monitoring a complex web of variables. Its unique AI is examining relative attractiveness and relative returns, the possible upside and downside risks, to propose the best possible portfolio mix for achieving the best potential outcome.

If you’re a registered or a certified investment advisor and want to learn more about how Columbus can help you make uniquely well-informed selections for your clients, contact us to learn more, including a trial of our monthly investment report.

Backtesting Strategies & Overfitting

Strategy backtesting is often used to test an investment strategy from historical stock or ETF information (usually price).  The belief is that if a strategy performed well in the past, then it should continue to perform well in the future.

Read More »