Can Models Create the Low Correlations That the Markets Have Lost?

It’s well documented that many financial advisors are in a quandary these days. For the past 25 years, give or take, many advisors had been constructing client portfolios based on the tenets of Modern Portfolio Theory, which enabled them to create “optimized” portfolios by allocating client assets among various asset classes, the projected returns of which had a broad range of low correlation to each other. The key factor that made those portfolio optimization programs—and the portfolios they generated—work was the notion that the correlations between those asset classes were static over time, and therefore, predictable. 

But do those correlations remain constant under all conditions? And what would happen if the correlations changed under certain conditions? Of course, we now know the answer to those questions, thanks to the Sub-Prime Meltdown of 2008-’09: No, and portfolio Armageddon. 

While the markets have largely recovered from that debacle, many, if not most, financial advisors have not. When virtually all asset classes (except Treasuries) took a nosedive, so did much of the faith in “well-allocated” portfolios, creating a philosophical void in advisors’ investment strategies.

While there’s a plethora of coverage about the problem, there’s a dearth of stories on viable solutions. (To make things worse, those high correlations have persisted since, and seem to be increasing over time. The current one-year correlations to the S&P 500 paint a troubling picture: S&P 500 Value 0.99, S&P Growth 0.99; Small Cap Equities 0.97; China 0.88; REITs 0.93; Emerging Markets 0.87; High Yield Bonds 0.86.) I’ve written about commodity ETFs and funds as one possible solution (January 2011 and November 2011 Investment Advisor), and I’ve recently come across another candidate with a more technical twist. 

In 1999, money manager Larry Connors and other experienced traders, including Kevin Haggerty, former head of trading for Fidelity Capital Markets, founded The Connors Group in Jersey City, N.J. Troubled by eroding market metrics leading up to the Dot.com crash of 2001, Connors and his team sought to create a quantified and systematic approach to portfolio investing and risk management. 

Their resulting portfolio models are largely based on the work of Nobel Prize winner Daniel Kahneman (author of “Thinking, Fast and Slow”) and Amos Tversky, whose work in the 1970s explored the flaws in the Efficient Market Theory, and led to the creation of Behavioral Finance. Connors reasoned that if conventional MPT asset allocation is not the solution, effective portfolio diversification could be achieved by diversifying investment strategies (through models) instead of, or in addition to, diversifying asset class, if those models were based on short-term strategies designed to take advantage of daily behavior in market pricing. 

To make this work, Connors and his team set out to create non-correlating strategic models using equities, ETFs and high-yield bonds. They created three types of models:

  • Those that identify equities that are undervalued in the short term due to investor over-reactions to general or specific news;
  • models that identify early stages of highly sought-after equities;
  • and models that identify over-valued ETFs that are in the early stages of the selling phase.

The result is over 28,000 active investing strategies (quantified models) that can be blended into well-diversified portfolios. These strategies break down into 7 categories:

1)     Weekly Rotation Strategies

2)     Semi-Monthly Rotation Strategies

3)     Selective Value Strategies

4)     Equity Trend Strategies

5)     ETF Trend Strategies

6)     Short ETF Strategies

7)     High Yield Bond Rotation Strategy

 

In November 2011, The Connors Group launched The Machine Advisor, which allows financial professionals to build portfolios out of

their diversified strategies (and in conjunction with traditional asset classes) with lower and more stable correlations to the broad market. The idea is to create more predictable, lower risk client portfolios. 

“With the Machine Advisor, you’re going to miss the highs and the lows of the market, generating more consistent—and higher—returns over time. If the market is up 39%, an advisor’s portfolio might be up 25%; but if the market is down 25%, they’ll be down, say, 6%,” Connors’ new CEO, Jim Lonergan, told me. “What’s more, when advisors go to sell their firms, which is a big issue today, they’ll have a system and a structured strategy that a new advisor can just step into.” 

Lower risk and higher returns, of course, was the original goal of MPT-allocated portfolios. Does Connors’ version work? 

You be the judge: From January 1, 2001 to December 31, 2011, a conservative 60/40 Balanced Blue Chip portfolio using a mix of Connors’ strategies returned a compounded annual 10.27% (net of fees and commissions) with a standard deviation of 7.82%; while the S&P 500 returned an annual 1.40%, with 16.43% standard deviation. That includes a 0.91% return in 2002 for the 60/40 BC vs. -21.59% for the S&P, and a 1.14% for the 60/40 in 2008 vs. -36.80% for the S&P. 

I realize that’s only two market cycles over 10 years, but if you like the idea of smoothing out portfolio returns while eliminating big losses, the preliminary indications are that Connors, Lonergan & Co. just might be on to something. 

Page 1 of 2
Single page view Reprints Discuss this story
This is where the comments go.