Ghosts in the Momentum Machine

As an analyst, I'm a fundamentalist at heart, focusing primarily on fund managers whose success owes to bottom-up research and strict valuation work. In advance of a panel I'll moderate at this summer's Morningstar Investment Conference, though, I've been researching the scholarship surrounding a technical strategy: investing purely on the basis of upward price momentum.

It's a fascinating topic, particularly for those who favor fundamental money managers, a group that, on average, has generally lost to the relevant bogies.

Contrary to that track record, the data on price momentum seem to show remarkable long-haul success. As I noted last month, Tom Hancock--co-head of GMO's global quantitative equity team--has crunched the numbers and found that, between 1927 and 2009, a simple strategy of investing in stocks with the highest trailing-12-month returns surpassed the broader market by 3 annualized percentage points.

And Yet �
A healthy dose of skepticism is in order. While momentum may look terrific under laboratory conditions, it can be devilishly difficult for actual investors to exploit. The strategy Hancock tested requires monthly rebalancing of a super-sized portfolio, one comprising fully the best-performing quartile of U.S. large-cap stocks.

That's an onerous task and an expensive one, given the level of portfolio churn. It's no surprise, then, that the universe of pure-play momentum funds is exceedingly small. Nor is it a head-scratcher that many funds using the tactic as an element within a broader strategy have struggled long-term-- Brandywine (BRWIX),  American Century Ultra, (TWCUX) and Turner Midcap Growth Investor  prominent among them.

Transaction costs drag on returns, after all. And, as with any strategy, price momentum's effectiveness waxes and wanes. Though it notched a strong return to form in 2010, the previous decade was exceedingly tough on the tactic. Those scars still show.

Blinded by Science
Substantial challenges exist even in the lab. Famously, correlation isn't causation, but when back-testing for a particular signal, that often becomes a distinction without much of a difference: The range of answers any set of data provides is always circumscribed by the questions an analyst asks, making it all too easy to simply find what you're looking for.

That risk of confirmation bias is nicely illustrated in the speech Vanguard founder Jack Bogle delivered as the keynote speaker at Morningstar's 2002 conference. Criticizing "the vastly oversimplified but typical way we look at long-term results," Bogle zeroed in on the vaunted small-cap and value-stock premiums, showing that, although it may seem in the aggregate as if those two asset classes have persistently outperformed, that's not what the data show: Relatively narrow portions of a time series spanning more than 70 years account for the bulk of the excess returns.

It's Academic
Could a similar dynamic be at work in the research surrounding momentum, with neatly rolled-up, aggregate-level data points head-faking in the direction of wrong--or at least incomplete--conclusions?

Well, sure. Any attempt to isolate the significance of a single variable (or even a small handful of variables) in a vast sea of data runs ample signal-to-noise risk. In addition to time-series static, for example, how much of momentum's long-haul outperformance may owe to style? Has the tactic's showing been powered by pockets of success in, say, growth stocks or mid-cap names?

Our colleagues at Ibbotson ask (and answer) that question in a recent white paper analyzing the influence of price momentum on the returns of composite mutual fund portfolios. I'm still scrutinizing the particulars of the study (which also examines the impact of investment in low-liquidity stocks), but the conclusion is striking: Regardless of where in the style box they reside, portfolios comprising funds that provide the greatest level of exposure to high-momentum stocks significantly outperform those with the lowest levels.

Back to Basics
Widening the analytical lens to include style represents a leap forward in terms of understanding the persistence of momentum, in part because it prompts related, though more fundamental, questions. What, for example, is the average manager tenure of the high-momentum funds that drive the bulk of the outperformance? Has the tactic generally fared better among higher-quality names or more-speculative fare?

Widening the lens even further to include these and other qualitative questions can further strengthen our understanding of momentum. They help to clarify and refine what the tactic has actually contributed and what portion of the gains ascribed to it may owe to other attributes--profitability metrics, say, or in the case of funds, managerial tenure.

Those factors almost certainly account for a greater than zero percentage of results that pass as effects of a momentum premium. How much greater is a question in need of greater study.