Century-Long Backtest Shines on a Few ETF Strategies

In a new article from Institutional Investor“Market Timing Is Back In The Hunt For Investors”–AQR Capital Management reviews the case for market timing and finds an encouraging track record. Citing the historical record from 1900, AQR’s Cliff Asness and two colleagues outline what is effectively a century-plus backtest of the value and momentum factors. In line with analysis from other researchers, they find that the numbers favor a degree of dynamic portfolio management for adjusting asset allocation through time based on this dual-factor framework. There’s nothing particularly new here, at least for anyone who’s familiar with the research on the topic of tactical asset allocation (TAA) and the related subjects. Nonetheless, this is a worthwhile read if only because it provides a long-run perspective on how the application of value and momentum factors provide a powerful foundation for managing an asset mix through time.

The authors are careful to point out that no one should confuse the results with a clear-cut case that market timing is easy. The standard caveats apply, they emphasize. Although the rearview mirror appears to provide an obvious roadmap for dynamic asset allocation, hindsight bias is probably making the process appear easier than it’s likely to be in real time going forward. For instance, it’s one thing to look back over a 100-plus-year period and proclaim that the turning points for valuation extremes are conspicuous. But investors in earlier generations didn’t have the benefit of knowing what we know now and so the ability to exploit the value and momentum factors in years past may have been limited and perhaps even missing altogether as a practical matter.

That’s a clue for wondering if we’re still in the dark to a degree because the future’s always uncertain. This may be an especially problematic issue for the value factor if assets are being repriced at a higher level generally compared with decades past. In other words, assets that appear cheap or expensive based on previous standards may no longer apply.

But those are standard warnings that apply to all backtests. Past performance is no guarantee of future results. Blah, blah, blah and caveat emptor… as usual. No less applies to a buy-and-hold strategy, by the way, and so everyone’s in the same boat. That said, the article makes a powerful case for combining the value and momentum factors in an asset allocation framework. Applying any one of the three is isolation of the other two runs up against serious obstacles; deploying all three as part of a larger strategic whole, on the other hand, forms the basis of a formidable toolkit for managing risk and generating attractive returns.

As an example, Asness and company point to the rolling 3-year performance for a 50/50 asset mix of US stocks and bonds since 1900 using a value and momentum strategy. Summarizing the results (shown in the chart below), the authors observe that “there are plenty of 3-year periods in which the full combo approach subtracts value (and we know good things get abandoned way too often when they suffer for three to five years). But at this horizon it adds considerably more often than it doesn’t, and historically it has been reasonably well behaved—that is, no superfat tails in either direction.”