Here’s Where Market Timing Works
Factor timing has its champions and skeptics — so academics set out to find an answer to the controversial practice.
When it comes to factor investing, timing matters.
A new academic paper published late in March shows that market timing works with factors. Researchers found that timed portfolios outperformed those that were static when they tested portfolio timing across 300 risk premia or what they call a “zoo” of factors.
In fact, the median improvement of a timed factor portfolio versus a static one was about two percent. Perhaps more compelling, the researchers found that a timed multifactor portfolio returned 20 percent more than portfolios that were left alone.
Factor timing has its champions, such as Research Affiliates’ Rob Arnott, and its detractors, like AQR’s Cliff Asness. The debate, as previously reported by Institutional Investor, questioned whether factor investing is too expensive, whether timing works, and whether simple diversification could achieve similar results.
Factor investing is no longer thought of as expensive, thanks to declining stock prices. In a recent paper, Arnott studied the performance of 19 factors worldwide and found that 11 of them are trading in the cheapest quintile of their historical valuations.
But other questions remain, including whether market timing can improve investors’ results.
Academics Andreas Neuhierl at Washington University in St. Louis and Otto Randl, Cristoph Reschenhofer, and Josef Zechner from the Vienna University of Economics and Business sought to answer the timing question in their latest publication, “Timing the Factor Zoo.”
“In this paper we conduct a comprehensive analysis of factor timing, simultaneously considering a large set of risk factors and of prediction variables,” according to the research. “Our analysis reveals that factor timing is indeed possible.”
The researchers used factor data from CRSP, Compustat, IBES, and FRED, aggregating information available between 1926 and 2020. Within that time frame, there are different start dates depending on the availability of data. For instance, price-based characteristics have the longest history.
According to the researchers, models have “taken a long journey” from a single factor focus “towards a heavily criticized factor zoo.” The idea that there could be some 300-odd factors that can be used as the basis for investment decisions has come under scrutiny. Researchers have questioned both small sample sizes and whether the results of certain studies can be replicated.
The researchers used a list of over 300 factors, including tail risk beta, inventory growth, profit margin, and size.
They also used a list of 39 timing signals, categorized into six groups: momentum, volatility, valuation spread, characteristics spread, issuer-purchaser spread, and reversal. According to the paper, momentum and volatility signals can be used to improve the portfolios’ raw returns, alphas, and Sharpe ratios. Other signal groups led to more variable results.
The researchers show that timing can improve both the performance of individual factors as multifactor portfolios. Improvements are highest for the profitability and value factors. Historical returns of factors and volatility are the best predictors of future returns.
Researchers found that a timed multifactor portfolio focused on small stocks returned an annualized 26.6 percent; the untimed multifactor portfolio returned 24 percent per year.
“Alongside the impressive gain in performance, we find decreasing maximum drawdowns and a reasonable number of firms in the portfolio,” the researchers wrote of the timed portfolio. Meanwhile, the timed large-cap factor portfolio also outperformed its untimed peer by roughly 2.3 percentage points.
“Our evidence reveals that factor timing is greatly beneficial to investors relative to passive ‘harvesting’ of risk premia,” according to the paper.