The Dirtiest Word In Finance: Market Timing

In 2015, Cliff Asness made the case that to earn attractive returns with proper risk-based diversification and low correlation to traditional markets, investors need to embrace ‘the three dirty words in finance,’ which he defined as leverage, short-selling, and derivatives. (Asness, 2015)

In deference to George Carlin’s seven dirty words, I believe that we should expand Cliff’s list to include the dirtiest word of all: market timing. (Note: Phil Huber @ Huber Financial is correct in pointing out that market timing is technically 2 words, but then the George Carlin reference wouldn’t work. So we’re sticking with dirtiest word and admitting defeat.)

To be sure, there are other dirty words in finance, like concentration, illiquidity and high cost, but I think that market timing may be the dirtiest of them all. Cliff used his list of dirty words to promote AQR’s Style Premia products (and some other funds, albeit less directly) and I believe that market timing is the key driver behinds AQR’s largest suite of mutual fund offerings: managed futures funds.

Managed futures, or trend following, involves buying securities whose prices are generally rising and selling assets whose prices are generally falling.  I believe that managed futures managers time the market (or markets: stocks, bonds, commodities, and currencies) using price trend as their timing signal (an example of this research is here).

Nearly everyone in the investment community, from academics to practitioners believe that successful market timing is impossible.  If managed futures strategies are essentially market timing, however, then successful market timing is not only possible but a potentially attractive addition to a well-diversified portfolio.

What is Market Timing?

While everyone may have an intuitive understanding of market timing, a clear cut definition is hard to find.  The landmark paper by Brinson et al. that turned market timing into a dirty word says that it is “the strategic under or overweighting of an asset class relative to its normal weight, for the purpose of return enhancement and/or risk reduction.” (Brinson, 1995)

Other definitions include the idea that market timing must include a forecast derived from factors like economic fundamentals, valuation, etc.  These definitions exclude managed futures as a type of market timing because trend following doesn’t require any forecasts or predictions since the strategy is backward-looking and reactive.

In my view, the definitions that include forecasts are too restrictive because the key factor in market timing isn’t the rationale, but the simple act of intentionally shifting a portfolio’s beta.  Some investors may use forecasts to add or subtract beta, but the forecasts aren’t a necessary condition.  Managed futures managers add and subtract beta based on price trends.

The following chart plots how trend followers may add and subtract beta based on market trends.  The left axis shows the rolling beta of a trend-following factor to the MSCI World Index.  For this chart, I used the equity component of a time-series momentum factor (TSMOM) created by academics (also affiliated with AQR) Moskowitz, Ooi and Pedersen related to their 2012 paper, Time Series Momentum.  (data are available here). The right axis depicts the corresponding MSCI World index return.

The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index.

Although the relationship isn’t perfect, it does appear that there is a relationship between the beta of a basic equity trend-following strategy and market returns.  Based on the TSMOM factor rules, a trend-follower would have added to their beta exposure as markets were rising and on the flip side, they would have decreased their exposure when markets were falling.  This relationship is, in effect, timing the market based on market returns.

Timing the Market, Poorly

Market timing gets a bad rap for good reason: it’s usually unsuccessful.

In 1966, Jack Treynor and Kay Mazuy were among the first academics to formally address whether mutual funds were successful at their market timing efforts.  They analyzed 57 open-ended mutual funds between 1953 and 1962, and only one fund successfully timed the market.  The researchers concluded that mutual fund managers could not anticipate major market movements and profit from them accordingly. (Treynor, 1966)

Since then, variations of this study have been conducted many times that include investment clubs, pensions, newsletters, professional market timers and asset allocation funds.  The results vary some by study, but the basic conclusion is always the same: on average, market timing underperformed buy-and-hold investing.

For several years, Morningstar has researched investor performance in a study called “Mind the Gap.”  They compare the time weighted returns of mutual funds to the dollar-weighted returns achieved by the fund’s investors.  Morningstar concludes that, “Investor returns fall short of a fund’s stated time-weighted returns because, in aggregate, people tend to buy after a fund has gained value and sell after it has lost value.”(Morningstar, 2016)

Maybe performance chasing by retail investors isn’t the same as market timing by professional investors, but a recent paper by AQR suggests that ‘Superstar Investors’ don’t get much out of market timing either.  The AQR researchers found that the returns for Warren Buffett, Bill Gross, and Peter Lynch are mostly explained by exposure to classic factors (market, value, credit, etc.). Only George Soros realized some true alpha through successful market timing.

It’s easy to see why market timing is a dirty word given the piles of academic studies, the poor investor returns and that even the superstars aren’t benefitting from it.

Timing the Market with Managed Futures

While the evidence was piling up against market timing as a whole, one group of investors was timing the market by following price trends: Commodity Trading Advisors (CTAs) and other trend followers in the managed futures industry.

The industry is relatively new dating back to the 1970s, and high-quality performance data is difficult to find.  BarclayHedge tracks the performance of CTAs in an index that started with 15 constituents in 1980, versus 532 today.  According to their website, the Barclay CTA Index gained 9.6 percent since inception, compared to 11.5 percent for the S&P 500.

Another well-known CTA performance tracker is the HFR Macro: Systematic Diversified Index that tracks trend follower’s dates back to 1990.  The HFRI index shows that since inception, systematic trend followers have earned 9.4 percent annually, an even match with the S&P 500.

This data likely has some backward looking biases, which we can evaluate by comparing the two data sets.   Since 1990, for example, the BarclayHedge returns only have a 0.4 correlation with the HFRI index.  If the data were solid, we would expect a much higher correlation between them, like over the past five years, when the indexes have converged.  Now the correlation is nearly perfect, but it was low and even negative in previous years.

Given the data problems and relatively short period, academic and commercial researchers have built factors and back-tests to try and understand whether the strategy is persistent and pervasive.  The Time Series Momentum paper referenced above by the chart found that 58 liquid futures contracts across four asset classes using a relatively simple strategy from 1985 through 2009 earned 16.1 percent, compared to 8.2 percent for the S&P 500 over the same period. (Moskowitz, Ooi and Pedersen, 2012)

Two years later, the same academics conducted a 135-year survey and found similar results.  Over the entire study period, the strategy earned 14.9 percent gross of fees and 11.2 percent net of a traditional “two and twenty” CTA cost structure.  The worst decade of returns, between 1900-1909, still earned 8.3 percent (gross).

An interesting (and difficult) question is why market timing may work for trend-followers but not for buy and hold investors.  I don’t have an answer, but offer two theories.  First, trend-following is typically systematic and the managers follow rules rather than their own judgment.  Market timing based on valuation, forecasts, etc. may be systematic, but I suspect that in the aggregate, it’s mostly discretion by managers and their clients.

Second, managed futures investors are typically very diversified, investing across stocks, bonds, commodities and currencies and frequently size their positions based on the risk contribution of each holding.  The aforementioned studies that found poor market timing results were mostly evaluating long-only equity managers that allocate by capital, as opposed to risk.  The long-only construct doesn’t allow managers to profit from falling markets, they can only lose less.  Managed futures managers, on the other hand, can switch from long to short across multiple asset classes.

Market Timing, Meet Buy-and-Hold

While the academic results are obviously very attractive, the live data is mixed with returns slightly lower than or indistinguishable from the S&P 500.  Keep in mind, though, that as Brinson defined market timing, it can be employed to either enhance returns or reduce risk.

If evaluated on a standalone basis, the live data for managed futures might not be attractive enough to warrant investment, but any strategy should also be considered in the context of a total portfolio.  Because managed futures strategies are lowly correlated with stocks, a portfolio that includes trend-following has historically enjoyed lower overall volatility without sacrificing much return.

Although it’s not possible to invest in a composite like BarclayHedge, let’s pretend that we could and take 20 percent of our portfolio from the S&P 500 and invested it in the BarclayHedge Index.  From 1980 to 2016, the S&P 500 earned 11.5 percent with annual volatility of 16.6 percent.  The return on the 80/20 mix would have been 11.5 percent, but because the correlation between the two asset classes was 0.55, the realized volatility would have dropped to 13.8 percent, thereby increasing the Sharpe Ratio from 0.36 to 0.51.

While the numbers differ, we get the same type of result repeating this process with the HFRI data.  From 1990 to 2016, the S&P 500 earned 9.4 percent with realized volatility of 17.6 percent and a Sharpe ratio of 0.28.  An 80/20 mix that includes the HFRI data would have returned 9.7 percent with realized volatility of 14.6 percent for Sharpe ratio of 0.36.

The academic factor TSMOM data shows remarkable results, partly because it starts in 1985 and ends in 2009, just before a run of poor returns from managed futures funds.  In this case, the raw returns for the managed futures strategy are substantially higher than the stock market at 17.6 percent, versus 10.5 percent for the S&P 500.

The academic factor TSMOM is attractive not only due to the returns, but because the realized volatility of the simulation was 12.0 percent and the correlation to the S&P 500 was -0.02.  As a result, an 80/20 S&P 500/TSMOM mix outperformed the S&P 500 by 1.9 percent.  The volatility of the 80/20 mix was 15.1 percent compared to 18.7 percent for the S&P 500, resulting in respective Sharpe ratios of 0.53 and 0.33 respectively.

Yes, You Can Time the Market

Yes, you can time the market, but you may or may not be successful.  If you time the market based on economic forecasts, valuations, elections, or your gut, it seems unlikely that you’ll either beat the market or lower the overall volatility of your portfolio.  If you time the market using some kind of trend-following approach like managed futures, you may or may not earn better returns; I think that the jury is still out.

There are some great track records of superstars from within the industry, but there is a silent graveyard of managers that closed their doors without anyone noticing (see this piece talking about RenTech’s exit out of managed futures).  The quality of the data has improved over the last five years, but that progress has been clouded by what has turned out to be a very difficult period in terms of performance for managed futures managers.

However, if you define success as diversifying your portfolio and lowering overall volatility, then the case seems relatively solid that managed futures, or market timing, works nicely with a long-only buy-and-hold strategy.  Don’t be tempted to time the market with your buy and hold portfolio, but if you want to time the markets with managed futures, adopt a buy and hold approach.

If you’d like to learn more about managed futures, here is a link to some research pieces by Alpha Architect.