A comprehensive look at market timing

In this post I’ll summarize a recent and very comprehensive study on market timing. It is probably the most comprehensive and robust look at market timing yet. Market timing is one of the biggest potential problems and complaints with TAA portfolios as I highlighted recently. Everyone wants a definitive answer – does it work or doesn’t it? As you might expect, there is something in the results for everyone, the die hard buy and holders and the die hard market timers. For those interested in implementing TAA portfolios there are several surprising findings in the study that are very applicable to a TAA approach with the potential to significantly improve results. Let’s dive right in.

First, a few notes about the study. You can read the entire study for yourself here. I highly recommend a detailed reading for advanced practitioners of TAA portfolios. The study is very recent, draft published in March, last updated a few days ago. The study is extremely robust and comprehensive. It uses monthly data on the SP500 index for a 155 year period – from Jan 186o through Dec 2014. The tests are conducted out of sample. The study includes the effect of transaction costs (0.25% one-way transaction costs). It tests 6 market timing rules over the study period – from the simple absolute momentum rule, to the well know simple moving average, to more complex rules like the double cross-over method. It tests the market timing rules not only over the whole study period but also over more realistic investment horizons, 5 and 10 year periods, using familiar portfolio stats such as sharpe ratio and sortino ratio. All right, enough about the technical details. Let’s get to the results.

The first set of results are for the entire out of sample test period. The study parameters were determined in a 10 year in sample period, then applied to the 1870 to 2014 out of sample period. Below is the results table showing mean monthly returns (not compounded) and other descriptive statistics.

I only include the 3 market timing rules that showed statistically significant results (the bold figures in the table); absolute momentum, reverse exponential moving average, and the classic simple moving average. In short, all three of the market timing rules outperform the market on a risk adjusted basis (sharpe ratio) over the study period (145 years, 1870 to 2014). Of the three market timing rules, the absolute momentum rule has the best risk adjusted performance. The simplest is often the best approach. Also, note that the absolute moment rule is the only one with positive skewness (in a very simples sense – more data points above the mean than below the mean), and significantly greater than the market. Before I move on to a better way to look at the results, let’s look at how the study determined the optimal look back period for the market timing rules.

The difference between the Rol (rolling) and Exp (expanding) columns in the results table above is the different method to determine the lookback period for each rule. The expanding rule looks at the entire study period to determine how many historical months to use for the market timing rule, e.g 10 month SMA. The rolling rule uses a rolling window of historical data. Ideally, the two would be the same. i.e. the optimal look back period would be stationary, it would be a single number. But it isn’t. The optimal look back period varies over time. The chart below shows the optimal look back period, in months, for the momentum rule and the simple moving average rule over time. The average for the momentum rule is 7 months and for the simple moving average it is 10 months but there is significant variation.

Now, let’s turn to a better way to look at the results. There are two important improvements to the results presentation. First, the presentation of annualized returns compared to the market and the returns over more realistic investment horizons (5 year and 10 years). After all, who has a 145 year investment horizon.

The next results presentation is much more applicable to real world investors. In these sets of results, the study looks at the annualized performance vs the market for the various market timing rules over 5 year investment periods and over 10 year investment periods. I also limited the presentation to the three rules that showed statistically significant results. Below is the results table.

The three market timing rules all outperform the market (positive mean values) and that outperformance increases the longer the investment period. For example, over a 5 year horizon, the absolute momentum rule outperforms the market on average by 1.95% per year using a rolling estimate of the optimal look back period. However, note that even over 5 and 10 year investment horizons there is still a significant chance the market timing rule will underperform the market. Over a 5 year horizon, only the absolute momentum rule has an out performance probability over 50%. Over a 10 year horizon, all three rules have a greater than 50% chance of outperforming that market, with all but one approaching a 60% chance. I also think it’s instructive to look at how the performance vs the market varies over time. Below I show the annualized performance over 5 year horizons relative to the market for the absolute momentum rule over the study period.

Notice that the greatest out performance versus the market for the momentum rule has come during the worst bear markets. It’s similar for the other rules. The worse the bear market the greater the out performance. That’s what you would expect and hope from a market timing rule I think. Of course, for that benefit you would have had to tolerate long periods of under performance. Whether it’s worth implementing market timing is entirely up to the investor and their goals. I would argue that at a minimum, market timing will probably help investors stick with their investment approach and thus lead to better returns and that for retirees where the biggest risk is big negative returns, market timing leads to higher SWRs versus buy and hold.

I’m going to leave the concluding remarks on this post to the authors of the study.  There is a ton that could be said about these results but I think they sum it up nicely and conservatively.

“…we did find support for the claim that one can beat the market by timing it. Yet the chances for beating the market depend on the length of the investment horizon. Whereas over very long-term horizons the market timing strategy is almost sure to outperform the market on a risk-adjusted basis, over more realistic medium- term horizons the market timing strategy is equally likely to outperform as to underperform. Yet we found that the average outperformance is greater than the average underperformance.”