“There is no science in this world like physics. Nothing comes close to the precision with which physics enables you to understand the world around you. It’s the laws of physics that allow us to say exactly what time the sun is going to rise. What time the eclipse is going to begin. What time the eclipse is going to end.” – Neil Degrasse Tyson

Investing is not physics. Not even close. There is no precise formula or indicator that can solve the mystery of markets. There is never certainty in terms of outcomes, only probabilities and whether you have the emotional fortitude to stick with a plan/process over time.

Our most recent research paper delves into moving averages and leverage (click here to download). We show that using moving averages can help manage risk over time and enable you to systematically employ leverage to enhance returns.

In the paper, we illustrate the performance of strategies using various simple moving average time periods that are popular among market participants: 10-day, 20-day, 50-day, 100-day, and 200-day.

The rotation works as follows. If the S&P 500 closes above its moving average, go 100% into the S&P 500 and use some leverage. If it closes below the average, go 100% into Treasury bills. Simple.

Going back to 1928, using any of these moving averages to time your exposure to equities, even without leverage, would have beaten a buy-and-hold of the S&P 500 with significantly lower volatility (ignoring transaction costs/slippage, etc.).

Table 1: Unleveraged Buy and Hold versus Unleveraged Moving Average Timing

(October 1928 – October 2015)

When leverage is applied, absolute returns improve, but more importantly, risk-adjusted performance remained very strong (note: the table below shows just the 200-day moving average using 1.25x, 2x and 3x leverage but the same was true for other time periods).

Table 2: Unleveraged Buy and Hold versus Leveraged Rotation Strategies (Oct 1928 – Oct 2015)

Chasing Optimization

When we presented these and other findings to the Market Technicians Association at their Annual Symposium and on a recent Webcast, there were a number interesting questions. Many were focused on how one could improve upon the results.

“Did you test the strategy using exponential moving averages? Did the returns improve?”

“Did you test the strategy using moving average crossovers? Did the returns improve?”

“Did you test the strategy using the slope of the moving average? Did the returns improve?”

“Did you test the strategy using x-day moving average? Did the returns improve?”

The short answer: sometimes yes, sometimes no.

You see, there are literally an infinite amount of permutations in terms of time periods and indicators that you could have used to achieve a better past return. But simply optimizing for the best past return with the lowest volatility was not the purpose of the paper. Our goal was to disprove popular notions about leverage and moving averages and to illustrate an anomaly that challenges some of the central tenets of the CAPM/EMH.

What would be the point of optimizing for the perfect past return anyway unless there was some fundamental reason why a certain time period/indicator should continue its outperformance in the future? Lacking such a reason, you are merely chasing optimization by trying to find the precise time period/indicator combination that would have given you the best past result.

Far too often we take for granted how long-term performance using various indicators can look similar while in the short-run that performance can look wildly different. This is true whether you are using technical indicators like moving averages or momentum, fundamental indicators like price to earnings/book/sales, or macro indicators. There is a cycle to everything.

Let’s run through an example to make this point clearer.

If you were launching the simple moving average rotation strategy outlined above (unleveraged version) in March 2009, the 10-day moving average would be the easiest choice as it had the highest return.

Since March 2009, how has the 10-day strategy performed versus other time periods?

It has been the worst performing time period, with an annualized return of 3.9%. The best time period has been the 100-day moving average (9.5% annualized) followed by the 200-day moving average (8.8% annualized).

Does this mean that the 10-day moving average is “broken” and you should switch to the 100-day or 200-day moving average going forward?

Perhaps, but it is equally possible that we were simply in a time period that has been unfavorable to short-term trend following and that has been more favorable to longer-term trends. In the next few years that could continue to be the case or it may change. We’ll only know the answer in hindsight.

The broader point here is that by chasing optimization and trying to find precision in field where there is none (investing) you may end up doing more harm than good. Why? Because if you have found an indicator that actually has value (which is very rare to begin with), obsessively tinkering with it runs the risk of canceling out any benefit. For if various permutations of that indicator go through cycles, you are merely chasing your tail in switching to the one that’s “working,” buying high and selling low, again and again.

I don’t pretend to know what the best moving average or time period is. No one knows because there is no such thing; it is constantly changing as the market environment changes and indicators go through cycles. If you like a 45-day/137-day exponential moving average crossover system better than the simple averages discussed in the paper, great. Who am I to tell you that it won’t work?

More important than your choice of indicator, though, will be whether you’re willing to stick with that indicator when it inevitably goes through tough times and periods of underperformance. Can you avoid the temptation to abandon your system or re-optimize to what’s currently “working”? Most can’t/won’t which is yet another reason why active managers fail; there is no consistency in terms of process and unwillingness to accept the bad with the good.

But what if the process/indicator really is “broken” and should be abandoned? Good question and one that every value manager faced back in 1999/2000.

(Click on image to enlarge)

There is no easy answer to this question. It ultimately comes down to whether you believe a) your indicator never had value in the first place, b) something fundamentally changed in the world, or c) or the market environment is just not favoring your style/strategy in the short-run.

Unfortunately, there is no easy way to differentiate between a/b/c and if it is indeed c the short-run may last longer than you think, testing the patience of you/your clients. Not a very satisfying conclusion but this is the best we can say; for we are not dealing with the laws of physics but instead the unpredictable world of investing.