Friday, February 20, 2015

Dow Theory Special Issue: Dow Theory versus Moving Averages

Which is a better timing system?

The post you are going to read was overdue, as it is vital for any trend follower. I could finally find some time to put some ideas together.

It is well known that moving averages (hereinafter, “MA”) are used by investors to help “timing” the markets. So, as with other systems (breakups, etc.) MA share similarities with the Dow Theory, as we all are involved in the business of avoiding drawdowns when the trend turns really negative, and hopefully, over time, try to outperform buy and hold.

What prompted me to write this post was an email that some time ago sent me renowned Dow Theorist Schannep, of “”. Schannep’s email contained a valuable piece of research concerning MA. You can find the study about MA, here.

The MA study was based on applying to the S&P 500 the following rules:

1. Buy the S&P 500 every time it crosses above and closes above its 10-month exponential moving average (EMA) on a monthly view.

2. Also, buy the S&P 500 every time its monthly relative strength index (RSI) crosses upwards through the 30 level.

3. Sell the S&P 500 every time it closes on an end-of-month basis below its 10-month EMA, and reinvest the proceeds in 3-month T-Bills.

So, we are comparing apples to apples, as the studies I have conducted in this Dow Theory blog have been based on the same index as the MA study, namely the S&P 500.

What follows are the observations I wrote to Schannep. Read them carefully, as you will gain many insights as to the net superiority of the Dow Theory over MA. Market analysts tend to lump together the Dow Theory with other “timing” techniques; however, while similar in objectives, the Dow Theory has no match.

So here you have my written observations, with very slight editing in order to make them more comprehensible to my readers:

Dear Jack,

Thx for forwarding me this piece of research. I do appreciate it, and I encourage you to send me more info you may deem interesting in the future.

My two cents, which might be of your interest:

I am not quite fond of using RSI in order to assess the long-term trend (i.e. the cyclical bull and bear markets spotted by the Dow Theory). We shouldn’t forget RSI is not a trend detector but an oscillator. In any instance, the author fails to mention the period used for calculation of his RSI (14 months? As the default setting in charting software is 14??). I found great use for RSI indicators for short-term trading, but this is another story…

   The RSI “rule” is a poor substitute of your capitulation indicator. I find much more solid yours.

  If a moving average does a decent work, I am leery as to adding an additional rule like RSI. It looks  great in back tests, but we are adding one degree of freedom, which makes this rule less likely to be so effective in the future when it comes to spotting long term bottoms.

   Clearly, timing reduces risk (the standard deviation of returns). Ditto for the Dow Theory.

   As you and I well know most of the years “buy and hold” outperforms timing systems, whereas timing systems underperform. While you and I understand the nature of such an “underformance”, most investors would not stick to the timing system when in a good year, the S&P clearly outperforms the timing system. The great problem for the average investor is not of skills but of psychology.

When it comes to the specific figures, the MA system is no match for the Dow Theory. It should convince us both (even more, since we are “true believers”) of the hidden treasure contained in the Dow Theory:

    The MA system trades last significantly less than those taken in accordance to the Dow Theory, be it the “classical” or yours. Remember that in an almost identical time period (from 1871 to 2012), the classical Dow Theory only resulted in 38 transactions versus 113 for the moving average (MA) system.

In real life, we should factor in at least 0.5% percent slippage and commissions per round trade. The less trades, the better. Thus, in real life, with so many trades, performance would be impacted to a larger extent in the MA system.

Furthermore, in many countries, the holding period that separates tax exempt capital gains (or reduced tax rate) and fully taxed gains is 1 year. For a taxable account, the longer the average transaction the better. The Dow Theory tends to fetch long-term capital gains and when losing tends to be shorter term (which is good). The MA system would be not as tax friendly as the Dow Theory in many Western countries, and in real life.

    Again, we see the beauty of the Dow Theory: Much larger outperformance (ca. +2%) with fewer trades and less slippage and commission.

    The MA system was in the market ca. 75% of the time; the Dow Theory only 2/3 or 66% of the time. With less time the Dow Theory manages to extract more profits from the market.

   The worst trade (greatest loss) for the MA system was a daunting -35.6% which is much larger than according to the Dow theory (of any “flavor” whatsoever). As I posted here, the greatest loss for the classical Dow Theory was -19.33% in 2008. If we follow Schannep’s Dow Theory, the largest loss was a mere -10.45% The average loser is very similar (ca. -6%). Don't forget that drawdowns kill the investor both financially and psychologically in real life. The Dow Theory does a remarkable better job at containing the worst case scenario.

    Only 47.8% are winners in the MA system, versus ca. 76% in the Dow Theory. This says a lot as to the accuracy of the signals and the ability to spot real breakouts and not fakeouts.

     Win/loss return: Again, it pales by comparison with the Dow Theory. Two worlds apart.

All in all, I get the feeling of riding a Porsche Cayenne with the Dow Theory versus a Mazda with the MA system. However, a Mazda is better than a bike. Your Dow Theory "flavor" is like the 4WD Porche Cayenne and with turbo 6 cylinders.

While the MA system is an improvement over naïve buy and hold, it certainly is years light apart from the Dow Theory.

Of course, everything comes at a price. It is much easier to program a MA system and disconnect your brains, than applying the Dow Theory, which ultimately is subject to human judgment and takes years to apprehend it (at least this was my case, and I still learn new facets every day). So psychologically, for many investors, the MA system may be easier to accept and understand than the Dow Theory.

What follows are some observations not included in my email, which I would like to add:

I agree with the study which used monthly bars in order to avoid the noise inherent with the use of daily bars, and the whipsaws resulting therefrom (false signals quickly reversed by the daily movement). However, the use of monthly or even weekly bars when using MA has a drawback: Even though they may have performed well in back tests, forward-looking losses could be scary if the market plunges by, i.e., 20% in a couple of days. MA based on monthly or weekly bars, may signal the exit too late. 

On the other hand, the Dow Theory is applied on daily bars, but it is based on patterns, not on one bar arbitrarily crossing a parametric MA. The close of one daily bar crossing a MA is just noise. However, under the Dow Theory, one bar is not noise, as it integrates within the pattern it helps to form. However, while not prone to “daily noise”, the Dow Theory allows me to react on a daily basis (Dow Theory signals). Traders say that your worst drawdown lays ahead in your career. When looking forward I feel more confident about loss containment when using the Dow Theory: I get the best of the two worlds: I can get signals on a daily basis (which helped us in the past to avoid crashes like the one of 1929 or 1987, as explained here), whereas, unlike with MA, it is not so easy to be blinded by daily noise. Thus, Dow Theorists get their bull and sell signals (which happens seldom, as well as with long term MA, Rate of Change, etc. when applied on monthly charts) on a daily basis before catastrophic losses occur.

Finally, we should remember that the Dow Theory does not have a parametric nature. MAs rely on parameters. What has worked in the past (i.e. a 200-day  MA), may not work in different market environments. On the other hand, the non parametric nature of the Dow Theory makes it more likely to perform well under a variety of market environments. While nothing is certain when it comes to investing, the Dow Theory is likely to be more stable and  to perform well (and avoid drawdowns) in the future than MAs. More about the importance of non parametric systems, here. The parametric issue is vital, and serious investors should cogitate (and look for resources on the web) about it.

Have a nice weekend.

The Dow Theory

No comments:

Post a Comment