Wednesday, June 26, 2019

Dow Theory Special Issue: Which markets are better suited to the Dow Theory (and Trend following in general) and the reason why

The power of trends and why it is not wise to disregard a Dow Theory signal based on the belief that the market is already overstretched.

This is another post that I was written on June 19th at Nashville’s airport while being stranded due to bad weather conditions. However,  it has been “matured” and edited today.

When I started this blog some six and odd years ago I even had time to apply the Dow  Theory to US bonds. While bonds have never been the main focus of this blog, I have been always convinced that the Dow Theory is fully applicable to bonds. Just take two related bonds (i.e. 30 years versus 10 years) and done.

Hamilton (who preceded Rhea) wrote that he had no doubt that the tenets of the Dow Theory could be applied to any organized market. More specifically, Dow Theorist Hamilton in his 1922 book “The Stock Market Barometer" wrote:

"The law that governs the movement of the stock market, formulated here, would be equally true of the London stock exchange, the Paris Bourse or even the Berlin Boerse. But we may go further. The principles underlying that law would be true if those Stock exchanges and our were wiped out of existence [...]"

As to whether Dow Theory can be applied to gold and other markets, Hamilton said:

"The averages of South African mining stocks in the Kaffir market, properly compiled from the first Transvaal gold rush in 1889, would have an interest all their own [...] The comparison of that average with the movement of securities held for fixed income would be highly instructive to the economist."

Source: William Peter Hamilton, "The Stock Market Barometer." Wiley Edition, pages 14-15.

All in all, the Dow Theory can be applied to other markets, not just to US stocks. I have personally tried to apply the Dow Theory to markets such as Crude Oil and Heating oil, or Wheat and Corn. Any movement in, i.e. crude oil should be confirmed by heating oil. Same with wheat and corn. My preliminary findings confirm Hamilton’s idea. I think there is some merit in applying the Dow Theory to such markets (so different, by the way, from US stocks). However, and while my research has not been exhaustive, my first findings seem to show me that, while still profitable, the results obtained are not as good as those obtained with US stock indices. In other words, there are more whipsaws (false signals: you get long to get immediately stopped out). In our time frame (i.e. 3 weeks to declare the existence of a secondary reaction, minimum movement 3% adjusted for volatility)  such markets tend to be noisier than US stock indices, “noise” being here, any movement that has enough magnitude (time and extent) to create a Dow Theory signal which later proves to be a whipsaw or, at best,  successful signal with, however, little follow through until being stopped out.

And I feel I know why such markets are “noisier” that US indices. Basically, US indices get moved by two forces. PER multiples and earnings. Interest rates tend to influence PER (lower interest rates, tend to be bullish for stocks). The PER cycle tends to span several years (PER is not jumping from day to day with knee jerk reactions to, i.e., a cut in oil production by Saudi Arabia). Something similar happens with earnings. While they change each quarter, when we take stock indices, changes tend not to be abrupt. This kind of close to “cyclical” pattern underlying the movement of US stock indices clearly favor the application of any trend following device, and particularly the Dow Theory. There will be more “signal” (the underlying forces of the economy) than noise. More predictability than erratic behavior. More likelihood that once a trend has been established, it will continue and not be reversed by any sudden piece of news (this "news" may result in stock market movements which lack enough magnitude to change the primary trend). On the other hand, markets like oil, grains, etc. are moved by countless factors. A drought, a flood, an import ban, an export ban, too many "shorts", rollover date for futures, whatever...Many more factors influence the price movement of such markets and hence, while trends continue to exist, there are more reversals, stop running, "noise", than with US stocks. More variables defining price are at play.

There is, though, a market whose underlying forces are even more simple than those that govern US stock indices. That is bonds. Bonds don’t depend on earnings. Bonds merely reflect interests. If interests go down, bonds go up and vice versa. And the underlying forces moving interest tend to be long term. Interests are not going up 1% (100 basis points) up or down one day based on one whimsical news. Thus, in theory, there should be more “signal” and less “noise” (less false signals, less whipsaws) when we apply the Dow Theory to bonds as compared to the Dow Theory applied to US stock indices.

I feel I wrote some months ago that I was conducting a test by applying the Dow Theory to the TLT (20 years US bond ETF) and IEF (7-10 years US bond ETF). They are close enough to be a good “pair” to apply the Dow Theory. My first preliminary results are very encouraging. There is only one downside. Since bonds are less volatile that US stocks, I feel the total money made is less than that made with US stock indices.  However, risk (losing trades) is also smaller.  Winning and losing trades tend to be smaller (when compared to US stock indices) and so is total percentage made. Maybe one day, when I have more time for blogging, I’ll be able to give final figures.

Another important thought of this post is to never forget to respect trends. Specially those trends that occur in non noisy markets such as US bonds and, to a somewhat lesser extent, US stock indices. At the end of 2018 many people thought that the bull market in US bonds was over. After all, interest had being going down (bonds up) since the 1980's. So many people though that a reversal of trend was at hand (that is a bear market in bonds, interests up). However, the Dow Theory was telling a different story. Depending on the way one appraises a secondary reaction (demanding at least 15 trading days on both indices) on 12/18/2018 a primary bull market was signaled. If one were satisfied with just 15 and 14 trading days on 11/19/2018 a primary bull market was signaled. In any instance, any way you appraise the secondary reaction that leads to the setup of the primary bull market signal, by the end of 2018 US bonds were in a bull market.

Well, this has been a raging bull market and as of this writing (June 26, 2019) both TLT and IEF continue making higher highs. Now, more than six months later since the buy signal was given, we learn that the Fed has stopped raising rates. However, markets in their collective wisdom, were already anticipating several months ago what we now know . If by the end of 2018 many considered the bull market in bonds as dead, now we now that, at least for now, the bull continues. Of course, I am not pontificating that I know the future, since one of the tenets of the Dow Theory is that no one can predict the time and extent of the primary trend. 

Here you have a chart displaying the two alternative primary bull market signals. One obtained by just demanding 15 and 14 trading days for a secondary reaction to exist (plus the extent requirement, of course) and the other one by demanding at least 15 trading days on both ETFs (the 3 weeks Rhea alluded to).

A market rich in signal and poor in noise: US bonds. Nice bull run

As an aside,  I also wrote some months ago that I am toying with generating signals with different time requirement for secondary reactions (i.e. by committing 50% of one's capital to signals given when just, i.e., 12 days are required for a secondary reaction and the remaining capital to signals given when i.e. 15 days are required for a secondary reaction). By doing this, we smooth out our results. However, one cannot take the definition of secondary reaction to extremely short time frames (i.e. 5 days) since then noise would prevail (Rhea also warned us about this). Some times it will be good to have an earlier entry and some times not. However, by dividing one's capital into different time-frames, trading results will be smoothed. More about that in some future post.

Hence, one could diversify across time-frames (i.e. signals based on longer and shorter definition of a secondary reaction) while diversifying the application of the Dow Theory across markets (i.e. 60% of capital to US indices and 40% to Bonds). My obsession with diversification is based on experience. Most investors fail due to long drawdowns which prompt them to throw in the towel in despair. However, drawdown duration relates to the number of trades. We need (winning) trades to get out of a drawdown. Longer term systems generate few trades and hence, when the drawdown hits, its length may span years, thereby testing the patience of the investor. More about the financial and psychological implications of drawdowns here and here.  This is why it is important to generate as many trades as possible without, of course, degrading the system. I am not saying to turn a long term system like the Dow Theory into a short term one, as noise would nullify the strategy. Nonetheless, if one could manage to create more "good quality long term trades", one would probably suffer shorter drawdowns. Hence the importance of Schannep's "capitulation" indicator, as I wrote here.

By the way, I know I owe my readers the next post concerning Schannep's "Capitulation" indicator. I could not write this post at the airport, since there I lacked all my tools. As soon as I have some more time, I'll write it. Too many things, too little time. 

The Dow Theorist

No comments:

Post a Comment