Dow Theory Special Issue: Capitulation. The ultimate bottom indicator (I)

How to catch falling knives without cutting yourself

There is a German saying which reads “Klasse statt Masse” (class instead of mass, so better little and good than a lot and of bad quality)

Since I don´t have much time to write, and, hence I cannot do lots of “Masse”, I hope that the next posts which I am going to pen, will have, to compensate, “Klasse”. At least the subject matter is very important for any investor. We are going to dissect Schannep’s  “capitulation” indicator and show its importance. I have alluded piecemeal to this important indicator (here), but the time was ripe for an in-depth study which enables us to apply it in real life.

The approximate “table of contents” for the next posts reads as follows:

1.      Why we need capitulation. The importance of having alternative entries when being a long term trader (expressed in trading parlance: “how to generate enough trades so that the expectancy of the trading systems materializes”)

2.      Definition of capitulation. Betting for a mean reversion.

3.       Performance after capitulation: Further advances and declines.

4.      Capitulation and the time requirement of secondary reactions.

5.      Measuring the risk when catching falling knives. So great?

6.      When and how much to buy. Transitioning from a 25% to a fully invested position.

7.      Cumulative capitulations. Is it worth to wait for a second one, if any?

8.      When to sell. Different exits.

8.1.           New lows. Exception: New capitulation.

8.2.           Primary bear market signal.

8.3.           Cyclical bear market signal. Exception: New capitulation.

9.      Capitulation coupled with primary bear market signal.

10. Capitulation coupled with cyclical bear market signal.

11. Capitulation within a primary bull market.

12. Analysis of past capitulation situations. Special reference to the 2008-2009 bear market.

13. Performance measures.

Please mind that as I progress with the writing of these posts, I might change the “table of contents”.

Without further ado, let’s get started.

1.      Why we need capitulation. The importance of having alternative entries when being a long term trader (expressed in trading parlance: “how to generate enough trades so that the expectancy of the trading system materializes”)

 Long term readers of this blog know that in addition to using the Dow Theory, I am also a short term (fully automated) trader for almost ten years. I have several computerized systems. Some of them are pure day trading (“in” at the open, and “out” at the close), some others (with less capital allocated) are very short term by nature with trades ranging from 1 to 5 days.

Please follow me, as I am not digressing.

All these trading strategies were tested and theoretically had a positive expectancy (that is, the ability to generate more profits than losses in the long run). And while there are no guarantees that past systems will continue to perform in the future, my systems, by being quite “raw” and “under optimized” have a good chance of not disappointing me. In the last nine years (the actual “forward” test), with real trades, they have been able to continue to perform.

Therefore, from actual short term trading I have gained two insights:

a)     The actual profit factor (profits divided per losses) is much less than that tested. The future never looks quite the same than the past. This holds true even with robust systems. The real Profit Factor (PF) of my trading systems hoovers around 1.25 (yes, a modest one, but this is real life) whereas it boasted a quite good PF around 1.6 when tested.

b)     And, very importantly, there are lots of dry spells. While my real PF of my trading systems hoovers around 1.25, the system may deviate from the ideal 1.25 PF for a long time, much more than would be "normal" statistically speaking. When things go well, I make profits, but when things go bad, I can be quite a long time making modest wins and modest losses without any meaningful performance, going nowhere. My drawdowns may last as long as six months, something which theoretically would be close to impossible. Why? Because I generate north of 1200 trades per year. So, theoretically a PF of 1.25 with 1200 trades per year should have drawdowns not lasting more than 3 or 4 weeks. I will not do the math here but take my word. However, in real life, drawdowns tend to last much more.  Why? Because even when trading short term, it takes time for a system (even a sound one proven in real life) to reassert itself, to deliver its assumed expectancy. Many times it is not in gear with market conditions and the amount lost in losing trades exceeds the amount won for a longer time that would be statistically normal. It is as if one had a coin where “tails” make me win fifty per cent of the time 125 $ and when “heads” I lose 100 $. Under such conditions theoretically the likelihood of being a loser after tossing the coin 300 times would be close to nil.  However, in real trading, it is a very common occurrence to be a net loser if one restricts oneself to the last 300 trades which implies a profit factor less than 1. Of course I cannot outwit the system. Only patience and faith in the soundness of the system is needed. On the other hand, when the system is in “sync” with the market, I can score for the last 300 trades a PF in the vicinity of 2 (which is darn good). All in all, my PF of 1.25 is the result of the average of long spells with a PF either below 1 or above 2. It is uncommon in trading to find average trades and average performance.

The takeaway is clear: In real life drawdowns may last longer than predicted by sheer statistics.  

If a short term system with many trades has had drawdowns (peak-to-peak) lasting six months (which implies ca. 600 trades), what can we expect of the Dow Theory with at best 2 trades per year? What happens when occasionally we hit a couple of losing trades? Given the paucity of trades, it can take quite a long time to make new highs. “Long time” might mean many years. This is a risk that can put the endurance and patience of the investor to the test. Of course, many years in the "red" does not only plague the Dow Theory, as the track record of buy and hold concerning extent and time in drawdown is even much worse. This is why, in the first place, some investors abhor buy and hold.

We must state in favor of the Dow Theory (any “flavor” whatsoever) that it has had a proven PF exceeding 10, which is something really unheard of in trading systems and which attests to the efficacy and soundness detecting trends of the Dow Theory (more about it here)

It is also an observation I made in my trading that you cannot have your pie and eat it too. If you want many trades so that theoretically you are lifted out of drawdowns quickly you should gravitate towards short term systems. However, there is a price to pay. The shorter the time frame the more noise and less signal (this is the subject for a whole book, and how to measure it). A higher signal-to-noise ratio implies a lower PF (which means money won exceeds money lost by a small measure), a lower PF implies that drawdowns will tend to last longer. I don’t have the exact calculations at hand but a PF of 2 may imply that given 100 trades the likelihood of remaining mired in a drawdown is below 1%. On the other hand, a PF of 1.1 and given the same number of trades might imply a higher probability of ca. 2-3%. The lower the PF the longer the drawdown given the same number of trades (in Latin: Ceteris paribus).  On the other hand, if the trader settles himself with a longer time frame which normally should result in a higher PF, then too little trades mean than in a given time (i.e. 6 months) one can be caught in a drawdown as well. As someone said: “Trading is the hardest way to make easy money”

The Dow Theory by having a PF exceeding 10 (which means 100 won by each 10 lost) gives us confidence to stick to the rules and weather the dry spells. Furthermore, such a high PF tells us that it takes few trades to exit a drawdown and make new highs. Such a high PF also tells us that the number of trades “out of sync” will tend to be few. Thus, if my short term trading system had a PF of 10, my drawdowns would last at most ca. 1 month (and this even accounting for the periods when the trading system is not in gear with the markets).

In other words, if the Dow Theory had a PF of just 1.25, it would be unworkable in real life. Given the scarcity of trades a drawdown could perfectly last 50 years, as it takes trades to exit a drawdown. The fewer the trades, the more time it takes for the system to show its real PF. As Keynes said “The market can remain irrational longer than you can remain solvent”. A low PF with a long term system is a recipe for disaster, as your drawdowns will last longer than your patience or paycheck. Please learn to respect the destructive power of drawdowns.

 

The Dow Theory by having an outstanding PF is workable. However, even with such a high PF, given the paucity of trades, it is perfectly likely that, i.e. following a couple of losing trades, it might take 2 or 3 trades to make new highs. But three trades may imply five or more years waiting time, given that each trade lasts more than one year on average and ca. 1/3 of the time the Dow Theory is out of the market, thereby not generating trades.

We know that since 1960 Schannep’s Dow Theory has had a maximum drawdown of ca. 10% (more about performance measures here). And hence the time in drawdown has been more or less bearable.  But in real life even such a  modest drawdown hurts and second guessing will plague most of the investors. Hence, in order to overcome the temptation to throw in the towel at the worst possible time, we must account for our own frailty which we can only conquer by, on the one hand, knowing our system (i.e. Dow Theory) inside out, and, on the other hand, by trying to generate as many trades as possible without damaging our long term PF. If we are able to generate more trades while keeping a decent PF, we will reduce the time spent in drawdowns. While not necessarily scoring a higher overall performance, our periods in the “red” will be reduced, and hence we are more likely to psychologically have the fortitude to persist. 

As an aside, I am exploring trading with the Dow Theory (in markets other than stock indices, such as 30 years bond and 10 years bond, gold and silver, etc.) with two alternative definitions of secondary reactions (i.e. one requiring three weeks, the other one requiring just two weeks) and with equal amounts of capital allotted to each concept of secondary reaction. By doing this the total number of trades slightly increases and we diversify a little bit (i.e. when a signal displayed by a two week secondary reaction fails, the signal taken with the three weeks secondary reaction wins). While overall performance may not be increased, it seems we smooth out the equity curve (drawdown reduction in both time and extent). Testing  two or even three alternative definitions of secondary reactions with equal amounts of capital allotted to each definition is tedious and time consuming as it cannot be computerized (or at least I don't know how to do it). This will be the subject matter for a future saga of posts if I have the time and energy to do it. One preliminary test with US bonds seems to show promising results. But I am still in year 2011....Test not finished yet.

The holy grail of the Dow Theory would be to find ways to increase trades without impairing the quality (PF) of the new trades. Please mind that I say "Dow Theory" as adding a mean reversal tool might fit into the Dow Theory as per the very writings of Charles Dow.

 

This is precisely what Schannep accomplished with his “capitulation” indicator. Capitulation is just a way of generating additional good quality trades (we will further delve into this, but the return of “capitulation” trades roughly mirrors that of “Dow Theory” trades). While occasionally, even good PF systems have losers, on average, the addition of “capitulation” trades will result not so much in boosting performance, but in smoothing the trading results, which implies less risk (loss) and time spent in losing mode (drawdowns). 

An additional benefit is that of diversification of trading strategies. While trend following may be undergoing a rough patch, mean reversal may be shining. There are no guarantees, but, in general, and over the long run, strategy diversification tends to result in smaller drawdowns both in time and extent than just using one strategy alone. Even though "capitulation" will not add many trades (16 "capitulation" trades since 1960) given that the Dow Theory itself does not generate many trades either, its effect in smoothing drawdowns and being useful when trend following stalls is likely to be manifested more often than not.

I must confess that given my aversion to “mean reversal” trading, I was not so enthusiastic about “capitulation”. Some years (and even months) ago I was more of a purist trend follower. However, our goal is to make money in the markets, and, after all, legendary trader Ed Seykota said “the trend is your friend except at the end when it bends”. Or as the Holy Bible says (Eclesiastes 3:1-8) says “[t]o everything there is a season, and a time to every purpose under the heaven: A time to be born, a time to die; a time to plant, and a time to pluck up that which is planted”. So, under very specific circumstances, there is a time not to be a trend follower.

If the system used to select mean reversal trades is sound, why not use it?

Well, the next post will be devoted to dissecting the “capitulation” indicator so that we become convinced that it is just another master stroke devised by Schannep which competes on par with his Dow Theory.

Readers of this blog stay tuned.

Sincerely,

The Dow Theorist