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

Definition of capitulation. Betting for a mean reversion

We continue with our saga concerning “capitulation”. I feel this saga is going to keep me busy for quite some time. Here you have the link to the first post.

http://www.dowtheoryinvestment.com/2019/03/dow-theory-special-issue-capitulation.html

Capitulation is a bottom detector tool devised by Jack Schannep. Thus, capitulation is by nature mean reverting. It does not follow the established trend but it tries to detect the change thereof. Capitulation has been devised only for the long side. When the market is severely oversold, capitulation bets for a change in trend. I see capitulation as an elastic gum which has been too stretched. Some contracting is due.

While betting for a mean reversal is kind of abhorring for trend followers, Jack Schannep makes clear that both Charles Dow and Robert Rhea were eclectic enough to depart from strict trend following on selected occasions.

I quote from Schannep’s website, which at the same time, quotes Charles Dow and Robert Rhea:

“[a] word [capitulation] Charles Dow never uttered but he clearly alluded to with his discussion of the final phase of bear markets when he described “distress selling of sound securities, regardless of their value…..” Robert Rhea described it as “a semi-panic collapse (and) it is wise to cover short position and even perhaps make commitments for long account” (emphasis in original).

Therefore, if the indicator is sound we should be pragmatic. Furthermore, as I will show in a future post, even if the indicator were to fail on a rare occasion in the future our losses would be well contained for three reasons: (a) On most occasions there are stops; (b) even when the knife continues falling, the initial commitment suggested by Schannep is modest (25% or 50% of total capital depending on the kind of trading); (c) the indicator signals a “buy” on extremely rare occasions when the market is really very oversold.

As I wrote in the first post of this saga, capitulation is important for three reasons:

a)     It creates more trades (not many, but something is something)

b)     The trades are very good quality trades (in the sense that the money made greatly exceeds the money lost, which amounts to a high Profit Factor).

c)     It introduces a measure of diversification to our trading by introducing some mean reversal trading. Strategy diversification, even though it does not necessarily increase absolute performance, results in smoothing the equity curve (less drawdowns). When trend following (strict Dow Theory) may be temporarily lagging behind, mean reversion may shine.

While I feel there is some minor part of the indicator which remains proprietary, the basic tenets of the capitulation indicator are as follows:

1.      One calculates a 10 week exponential moving average of the closing prices.

2.      One calculates the ratio between today’s close and exponential moving average at the close.

3.      If the ratio reads 0.9 (that is for example, current close at 90 whereas the exponential moving average stands at 100) and such a ratio is confirmed by the INDU, NYSE and S&P 500, then we have reached “capitulation”.

Once we have capitulation we are advised to buy on the very same day close (with equations one can calculate the closing prices that would render the ratio at 0.9 or below that figure one day in advance so that one can buy at the close) or at next day’s open.

I created a capitulation indicator that plots the current reading of the ratio between the current close and the 10 week exponential moving average. Since weekly bars can be tricky (as the current bar is not completed until Friday), I work with daily bars and avail myself of a 50 days (5 weeks) exponential moving average.

The coding in Easy Language® reads as follows:

input: length(50);

value1=average(c,length);

If value1>0 then value2=c/value1;

Plot1(value2, "ratioMA");

With the aid of TradeStation® I created three charts each one containing one index (INDU or NYSE or S&P 500). Below I plot the indicator which shows me the ratio. When the reading of the three charts falls below 0.9 then we have reached capitulation.

My own indicators faithfully signaled capitulation the very same day it happened (12/24/2018), so I needed not Schannep’s “crutch” in order to know when capitulation occurs. 

Here you have the one chart depicting the situation on 12/24/2018 (blue circle highlights the date) where the ratio fell below 0.9 on the three indices. As I told you, I use daily bars and my software calculates de 50 day (10 weeks) exponential moving average.

 

Capitulation on 12/24/2018. Chart created with TradeStation (r)

Schannep in his book analyses in depth the concept of capitulation. Readers really intent on applying this indicator are well advised to read chapter 8 (pages 79 to 95) of Schannep’s book “The Dow Theory for the 21^st Century”. There are no shortcuts. This post is not enough. Go and read the book.

Schannep in his book makes a conclusive case concerning the soundness of his indicator. He compares the accuracy of “capitulation” when signaling bottoms against other well-known similar indicators (i.e. the put/call ratio) and, if the past is to serves us as a guide, capitulation emerges as the most accurate bottom indicator by far.

Furthermore, there is a thing I particularly like about “capitulation”. It is its simplicity. Other competing indicators are more complicated by design. The less parameters, the better when it comes to creating a robust indicator. A ratio between the closing price of 0.9 or less between the closing price and the 10 week exponential moving average very seldom occurs. It really measures a much stretched elastic gum. Since 1960 there have only been 16 instances of capitulation. Hence, the accuracy of the indicator. So clearly, capitulation does not result in overtrading.

More about “capitulation” here on Schannep’s website.

Well that’s all for today.

Next post will deal with what follows capitulation. We will be studying further advances and declines (on rare occasions) following the capitulation signal and the time such advances last.

Sincerely,

The Dow Theorist

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

Performance after capitulation: Further advances and declines

At last I found some time to pen a new post concerning capitulation. 

The proof of the pudding is in the eating. In this post, we will describe what happens after capitulation has occurred. Do we get strong rallies after capitulation? If yes, then the indicator has some value. 

As a reminder, I have already written two posts concerning capitulation. If you are new to this blog, please read them so that you get acquainted with capitulation.

http://www.dowtheoryinvestment.com/2019/03/dow-theory-special-issue-capitulation.html

http://www.dowtheoryinvestment.com/2019/03/dow-theory-special-issue-capitulation_22.html

Disclosure: all the numbers I will show have been derived from Schannep’s book (“The Dow Theory for the 21stCentury) and, more importantly, their updated version to be found on Schannep’s website, "The Dow Theory.com".

To begin with, capitulation is not signaled very often. As I derive from Schannep’s website from December 31st 1953 until December 24^th 2018 (last occurrence) there have only been 16 instances. Hence, “capitulation” will not result in “overtrading”.

We will highlight two features that attest to the accuracy of capitulation detecting impending rallies. These features are:

·        Closeness to detecting bear market bottom.

·        Subsequent rally after the capitulation signal.

As to the closeness or accuracy of the capitulation indicator detecting the bear market bottom (that is catching the falling knife), it has called the exact bottom of the bear market on 4 occasions. The first one, the day of the 1987 market crash (October 19^th, 1987). The second one, on August 31^st, 1998  (the Asian financial crisis, another day when panic was the prevailing mood due to the suddenness of the collapse). The third one on October 9^th, 2002 (the last confirmed bear market bottom of the 2000-2002 bear market) and, more recently, last December 24^th, 2018, a date when, to say the least, optimism was not prevailing on the markets. On such four instances capitulation suggested the opening of a long position on the very day of the market bottom. Of course, in real time it is difficult to act on such a “buy” signal, as despondency and fear prevails. Hence, the importance of knowing inside out the entrails of this indicator, so that the trader overcomes fear when prompted to act. In theory it is very easy to look at the past track record (and especially at market crashes) and think that one will keep calm next time and pull the trigger. However, in real life, it takes a deep conviction to isolate yourself from the hysteria and go long. In the end of the day, Schannep’s long term outperformance is being built by buying on really nasty days (this also includes buying “normal” Dow Theory signals given when everyone is pessimistic). All in all, as with short term trading, you cannot “pass” any single trade because you don’t know which one is going to be the big winner.

Of course, not all capitulation signals are given at the exact day of the bottom. The median time from the signal to the actual bottom is 2.5 days, whereas the average time to the bottom is 13.4 days. The noticeable divergence between the median and the average is due to the fact that most of the capitulation signals are given very close to the market bottom. 9 out of 16 signals have been given within 3 days or less to the actual bottom. That’s accuracy! However, there have been some other instances where the final bottom was more distant in time (i.e. on July 19th 2002 it took an additional 57 days to reach the bottom). Such rare occurrences increase the average time to the bottom but, since most of the signals are close to the bottom, the median remains small (2.5 days). Therefore, the median is more representative of the accuracy of the indicator than taking the average.

Furthermore, we should not get too scared with those signals that were given too early as the decline that follows the signal until the final bottom is made tends not to be that big. In other words, even if in some instances we are a little bit early what happens afterwards is not catastrophic. Normally, there is no dramatic decline following the capitulation signal (the exception is the bear market of 2008-2009 which will be the object of a specific post, as it is good to see how Schannep sailed such turbulent waters). More specifically, the median percentage decline that follows the capitulation signal amounts to 1.4%. If we take the average, the subsequent decline averages 4.4%. Once again the average is larger than the median as in most cases the subsequent decline following the capitulation signal is muted. However, there are some few instances (i.e. the horrible bear market of 2008-2009 with three capitulation signals) where the decline following the signal was big (20%, 8.8% and 8%. More about such signals in a future post). However, we must bear in mind that Schannep does not recommend to go long 100% after a capitulation signal. Depending on the way you apply Schannep’s Dow Theory either a 25% or 50% position is opened at the close of the capitulation day (or next day’s open). In other words, even in those rare instances where catching the falling knife results in sizeable declines following the signal, the total loss is well contained. As an appetizer to my future post concerning the bear market of 2008-2009 and how Schannep sailed through it, it suffices to say that we greatly outperformed buy and hold.

As to the rally that follows capitulation, Schannep has tabulated the price advances in a 6 months, 1, 3 and 5 years window. For the full details, please go to Schannep’s website.

After six months the median advance amounts to 17%. The average amounts to 15.9%.

A six months window is important, as short lived Dow Theory signals tend not to reach one year. Hence, it is good to know what is likely to happen in case the rally fizzles out after 6 months. In other words, if we act on capitulation day, we are “long”. As per Schannep’s Dow Theory we will get out either if the cyclical bear market definition is met (-16% on both the Industrials and the S&P 500) or a primary bear market signal is given. As I wrote here, roughly just 5% of the exit signals correspond to the bear market definition being met. The overwhelming majority of signals correspond to the “typical” Dow Theory signal.

 Hence, while things can change in the future, we can reasonably infer that roughly 5% of the “exits” are going to be given by the bear market definition being met, which implies a loss from the highest point of the rally of ca. -16%. Please mind that this is not the total likely loss of the trade, since even if we got a modest +10% rally, the -16% is counted from the top, not from the entry on capitulation day. The average rally amounts to 15.9% from capitulation day. Hence, a loss from the top of ca. -16% would basically result in a very modest losing trade. 

As to the majority of “normal” Dow Theory exit signals we also know that on average Schannep’s Dow Theory exit longs at an average distance of 7.7% of the tops. In other words, even if we got a very modest rally of, say, 10% following capitulation day, that trade would remain a modest winner.

Thus, even under a quite adverse scenario trades taken under capitulation signals are very likely to be winners. 

After 1 year the median advance amounts to 26.5% and the average 25.7%.

A one year time frame is important because the average trade taken under Schannep’s Dow Theory tends to last between 1-2 years. In other words, a one year window shows us whether during the first year of life of a “normal” Dow Theory trade we will have enough tailwind. Please mind that on capitulation day, the trader is advised to open a 25%-50% commitment. However, a 100% long position will not be achieved until a bull market definition (+19% on both the Industrials and S&P 500) is met or a Dow Theory signals has been given. Hence, “capitulation” trades do not lie in a vacuum, as they either get closed out as losers or evolve into a full buy signal. Hence, it is important to know whether the new rally is likely to become a powerful one or not.

Of course, a ca. +26% one year rally does not imply the absence of turbulences (declines) during the development of that year. Thus, even though, we can have a powerful rally following capitulation day, it could happen that in between a cyclical bear market definition would be met, and hence the long position is to be closed out. However, this has only happened one time in the past (September 20, 2001). The other 15 capitulation trades were not aborted by a bear market definition signal, which means that the average ca. +26% rally did not meet a turbulence (decline) exceeding -16%. This attests to the strength of the rally that follows capitulation.

It would also be possible that, even if no bear market definition (-16% on both the Industrials and Transports) is met, a typical Dow Theory primary bear market signal is given and hence the investor cannot ride the average ca. +26% rally till the end (stopped out). However, even if one were stopped out, the likely outcome is a modest winning trade, as we know that, normally, we are “out” at a distance that normally averages -7.7% from the top. Of course, there are no guarantees and trading with real money is not easy. However, deep knowledge of the “odds” and the soundness of Schannep’s Dow Theory help us harness the strength and will not to flinch.

Once again, I encourage my readers to go to the original. Read Schannep’s book at least 5 times until it becomes dog-eared. Additionally, go to his website and immerse yourself in the numbers.

 

Sincerely,

The Dow Theorist