Monday, February 8, 2016

Dow Theory Special Issue: Putting the Dow Theory under Stress-Test (III)

Additional remarks concerning performance under secular bear markets

As a consequence of my second post concerning the Dow Theory stress-test, Alpha Dow posted this comment:

Hi there. First of all -- I love the site. I'm an 18yr disciple of Dow Theory and a subscriber to Schannep and a practitioner myself. There is a dearth of material out there about Dow Theory so you are doing the world a great service. I would like to say that although I agree with most of the article, it is at risk of sounding a bit facile. During the most recent decline from the May high to the October low (approx 14% on the Dow), I believe Schannep was at least 50% invested and most people are probably even more so. I may be wrong but JS himself says people use his service in different ways. I certainly don't just buy the ETFs that he does. I use it as a "regime filter" to know when to be aggressively long and when to play defense. I appreciate that this study is only for academic purposes and to elucidate the merit of the theory, but I think it risks sounding a bit pollyannaish to suggest a 6% loss is all that can be expected in a secular bear. We haven't even talked about transactions costs, slippage, emotions and taxes!

This post further elaborates on Alpha Dow’s comment.

Schannep’s mixes his Dow Theory with his proprietary “timing indicator”, and hence signals derived from the Dow Theory just weight “50%”. In other words, if one gets a Dow Theory “sell” (primary bear market) signal, whereas the “timing indicator” remains on a “buy” mode, he will only sell ½ of the position. This is why as of this writing Schannep remains invested at 50%.

This blog focuses exclusively on the Dow Theory, and does not use “Schannep’s timing indicator”. It is not because I find it defective, but, rather, because the “timing indicator” is not Dow Theory and the exclusive focus of this blog is the Dow Theory. The focus of my post was to assess what to expect under secular bear markets when exclusively applying the Dow Theory. I want to isolate its performance.

In the past, Schannep’s timing indicator has quite closely mimicked his Dow Theory signals (i.e. 1987 crash). However, there is no guarantee that at any critical future juncture it will continue to do so. Hence, at certain instances, such as the current market environment, the Dow Theory may signal a “sell”, whereas the “timing indicator” remains at a “buy”.  It is clear that when mixing both timing devices, there is a higher risk of enduring larger losses. If the current decline turns into a “big” decline, those remaining at 50% invested will endure a larger loss than those who only applied the Dow Theory, and run for the exits on December 11 (date of the primary bear market signal).

I am not saying that using the timing indicator is worse than just applying the Dow Theory. In real time nobody knows what is going to happen, and hence, should the current decline fizzle out, and a new powerful rally emerge, then it would have been wiser to remain invested at 50%. In other words, sometimes it is better to just use the Dow Theory; sometimes it is better to mix it with the “timing indicator” in order not to be so “reactive”.

Of course, the downside of using the “timing indicator” will be a slighter greater tendency to larger losses when markets decline in earnest. So mixing the Dow Theory with the timing indicator will result in the following profile:

a) less whipsaws, as one remains partially invested when there is a failed Dow Theory sell signal (“failed” meaning that the next buy is at a higher level than the sell signal). 

b) slightly higher losses, when the Dow Theory alone did not whipsaw (that is, when the next buy was at a lower level than the sell signal).

We should bear in mind that most Dow Theory sell signals are successful, that is, the next “buy” comes at a lower price level.

Schannep’s own book, gives us an idea of the losses to be expected when using the “timing” indicator versus just using the Schannep’s Dow Theory.

If we go to page 182 of his book (Table 14.2) we see that following a Schannep’s Dow Theory primary bear market signal, the market has further declined -14.6%. On the other hand, the timing indicator further declined -12.5% (both figures were derived by Schannep with data until 2007). In other words, the Dow Theory signaled a “sell” (exit) ca. 2% points before than his timing indicator.

In other words, if I say that Schannep’s Dow Theory tends to exit at ca. -6% from the top, it is reasonable to infer that the “timing indicator” will tend to exit at ca. -8% from the top. Please mind that we are talking about averages and not certainties.

The fact that the Dow Theory exits “earlier” than the “timing indicator” will result in a profile of more contained losses. I am not implying that this is better, since we have to also look at returns. Maybe the timing indicator endures higher losing trades but, at the same time, winning trades are larger. If we go to page 181 (Table 14.1) of Schannep’s book we will see that the timing indicator does not outperform Schannep’s Dow Theory. The combined use of the “timing indicator” and Schannep’s Dow Theory (giving to each respective signal a 50% weight) results in a very similar performance to that derived from just using Schannep’s Dow Theory. Thus, we can gather from Schannep’s book that the combined use of the “timing indicator” and the Dow Theory:

            a) Result in very similar performance to just using alone the Schannep’s Dow Theory.

            b) Losing trades will be slightly higher.

            c) It may be convenient in some instances to keep a foot (50%) in the market and   avoid whipsaws.

Personally, and here enters the psychological factor Alpha Dow mentions in his comment, I abhor overstaying weak markets, and hence I feel more comfortable just using the Dow Theory (and I keep it simple, less complexity). Furthermore, when looking forward, it is more unlikely that Schannep’s Dow Theory be caught unaware by a sudden crash than the “timing indicator”. The fact that the timing indicator avoided past crashes unscathed does not guarantee that it will continue to do so in the future. Of course, the same applies to the Dow Theory, but there is a difference: ca. 2% average more responsiveness (getting out earlier from the top) means that it is more likely for the Dow Theory to escape future crashes.

I have studied Schannep’s Dow Theory whole record of buy and sell signals. Most of the buys and sell signals tend to be signaled at ca. 5-7% from the respective bottom/top. Of course, there are exceptions, but given the definition of secondary reaction given by Schannep (3% minimum movement) it is highly unlikely that buy/sell signals will be signaled at, i.e. 14% levels. This implies that even under very weak markets (2008, 1974) the Dow Theory will signal the exit at close levels from the top. Of course, nothing is carved in stone, and one day (hopefully very far from now) the markets will deliver us a bad surprise; however, Schannep’s Dow Theory puts the odds on our favor, and hence it is not too optimistic to say that a loss of ca. -6% from the top is what we normally can expect. This is not true, if we mix the Schannep’s Dow Theory with the “timing indicator”.

It is true that real life investment is not easy. Somebody said that “trading is the hardest way to make easy money”. It was not my intention to convey the idea that Schannep’s Dow Theory is a panacea under secular bear markets. No, it is not. Investing is hard, and we have to keep emotions at bay. I can say in full sincerity, though, that, at least for me, is the only technical device I have found I fully confide, and which I am able in real life to fully respect and not outsmart. In other words: I pull the trigger when needed with no second-guessing. Of course, this need not necessarily apply to other investors. So if my post expressed the idea that investing with the Dow Theory is facile, it may partly be because to a large extent I do believe so. I regret that in the past I did not know about such a facile investing technique. For me it is like a puzzle that suddenly fits in. Of course, it took me many years of hard work to find it “easy” and have an almost blind confidence in its ability to deliver positive results in the future. I did not find it easy when I read Rhea’s book for the first time.

There are other issues mentioned in Alpha Dow’s comment:

a) Transaction costs: Yes, we have to keep an eye on them. Nowadays, though, brokers’ commissions tend to be very low. Furthermore, the Dow Theory has low turnover. It is not day trading.

b) Slippage: This is an issue that affects all market timers. As to my experience it depends a lot on the broker and the issue traded (it is not the same to sell the SPY, than an illiquid stock). However, even under very adverse circumstances, slippage does not exceed 0.1% of the stock price. A round trade should not exceed 0.2%. Since, on average, the Dow Theory trades less than once a year, slippage, even outsized one, should not be so detrimental to performance. There is an additional “slippage” issue when using the Dow Theory. We know the buy and sell signals at the close but we act at the opening next day. Hence, there is a difference between the closing price, and the real fill at the open. This is the subject for a future post, but I can tell you that there is a negligible difference in performance between results calculated using the “close”, and the “real ones” using the opening price next day. So waiting for next day’s open does not degrade much performance.

c) Taxes. Well, all timing systems are afflicted by taxes (and even to some extent buy and hold). However, there is a good thing with the Dow Theory: The big winning trades tend to be made in time frames exceeding one year; the losing trades (which are not subject to taxation, as are “losses”) tend to have shorter term durations. I am not giving tax advice, as the laws of each country are different. However, in many countries long term gains tend to benefit from a reduced tax rate or be tax exempt. The Dow Theory may capture a significant share of tax reduced capital gains. Please bear also in mind that the Dow Theory, unlike moving averages, is not prone to whipsaws (that is repeated contradictory signals flashed very often), and hence it is not so likely. More about the “built-in” advantage of the Dow Theory regarding whipsaws, here

The Dow Theorist

Tuesday, February 2, 2016

Dow Theory Special Issue: Putting the Dow Theory under Stress-Test (II)

Performance under secular bear markets

On December 7th, 2015 I started what is going to be a new “saga”. Its aim is to put the Dow Theory under a “stress-test”. We will try to think out of the box and imagine adverse scenarios which hitherto have not occurred yet. How would the Dow Theory perform under turbulent and uncharted waters? More about the reasons and premises underlying the present study, here.

There are many market environments which we could label as “adverse”. Of course, “adverse” environments, are also detrimental for “buy and hold” or any other trend following method (moving averages, breakout systems, etc.). Let’s be frank, the best environment for us all is a rising market. If stocks go up, all investing strategies will fare well, even buy and hold.

There are many “adverse” environments: sudden crashes, secular bear markets, weak bull markets prone to failure and reversals, dried up liquidity, etc.

In this post, we will focus on secular bear markets.

A secular bear market is a long period of time (i.e. 10 years) where the stock market goes nowhere (or even with a downward bias). Sharp declines (real bear markets, not just pullbacks) are punctuated by bull markets but, overall, the market is unable to make any meaningful gains. More about the concept of “secular” bull/bear markets, here

A secular bear market is clearly “headwind” for stocks. It is like a huge gravitational force which prevents stocks from gaining ground. Accordingly, buy and hold suffers.

I am not going to reinvent the wheel. In 2013 I analyzed in depth the performance of the Dow Theory under secular bear markets, as was explained here. Hence this post will borrow heavily from the studies I performed in the past. My 2013 study of performance under secular bear markets was a comparison between Schannep’s and the Rhea/classical Dow Theory. As you may expect both “flavors” suffer under secular bear market conditions. However, “suffer” is not tantamount to “losing”, as both flavors managed in the past to post positive results under secular bear markets, and outperformed buy and hold (both in absolute returns and risk-adjusted).

For the remainder of this post (and futures ones of this “saga”), I will exclusively focus on Schannep’s Dow Theory, which is the “flavor” I use in real life (and for good reason, since it outperforms the classical/Rhea Dow Theory).

While defining “secular” bull and bear markets is subjective, I have tabulated the following periods:

From 1954 to 1967: SECULAR BULL
From 1968 to 1981: SECULAR BEAR
From 1982 to 2000: SECULAR BULL
From Mid 2000 to 2013?: SECULAR BEAR.

During secular bear markets, Schannep’s Dow Theory managed to make an average of +5.46% on each trade taken. So, on an overall basis, the Dow Theory made money even when stock indices remain flat at best. We may hope that in the future the Dow Theory will manage to extract some profits out of the market even when there is no meaningful long term movement, and when every now and then huge drawdowns hit the investors. 

Of course, averages are misleading; we should study losing trades under secular bear markets. An average gain of +5.46% per trade, means nothing if the investor has to endure a -40% drawdown in between. Well, the average losing trade amounted to -6.48%. Hence, on average, losses were well contained, and did not endanger capital integrity.

Furthermore, we should focus on the largest losing trades. An average loss of -6.48% may not be indicative of the real dangers besieging the investor. The largest loss endured by Schannep’s Dow Theory during secular bear markets amounted to just -10.45% (Jan 30, 2009). Please mind the date: The Dow Theory largest loss coincided with one of the most vicious bear markets. A largest loss of -10.45% is still bearable, and can be quite easily recovered.

Let’s take a look at the second largest loss. It amounted to -8.33% (Jan 26, 1970). As we can see losses are bearable, and have nothing to do with the drawdowns underwent by buy and hold.

Therefore, if we are to judge the performance of the Dow Theory according to past secular bear markets, we can tentatively conclude that it remains profitable, whereas losses were kept at bay. This is good news, as past secular bear markets where not lenient in the least to buy and hold investors.

However, I feel that past secular bear markets were to some extent favorable to the Dow Theory. Let me elaborate: As I wrote here, the Dow Theory outperforms buy and hold when markets decline (by cutting losses short).

Hence, secular bear markets characterized by deep declines, and subsequent recoveries, are not necessarily “bad” to the Dow Theory. We know that Schannep’s Dow Theory tends to exit at a close distance from the top, at ca. -6%. If during a secular bear market (as we saw in the 1970’s and 2000’s), we experienced two big bear markets in the vicinity of -50%, this would be “good news” for the Dow Theory, as we would exit roughly at -6% from the top, and the market has still a lot to decline. If after the -50% decline (as in 1974, 2009), the market rebounds (even modestly), the Dow Theory stands to make good money.

Let’s imaging a decline of -50%. The Dow Theory manages to exit at just -6% from the top. After such an exit, the market further declines for a total loss of -50%, Following the market bottom, a weak bull market of +35% sets in (which, hence fails to even match the last recorded high). While such a market action is disastrous for buy and hold, for the Dow Theory this is a great environment, as 100 units would become:

100 minus 6% (exit signal, primary bear market) = 94 units left to the investor.
Normally, the Dow Theory flashes its buy signals (primary bull market) at ca. +6% from the bottom, hence a +35% recovery might entail ca. +23% “extractable” profits from the rally (35% total advance from bottom to top, minus ca. 6% “lost” in the entry, minus ca. 6% “lost” in the exit) . Hence we have:

94 units x 1.29 (+35% rally minus 6% “lost” as the entry was 6% “late”) = 121.26 units paper profit
121.26 units minus 6% (exit signal, new primary bear market) = 113.98 units, that is +13.98% profit, whereas buy and hold would have lost:
100 units -50% = 50 units left to the investor.
50 units + 35% recovery = 67.5 units, 
67.5 units minus subsequent -6% decline = 67.5 x 0.94 = 63.45 units, hence a -36.55% loss from the first top.
Please find below graph displaying the core idea:

Secular bear market action with downward bias. The Dow Theory manages to remain positive.
Hence, the Dow Theory is able to make money in an ever declining market provided the reentry price remains below the exit price (something which tends to occur on most of the trades taken in pursuance of the Dow Theory).

The only requirement is that stocks should further decline (which they almost always do) after the primary bear market (sell) signal, and the next buy signal (primary bull market) should be signaled close enough from the bottom (let’s say 5~7%) so that one “rebuys” stocks at a lower price than previously sold.

Since we know that we tend to exit at ca. -5~-7% from the top and re-entry at ca. +5~7% from the bottom, we infer that in most instances a bear market swing from top to bottom exceeding ca. 10~14% will be ample enough to avoid losses. Any such a swing will result in profits, whereas swings not exceeding this threshold will be losers (albeit very small ones). More about “fibrillation” (when the stock indices remain for a long time caught in a narrow range displaying false signals in a future post of this saga).

Past secular bear markets underwent bear market swings of enough magnitude so that the Dow Theory (even the classical one, which normally enters and exits trades not so close from the bottom/top, as Schannep’s) managed to be profitable. Rollercoasters are not necessarily bad for the Dow Theory, just those of modest percentage swings.

I feel it is likely that future secular bear markets will witness ample swings, as those seen in the past, and not just “fibrillation swings”. In other words, I think that future secular bear markets will, to some extent, resemble the “swing pattern” of past instances, and, hence will allow the Dow Theory to deliver positive results. Why am I of this opinion? The answer will be given in the next post of this saga when I discuss “fibrillation”.

Until then you have material to cogitate.

The Dow Theorist

Tuesday, January 26, 2016

Dow Theory Update for January 26: Secondary (bullish) reaction for Gold and Silver against primary bear market signaled today


The primary and secondary trend is bearish, as explained here:

Here is an additional post concerning the likely decline to follow primary bear markets signals:

No secondary reaction has developed yet, albeit it could happen any moment now as explained here.


The primary trend is bearish as explained here.

However, today, SLV and GLD made higher closing highs. SLV has rallied since its lowest closing low of December 12th, 2015 and GLD since its December 17th, 2015 closing low. This amounts to 28 trading days for SLV and 26 trading days for GLD. Thus, the time requirement for a secondary reaction has been amply met today.

As to the extent requirement, both SLV and GLD have rallied more than the minimum move requirement .

As readers of this Dow Theory blog know, when one is dealing with stock indices such as the Industrials, the minimum meaningful movement to consider the existence of a secondary reaction is 3%. However, given that SLV and GDL exhibit a different volatility pattern than indices like the Transports or Industrials, we have to conduct a volatility adjustment that takes into account their volatility.

To calculate the volatility multiplier we take the 30 days average of the daily volatility (percentage change) of the SPY. We do, likewise, with SLV and GDL and divide SLV and GDL volatility by SPY’s volatility. Once we have the multiplier, we apply it to 3% (the minimum volatility for the SPY for a meaningful movement) to determine the minimum movement for SLV and GDL. The spreadsheets below give you the full math:

30-Days avg volt

30-Days avg volt





Min move


The following spreadsheet shows the price advance of both SLV and GLD


SecReac High

Sec Reac Low


% Move


Hence, we see that both ETFs have rallied much more than the minimum volatility-adjusted movement.

Hence, the extent requirement has been met.

All in all, a secondary (bullish) reaction against the primary bear market has been signaled today.

Here you have an updated chart (blue rectangles display the secondary reaction).

Secondary reaction against primary bear market for SLV and GLD


The primary trend is bearish, as explained here.

The Dow Theorist