Thursday, September 1, 2016

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

What happens to the Dow Theory when the market enters into “fibrillation” (Narrow ranges)? Looking into the future

The following conclusions may be derived from my first post concerning the Dow Theory and narrow ranges:

1) Historically, the maximum drawdown caused by cumulative losing trades amounted to less than 19%. 

2) This drawdown was not caused by narrow ranges but by a powerful bear market (2008-2009). As an aside, buy and hold lost ca. 50% during the same period. 

3) Thus narrow range conditions have not been able in the past to result in decimating Dow Theorists (no death by one thousand cuts).

4) Even under narrow range conditions the Dow Theory managed to experience modest losses or even modest gains. An attempt at profiling the likelihood of profit and loss depending on the extent of the range pertaining to each specific buy/sell pair was made. 

5) The “Rhea//Classical” Dow Theory is less likely to be whipsawed under narrow range conditions, as it is less prone to giving signals. However, one must not forget that the average losing trade under the classical Dow Theory is larger than that of Schannep (see here). Hence, even under narrow range conditions there is a tradeoff. The classical/Rhea Dow Theory is likely to generate fewer trades, but the losers are likely to be larger.

Therefore the past (+110 years) of market price formation did not result in patterns (repeated narrow ranges with losing trades that pile up) decimating Dow Theorists.

So far for the “past”.

In this post, I will explore the future. Short term traders rightfully say the “your worst drawdown is always ahead". As Zero Hedge says "On a long enough timeline the survival rate everyone drops to zero". And this is true. The more we trade, the longer we are trading (be it short term or long term), the more likely that sooner or later we will encounter never-seen-before market conditions. Thus, we have to imagine how would fare the Dow Theory (especially Schannep’s) if we were to encounter a previously unheard-of long lasting period of narrow ranges, which result in a long string of losing trades.

What if the market continues ranging (as it had in the last 1 and ½ year) for 4 additional years? Unlikely, yes, but we have to brace ourselves for that possibility and know what should we expect to lose (or, surprise, maybe gain) under such conditions.

Such a continued state of repeated narrow ranges, each resulting in failed or almost failed trades is what I call “fibrillation”.

The last +110 years were characterized by no instance of  severe “fibrillation”, which means that in most instances, following a primary bull or bear market signal, the market, true to “momentum” had farther to go (either up or down). Thus, I have shown that, according to past observations, following a primary bear market signal the market declined on average a further -12.52 % (this figure has been updated after the last primary bear market signaled on June 24) which implies no fibrillation (if we add, ca. 7-8% decline from the top until we get the primary bear market signal, and a further -12.5% to the bottom, we get an average downward swing of ca. 20%, which keeps us in the “profitable swing zone”). Please see the table of my previous post reproduced below.  Thus, movements in the vicinity of 20%, in spite of being "narrow" if we compare it with most swings, are sufficient to result in profits or, if the entry was far above the average entry ca. 7-8% off the primary bear market lows, losses are to be modest .

Thus, to begin with, it is not outlandish to say that in most instances swings of only 20% may be considered “narrow”. This is the average downward swing. Upward swings (average distance from bottom to top until the trend is reversed by a secondary reaction which leads to a primary bear market) amply exceeded in the past 20% (I don’t have the exact data but just look at a chart and the upward bias of stocks –at least for the last +110 years- is evident). Thus, a movement of ca. 20% (which accommodates both a "buy" and a "sell") is by all standards a “narrow” movement.

Therefore, even if the future will face us with a long string of ranges of ca. 20%, we can remain quite confident that we will be profitable on most trades. Losing trades should not be life-threatening. While nothing is certain, narrow ranges of ca. 20% should result on average in trades yielding ca. 6%. Some will be more, some less.

From the last post (and the table above) we know that danger begins at ranges of ca. 15% or less. At so reduced swings (by “swing” I mean a movement not interrupted by a secondary reaction leading to a primary bear market signal) it is quite likely that, on average, we will begin to lose money.

What if the future results in many years of “ranging” with the specific swing of each Dow Theory signal being of a magnitude of less than 15%? Would a string of losses cause a drawdown similar to those of buy and hold?

The answer is NO. It is highly unlikely.

For both the "Classical" and even Schannep’s Dow Theory, paucity of signals is the hallmark of narrow ranges, since the completion of the whole sequence of price formations that lead to a primary bull and bear market signal takes time. As far as Schannep's Dow Theory is concerned, we know that at the very least, we need an average of 8 trading days (average of the three indices) and at least two indices should have corrected the primary movement two full calendar weeks. Furthermore, we know that the subsequent rally/pullback must last at least 2 days and have an extent of 3% (which in most instances result in taking more than 2 days to be completed, as 3% is not normally made in 2 days). Furthermore, once the 3% rally/pullback has occurred, we need at least one trading day to break the secondary reaction closing lows/highs. To add an additional hurdle (more time), all these movement must be confirmed. All in all, there is an existential, absolute minimum of time between each Schannep’s Dow Theory signal:

8 trading days + 2 days for rally/pullback + 1 day to break the secondary reaction lows/highs = 11 trading days.

Of course, as far as I have been able to examine, no such “high speed” Dow Theory signal has been signaled in the past, and is quite unlikely to occur in the future. A secondary reaction which has developed in the record short time of 8 trading days implies powerful momentum (it is not to the same to advance or decline i.e. 5% in 8 trading days than in 20 days), and hence, such momentum in most instances, imply that the counter movement (the rally/pullback following the secondary reaction) rarely will be of enough power to retrace 3% in one index in just one day. On the other hand, if the secondary reaction develops on weak momentum (in other words, it takes, i.e. 4-6 weeks to make 5%) in most instances the subsequent rally/pullback will have higher odds (less momentum against it) to materialize in a shorter time. Hence, you can see an emerging pattern: If the secondary reaction is powerful (little time to form) it is not so likely that the countermovement will be powerful too. Conversely, if the secondary reaction lacks momentum (many days to reach the minimum 3% threshold) it is also likely the countermovement will take less time to move 3%. It is rare that we get an 8 days secondary reaction followed by counter momentum of 3% in just two days, and, once again, counter movement of 3% in just one day. Theoretically, 11 days is the “physically” shortest time to produce a Dow Theory signal. In practice this has never happened (as far as I know) in the past. As per my experience, even when we get narrow ranges, it takes at least a couple of months for a signal to be flashed. But beware: a signal is not a closed trade. Even under “accelerated” signals, we need a buy and, later, a sell signal to close a trade. Thus, it is not unreasonable to say that, even under a narrow range which results in “trade acceleration”, we should not expect more than one “round trade” every four months.

Thus, under an extremely “crazy”, never-heard-of, scenario, we could get three or four “buy/sell” signals pair in one year. Let’s assume that the specific ranges for each signal do not manage to exceed 10% (this is quite a detrimental assumption). This would result in a probable average loss of ca. -4% on each trade which translates into a cumulative -12% (assuming 3 round trades). Furthermore, actual losses would be smaller due to (a) capitalization factor, as four consecutive capitalized losses of 4% each amounts to a capitalized loss of –11.53%; (b) some dividends would be cashed in during the time invested in the market.

Let’s further imagine that such horrible conditions last for 4 consecutive years. Let’s imagine that we get no respite; not even one modest winning trade. Our capitalized loss would amount to -38.73% (again capitalization also works in reverse when losing hence diminishing the total amount lost).

Therefore, even under extremely challenging conditions, it is quite unlikely for the Dow Theory (including Schannep’s) to undergo the drawdowns which commonly afflict buy and hold. Furthermore, the recovery from such a reduced drawdown would be swifter (it is not the same to recover from -50% which means a 100% price advance, that recovering from ca. -39%). Finally, such an extreme occurrence might happen every 200 years (in more than 110 years it has never occurred). I feel that a strategy that is likely to undergo a -39% drawdown once every 200 years, is safer (at least for me) and more “drawdown-resistant” than buy and hold for which drawdowns of -50% or more (see 1932) are frequent occurences.

Of course, everything comes at a price. Narrow ranging periods will normally imply (temporary) underperformance for Dow Theorists (and trend followers in general) versus buy and hold. And, be sure, we will get clobbered by a crowd on “unbelievers” during the time that "buy and hold" shines. We will make modest losses or marginal gains, whereas buy and hold might even eke out modest profits if there were a slight upward bias (as it has been happening since the beginning of 2015). However, this lack of correlation with buy and hold is good for investors in order to diversify. It is not carved in stone that the market must always oblige trend following. However, deep knowledge of the Dow Theory, its patterns, magnitude of swings, etc. helps us remain calm and confident, as the Dow Theory is indeed solid, much more solid, by the way, than normal breakout systems or moving averages (more on that at the end of this post).

Recent market activity is illustrative. Since roughly February 2015 the S&P 500 has been going nowhere with repeated false breakups and downs and extremely modest follow through following each Dow Theory “buy” and “sell” signal. This has been a challenging market environment. Our “real time” experience has resulted in just the following signals:

Long (invested) from previous buy signal of October31st, 2014

Please mind that the proper way of counting is coupling buy with sell signals. You buy a house, and you sell it some years later. You don’t start with the date you sold a house and when you bought a new one.

Thus in our case, the sell signal of August 2015 is not to be considered a whipsaw, but merely the exit (a small profit) related to the previous buy signal of October 2014.

Thus, since the beginning of the “ranging” (February/March 2015) as of this writing we have had two completed pairs of “buy/sell” trades.

The current “buy” (August 11th, 2016) has not been matched by a “sell” and hence we don’t know whether fibrillation will continue or we will finally get a decent trend which will result in profits, and almost as important, in a belated sell signal (hence decreasing the time frequency of signals).

Thus, as of this writing, we know the following.

  •       Roughly in February 2015 the uptrend stopped; only marginal progress was made by US indices.
  • As a consequence of the end of the uptrend as primary bear market signal was signaled on August 20th, 2015. Such signal ended the “buy” made in October 2014.
  • Thus the first ca. 6 months (February to August) of narrow range merely served to end a profitable trade, and not to whipsaw us, Dow Theorists. The end of the trend (as with any end of any trend) served us to exit a trade which did not originate during the narrow range period.
  • As from August 2015, the real “problems” (whipsaws due to narrow ranges) begin. 
  • From August 20th, 2015 to June 24th, 2016, that is in ca. 10 months, we have had two completed “buy/sell” pairs which amounts to two buy signals (which imply two bullish secondary reactions against primary bear markets) and two sell signals (which imply two bearish secondary reactions against primary bull markets).
  • The first trade (October to December 2015) lasted ca. 2 months. Once again we see that, even though there was no real follow through after the entry due to the "narrow range" condition, it is not so easy to reverse a signal (2 months is more than the theoretical 11 days).
  • We were out of the market (time between the “sell” of December 11th, 2015 and the “buy” of April 13th, 2016) ca. 4 full months. This is important: Even under a trendless market, it was not so easy to complete the setup leading to the next buy signal of April 2016. This is quite far from the theoretical 11 days for a signal to be signaled.
  • The second trade from April 13th,.2016 to June 24th, 2016, lasted ca. 2 months. Once again we are quite far of the dreadful “11 days”.
  •  If we average (I know, I know: two observations is not statistically significant, but it gives us a pretty good idea of what to expect, as to losses and time development when markets are lull), we get less than -1% loss for each trade and 2 round trades in ca. 10 months (from August 2015 to June 2016); this amounts to ca. one buy/sell every five months. Nothing to do with 11 trading days.  

Since “losses” are made when the pair “buy/sell” is completed, we can see during the ca. 10 months that span from the end of August 2015 to 24 June 2016, we just had two whipsaws. 

The first one (trade closed on December 11th, 2015) resulted in a theoretical win of 1.23%. Strictly speaking it is not a whipsaw, as it was a winning trade (very modest one, but winner after all).

As to the trade that was closed on June 24th, 2016, there was a -2.34% loss. The cumulative loss (no dividends included and no commission/slippage factored in) amounts to -1.14%. Here you have a table displaying the math:

Hence, even under a quite adverse scenario (maximum range from close to close of ca. 16.75%) and specific ranges for each trade of less than 15%, the Dow Theory did not result in great losses. In bears mentioning that the Brexit’s sudden gap of June 24th, 2016, which suddenly reversed a mildly bullish condition into a bear market signal, is really an adverse and rare market condition. So we are now really undergoing a “life” stress test.

Here you have a chart displaying the narrow range which has engulfed stocks for the last 1 and ½ year (orange rectangle) and the two specific narrow ranges (grey rectangles) which pertain to each specific buy/sell pair (trade). The blue arrows display the date of entry, and the red arrows the date of exit.

Ranging environment. Tough for trend following

Please mind that the grey rectangle on the right side of the chart does not engulf the latest high made by the SPY before the Brexit breakdown of June 24th, 2016. SPY's closing higher high was not confirmed by the other indices and hence from a Dow Theory standpoint it did not qualify to end the then ongoing bearish secondary reaction against the then existing primary bull market. Thus, the grey rectangle  merely includes the last closing highs confirmed.

As to the current “buy” of August 2016, since it has not been closed, we cannot know the final extension of its swing and whether it’ll be a narrow range or a good upward thrust.

What would happen if the present market conditions were to last, let’s say, 3 years more (so ca. 4 years in total)? Even though nothing is certain, if we assume an average loss per year of ca. 4% due to a couple of losing whipsaws a year (more loss than we have hitherto endured) and we assume four years of pain, we are talking of a total loss of ca. 16% (somewhat less after allowing for capitalization) which is much less than the drawdowns endured by buy and hold.

And if a “crash” comes, that would be very good news to the Dow Theory, as a “crash” is “trend”. A “crash” is not a trading range. Please mind that most of the outperformance of the Dow Theory is predicated on bear markets. We exit near the top of the primary bull market and buy at a significant lower level, post-crash). While painful to see (we are humans after all, and one day one crash will entail the destruction of civilization, and the end of the Dow Theory, so even market technicians, feel uncomfortable when the market hints at Armageddon), the market environment of 2000-2003 and 2007-2009 is the good one for the Dow Theory (provided the financial system survives): It gives us the opportunity to exit quite close to the top (let’s say at. Ca. 7-8% as it has been determined by Schannep) and, after the dust hast settled, buy at significant lower levels and quite close to the bottom (ca. 7-8%).

Bottom line: While the last months have clearly been adverse for the Dow Theory (and for any kind of trend following), the Dow Theory is unlikely to suffer deep drawdowns. Until a big decline comes, or until a big uptrend materializes buy and hold is likely to temporarily outperform (specially when there is a light upward bias). Real time experience remains within the boundaries of our theoretical considerations. Narrow ranges per se are unlikely to result in major losses. Furthermore, round trades are not signaled with such a frequency as to cause death by one thousand cuts.We have more to fear from powerful bear markets and gapping down action (something which will be dealt with in the last post of this saga).

Some readers might be questioning how fared moving averages under the same time period. Not very well. The chart below shows a 200 day moving average and the corresponding trades (“buy” when close is above MA, “sell” when close is below MA). There have been 6 trades (all of them losers) und, as with the Dow Theory, now there is a “buy” which hitherto has not been reversed. 

200-day MA unable to cope with narrow range scenario.

Thus, we can see that a 200 day moving average has undergone six whipsaws, whereas the Dow Theory only two or even one, as strictly speaking one of our “narrow range” trades was a modest winner. More about the net superiority of the Dow Theory versus moving averages here.

The next post of this “saga” will “stress-test” the Dow Theory under abortive cyclical bull markets. To this end, we will define and follow Schannep’s definition of “cyclical” bull/bear markets (which in my opinion is the best and most actionable one). Readers of this blog stay tuned.

The Dow Theorist

No comments:

Post a Comment