Saturday, October 20, 2012

A primer on secondary reactions under Dow Theory

My detailed answer to a comment left by bstart.

Bstart, an attentive reader of this blog, left an interesting comment which you can find here:

I quote bstart:

"Based on your notes and chart of DJT of I interpret the SLV behavior differently. Is the following analysis right?

1.     5/16/12 - 26.37 SLV low

2. Reaction to 6/6/12 - 28.51 - higher than 6%, 10 trading days

3. From there expect pullback and move to higher highs

- Instead 6/28/12 25.63 new lower lows (primary bear continues)
- this is new leg in bear market
- look for a new secondary reaction retracing 6%, 10 trading days

4. new reaction on 7/30/12 - 27.36 (sorry for the wrong month !), 6.74%, nearly 1 month

5. pullback to 08/02/12 - 26.31

6. signal 08/16/12 - 27.37 - new bull market signal

Is the analysis right?"

My answer said that points 1 to 4 were OK.

However, where it got tricky and thorny was from point 5 onwards.

So basically, bstart is suggesting an alternative way to appraise secondary reactions, namely taking into account (in a bear market) the rally stemming from the last recorded low.

Here is my answer, which is addressed to bstart but is meant to benefit my entire readership:

Please bear in mind that appraising secondary reactions is not math. Rhea wrote:

“Probably no two students would agree on any rule for selecting and tabulating the important secondary reactions."

All Dow Theorists, past and present, agree that gauging secondary reactions is not an easy feat. Furthermore, it appears that its rules (i.e 10 trading days, etc.) are bent quite often. Even Rhea when he conducted a study of all secondary reactions that occurred until 1931, he labeled as “secondary reaction” movements that didn’t reach 10 trading days, much less the 3 weeks times he advocated in his book. Other practitioners drop the retracement requirement altogether (i.e. Martin Pring). Personally, I find that the Schannep way strikes a decent balance between respecting basic Dow Theory rules (i.e. minimum time requirement, minimum extent requirement and technical requirement) and responsiveness (i.e. by accepting 10 days and even less in very exceptional circumstances).

So, while as from point 5 I will depart from your opinion, I am not meaning that you are wrong, or I am right. Furthermore, your analysis is so well structured that you forced me to check my own analysis again. This is totally positive for us both since we strive to be good practitioners to succeed, hopefully, in the market. I don’t want to be a proud idiot.

What now follows is only applicable when I am dealing with “classical” Dow Theory (that is only two markets: GLD, SLV; DIA, IYT; etc.).

Starting point: The principle of confirmation is vital in Dow Theory.

The last rally that was in “sync” (confirmed) was the one that finished on 6/6/2012. It is not necessary that confirmation is given on the same day, but it doesn’t hurt. So the last rally (counter movement against the hitherto primary bearish trend) finished on 6/6/2012.

While we can talk loosely of “secondary reaction” in one specific market, under Dow Theory, we shouldn’t talk of a secondary reaction if there is no confirmation.

Hence, the only rally we had with confirmation was the one that finished on 6/6/2012.

It is true that SLV hit new lows on 6/28/2012. However, GLD refused to confirm. Had GLD confirmed then we’d had a new primary bear market leg, and you would be right counting the next secondary reaction from this point (but for both GLD and SLV).

Since SLV new lows were unconfirmed, the only confirmed rally I can consider is the one that started from the confirmed lows of 05/16/2012. Hence, the highs made by SLV and GLD on 06/06/2012 remain the only valid “relevant” highs.

Look at the recent gold and silver highs of Oct 4. In that day, I wrote that we had to “reset” the clock in order to count the days for a new secondary reaction. Why? Because such new highs were confirmed and hence negated importance to any previous pullback. The already incipient secondary reaction was aborted. It is the same principle in action.

So both the starting point and the ending point of a secondary reaction must be confirmed.

Back to our “secondary reaction”. To sum up:

The last confirmed bear market low (which is the starting point to count a reaction) was made on 05/16/2012. It would also have been OK if one market made its low on that day and the other on a similar date. The important thing is to look for the last confirmed lows to determine the staring point of a likely reaction (and in a bull market, the last confirmed highs, as we saw on Oct 4).

The last confirmed high was made on 6/06/2012.

This confirmed movement in both the starting point and finishing point is what I’d consider a secondary reaction.

Since SLV made a new low on 6/28/2012 never to be confirmed by GLD, this is what made refuse considering the lower high made by SLV on 07/30/2012 as the “relevant” high to be broken.

But your analysis has a big germ of truth on it. What if there is a divergence that lasts several months when, i.e. SLV continues making lower lows and GLD higher highs? Should it be still considered as the relevant high to be broken by SLV the highs made on 6/06/2012? Do secondary reaction highs become fixed or lose relevance with time? Should we consider newer, albeit lower highs, such as those of 07/30/2012 for SLV? These are legitimate questions and unluckily I have not found the answer in my Dow Theory library (I think I have all books about the subject). However, digging very deep in Rhea’s book (it dawned on me the fifth time I read it, and because I was already in the markets) there are some useful tricks that help us determine “relevant” high and low points. But this is the issue for a future post.

However, in this very specific instance, I feel that the secondary reaction highs were new and fresh enough not to suggest an alternative interpretation of the Dow Theory.

Now you could raise a valid point: “What about your analysis of the Transports you made here? In this case you, Dow Theorist, clearly accepted a lower high as the relevant high and were not so strict with the principle of confirmation.
My answer is as follows: When dealing with stocks, I go full steam with the Schannep “flavor” which includes three indices and merely demands (as with “classical” Dow Theory) confirmation from just two of them.

The primary bull market in stocks was signaled on June 29 when the SPY and DIA (Industrials) broke above preceding secondary reaction highs. The Transports didn’t confirm. So from this point on, I considered the Transports on their own. I didn’t need them any more in order to qualify the bull market. Since I didn't need to resort to “confirmed new lows” as the starting point of un upward secondary reaction or “confirmed new highs” as the finishing point thereof, I just focused on rallies and declines occurring in the Transports in isolation. Hence I loosely used the expression “secondary reaction” referring to the action of the Transports alone. But we have to bear in mind that the market was already in a bull market. This is why even now I am focused on the last high made by the Transports as a valid entry point for latecomers (read this post) . It is a technical valid entry point, albeit it has been determined in isolation. However, this is OK, since we know that the primary trend of the market is bullish, so with the Transports I don’t need to require triple confirmation, but merely sound technical action on its own.
I know a saga of posts concerning the appraising of secondary reactions under Dow Theory is due. They are vital because the more proficient we become in spotting them, the better we will be able to detect primary bull markets and to put our trailing stops.  I hope I find one day the time to do it. In the meantime, I am too busy with real-life  activity and keeping updated this blog with ongoing market action.

I do thank emails and comments that prompt me to check my premises. We all win.


The Dow Theorist.

1 comment:

  1. Thank you for the useful clarifications.