- This blog has a copy of all header posts that I publish anywhere, so that those interested in seeing what my thoughts are on the markets can find them easily.
- I will be answering questions and responding to comments, so feel free to respond to any posts and I will see your comment even if it is not on the most recent post.
- If you're interested in seeing any intraday charts I post, I do that on twitter, and my twitter handle is @shjackcharts.
- The charts in the posts are as large as I can practically make them. if you would like to look at one more closely, click on it, and the link will take you to a larger version at screencast. If you click on that again, you will get a full page version, and can use the resizing function on your browser to enlarge parts of interest further.

Saturday, 29 June 2013

Brave New World Series: 2 - Fleas and Fibonaccis

I started writing a post talking about the 2011-13 rising wedge target and after a lot of work I have divided that into two posts that I will post on today and tomorrow.

This first post will lay out the theoretical basis for my analysis and is going to be a little dense. If you don't much care for math and TA theory you can skip onto the next in the series which I will be posting tomorrow and see the short version with an explanation of the current setup and the two previous examples on SPX that I am treating as comparable.

Before I look at the other two examples I need to explain my view of bull and bear markets historically on SPX, as my first post in this series looking at secular cycles on equities was considering these cycles from a conventional TA standpoint. You can see that first post here. What I'm going to do today is to look at both primary and secular cycles from another standpoint that works better in my view.

Historically there are two main schools of market analysis and both have strong merits in my view. The first is fundamental analysis (FA), which looks for investment instruments that are relatively good value compared to other comparable instruments, in the expectation that concentrating on such relatively undervalued instruments will allow practitioners to outperform the market. This is a sound strategy and I've seen good practitioners of this outperform the market fairly consistently in equities, though trying to do this between asset classes seems to have much more mixed results.

The other school of market analysis is technical analysis (TA), and the foundation of TA in my view is purely mathematical, based on the natural tendency of much in the world around us to fall into mathematically based structures. Classic examples of this of course are Pi and the golden ratio, both extensively studied by mathematicians since antiquity and the golden ratio is the source of Fibonacci numbers and ratios which have been derived from it. TA uses the historical performance of mathematical structures on markets to predict future moves and again, good practitioners can outperform the market fairly consistently.

Both forms of analysis are attacked as pseudoscience by many academics on the basis of the efficient-market hypothesis, which has three variants, but essentially argues that markets incorporate all available news and information at any given time, and as future news is inherently unpredictable, markets must therefore by extension also not be predictable. The weakness in this theory is that it's either obviously untrue, or that patterns forming in markets somehow anticipate the news. A good example is the chart I posted on 23rd May showing the broadening ascending wedge on SPX. You can see that here. Since then that wedge has tested support, bounced, then broken down to make an almost perfect 38.2% fib retracement, and that's very much what you would have looked for after seeing that chart. This sort of thing happens too often to be chance so either patterns like this somehow anticipate future news, or the news is just nowhere near as important as everyone seems to think. The media, all seemingly enthusiasts for the efficient-market hypothesis, spend a lot of time weaving the news and market moves together into an ongoing narrative, but correlation is not causation, and I think it's obvious that news and information can only be part of the overall picture here.

I would compare efficient-market hypothesis with rational choice theory, another very weak theory that is in effect a rough working assumption to underpin economic theory in derived fields until assumptions that better describe reality can be developed.

In terms of schools of TA there are quite a number of those, and I'm not familiar with them all. There's more than one way to skin a cat however and I've seen some really impressive work done with Elliot Wave, volume levels, bollinger bands, moving averages, tick analysis and gap analysis. My own approach is mainly based on trendlines, pattern setups, fibonacci retracements, momentum indicators, moving averages and bollinger bands, and today I'll be looking at bull and bear trends of various sizes from a trendline and pattern basis combined with fibonacci retracements.

Looking back over the lifetime of SPX, and I'd stress that in terms of the size of some of the patterns we are looking at here, that can be a very small statistical sample, I would make two rules governing bull moves on SPX of primary degree. Primary and secular bull trends differ from bull trends of lower degrees in that cumulative inflation becomes increasingly important, but too small to be a factor in the smaller degrees.
  1. All primary bull trends are trendline pattern based and come in four related forms on the linear charts. The first three are all variants on a rising channel and they are the rising wedge (channel narrowing inwards), a rising channel (parallel trendlines) and a broadening ascending wedge (channel broadening outwards). The fourth type is a support trendline only and that may just be one of the other three where insufficient data is available to identify the upper trendline. This doesn't apply to secular bull moves as these are too large to be charted on linear charts.
  2. All primary (and secular) bull trends end with a fibonacci retracement of 38.2% or more of that bull trend, after which a new bull trend will begin unless there is a fibonacci retracement of a secular bull trend to follow. Bull trends can be composed of a single trendline pattern followed by a fibonacci retracement of that move, or a combination of more than one trendline pattern, with that trend ending with a fibonacci retracement of the move covered by the combined patterns.
Otherwise all trends of any degree follow similar rules in the higher and lower degrees. Clear patterns and trendlines are visible down to the smallest scales on SPX and up to the highest. In effect from highest to lowest all these trends open up on closer examination like Matryoshka Dolls, with smaller setups following identifiably the same rules down to the limit of detection. This is obviously a similar position to that taken in Elliot Wave Principle, and that has stood up to my inspection to date. The difference in the smaller degrees is that there can be long bear trends, which aren't there in the higher degrees.

This means that I am discarding the definition of a primary bull or bear market being a move of over 20%, which is an arbitrary boundary that regularly doesn't fit the trendline and fibonacci structure of the overall moves, and I am also discarding the idea that secular bear markets look much different or last longer than primary bear markets. Given that the highs and lows in all three secular bear markets to date were made in a single bear market in three years or less that seems a reasonable working assumption.

Looking again at the SPX since 1925 on  this basis, and ignoring the 1929-32 market as I don't have the data on the prior move to assess it as a fibonacci retracement, I therefore have the following secular markets since then:

  • 1932-72 Secular Bull Market, 40 years, trough to peak 2667% rise
    • Rising Wedge into Diamond Top
  • 1972-4 Secular Bear Market, 2 years, peak to trough 50% decline
    • Part of Diamond Top, failed so continuation (overambitious downside target at 0)
    • Almost exact 50% fib retracement of 1932-72 on quarterly close basis with pinocchios
    • Multi-quarter test of 100 quarter MA
  • 1974-2007 Secular Bull Market, 33 years, trough to peak 2485% rise
    • Broadening Ascending Wedge into Double-Top
  • 2007-9 Secular Bear Market, 2 years, peak to trough 57.7% decline
    • Part of Double-Top, failed so continuation (overambitious downside target at -120)
    • Just under 61.8% fib retracement of 1974-2007
    • Multi-quarter test of 100 quarter MA with pinocchio to final low
  • 2009-204(?) Secular Bull Market in progress, trough to peak 153% so far

On the SPX quarterly (LOG) chart from 1925 that looks like this:
I would break down the primary bull and bear markets on SPX from the 1974 secular low to the 2007 secular high as follows:
  • 1974-80 Primary Bull Market, 6 years 2 months, trough to peak 132.8% rise
    • Rising Channel into Double-Top
    • (Disregarding the 20.5% decline in 1976-8 as less than 38.2% retrace of prior move)
  • 1980-2 Primary Bear Market, 1 year 9 months, peak to trough 28.5% decline
    • Declining Channel into W Bottom
    • Almost exact 50% fib retrace of 1974-80
  • 1982-7 Primary Bull Market, 5 years, trough to peak 233% rise
    • Rising Wedge broke up then Broadening Ascending Wedge into Double-Top
  • 1987 Primary Bear Market, 2 months, peak to trough 36% decline
    • No identifiable pattern between Double-Top and Double-Bottom
    • Almost exact 50% fib retrace of 1980-7
    • Exact test at 200 week MA
  • 1987-90 Primary Bull Market, 2 years 9 months, trough to peak 70% rise
    • Broadening Ascending Wedge into Double-Top
  • 1990 Primary Bear Market, 3 months, peak to trough 24.5% decline
    • Falling Wedge into IHS
    • Almost exact 50% fib retrace of 1987-90
    • Multi-week area test at 200 week MA
  • 1990-2000 Primary Bull Market, 9 years 6 months, trough to peak 428% rise
    • Rising Wedge, into Broadening Ascending Wedge, into Broadening Ascending Wedge into Double-Top (and slightly larger H&S)
    • (Disregarding the 22.4% decline in 1998 as less than 38.2% retrace of prior move)
  • 2000-2 Primary Bear Market, 2 years 4 months, peak to trough 50.5% decline
    • Declining Channel into Double-Bottom
    • Almost exact 61.8% fib retrace of 1990-2000
  • 2002-7 Primary Bull Market, 5 years, trough to peak 105% rise
    • Rising Wedge into Double-Top (and larger H&S)
I would break down the primary bull and bear markets on SPX from the 2009 secular low as follows:
  • 2009-11 Primary Bull Market, 2 years 5 months, trough to peak 105% rise
    • Rising Wedge into H&S
  • 2011 Primary Bear Market, 5 months, peak to trough 21.6% decline
    • Declining Channel into W Bottom
    • First bottom almost exact 50% fib retrace of 2009-11, second bottom pinocchio through
  • 2011-? Primary Bull Market in progress, trough to peak 57% so far
    • Rising Wedge broken up with 1965 area target
Here on the the SPX weekly chart since 2006 is my breakdown of the primary market patterns on SPX since the 2009 low. It's worth noting that the trough to peak rise so far has only been 57%, with the only primary bull market since 1974 that rose less than 100% being the 1987-90 bull market at 70%. SPX weekly chart since 2006:
To illustrate my point about similar patterns and rules being apparent at lower degrees here is a chart of the SPX 1min over four days that I posted on twitter earlier this week with a jokey poem about fleas. SPX 1min chart:
There are some notes for this analysis.

As usual my view here is a purely technical analysis view of the markets. I'm not convinced either of the value of adding economic forecasting, or my ability to do so with accuracy much better than the many highly qualified pundits and forecasters who already generally fail to manage to do this successfully, so unless there is a fundamental case that seems particular clear (RIMM/BBRY 2011 or JYPUSD 2012) I generally won't mention it.

I have only charted SPX in great detail back to about 1970, though I have checked SPX for primary rising wedges that have broken up back to 1925. I may need to modify rules slightly by the time I have fully charted SPX in detail back to 1925.

There are only two completed secular cycles on SPX going back to 1925. Statistically that is a very small sample and while those secular cycles look remarkably similar so far, the next one may not. As waiting a few centuries for a larger sample is impractical I am deriving rules from what we have so far, but would warn that an assumption that the next secular cycle will look very similar to the last two would be statistically unsafe.

I am aware that in the classic reference Technical Analysis of Stock Trends by Edwards and Magee it is stated clearly that pattern trendlines from a bull trend cannot extend back into a previous bear trend and vice-versa. I have tested this proposition carefully and consider this to be demonstrably untrue. I do this all the time and it works very well and helps clarify pattern structures. Even the greats in their fields get it wrong some of the time. 

No comments:

Post a Comment