Sunday, May 03, 2009

A Better Market Exhaustion Tool

I remarked a handful of days ago about hitting a Demark Combo and Sequential sell signal on the S&P. Demark counts rising or falling prices in stocks that meet certain criteria. The premise of his work is based on price exhaustion. Although I would argue there is no substantiation of his work beyond anecdotal analysis. Most of that anecdotal analysis has been since quantitative trading became very popular over the last decade.

As we remarked similarly with volatility, volume is one of the least understood and analyzed tools in quantitative analysis. And, it is almost always completely discounted by chartists and other forms of technical analysis. Over the years I developed a count-based algorithm that is substantially based on market activity rather than made-up rules using counts to measure exhaustion in the market. ie, Demark's tools. Behind the visual representation below is a count-based algorithm of specific volume characteristics on the New York Stock Exchange. But the market is the determinant of counts in my work as opposed to Demark's work which uses rules that really don't reflect market reality.

On that note, certain patterns or rhythms exist within markets. These patters are very similar regardless of time frame. In other words, whether it be a monthly, weekly, daily or intraday view there are certain rhythms to the market that are common across all time frames. Below we have the same count-based volume algorithm overlaid on an intraday, daily, weekly and monthly chart of the S&P 500. I use the S&P simply because it is most representative of the overall U.S. and global economy.

As you can see, the same exhaustion patterns clearly exist in all time frames. And we clearly see exhaustion often sets in at the same counts: 10-11 with price peaks and 0 at price troughs. The data is not curve-fitted rather I have let the market determine its own cyclicality. In other words, there is nothing that would keep counts from achieving levels of 20, 40 or any other arbitrary number other than natural market rhythms. And that same natural rhythm ironically appears across all time frames.

So based on its construction, a low count reading can be interpreted two ways. One is that we have reached a price exhaustion point and are ready for a few rally (readings of zero are nearly a surety) or that the upward price movement has characteristics of temporary exhaustion. A high count reading implies temporary market exhaustion. In decades of analysis and years of analysis on intraday data, market exhaustion has never reached over 11.5 on any time frame. And, the most powerful seldom exceed 10.5 On an intraday view, exhaustion may last days. On a daily view it may last weeks. On a weekly or monthly view, it may last substantially longer.

Notice on the first intraday chart, that price was still declining into the early March low as the volume count started increasing. This was signalling a coming rally of some sort. Notice we also see a chronically low count reading over the past two weeks. I attribute this to temporary exhaustion setting in as the market has made very little progress during that time. But we shall see.

On the second chart, a daily view, notice how the count did not advance when there were so many calling a bottom again and again in September, October and November. And notice how the price count started upward before the March 2009 low confirming a possibly imminent rally. Also, notice how the price count appears to have peaked a few weeks ago. This is potential confirmation of the intraday chart where price has continued higher without the count increasing. This is lending credence to the potential for a short term price top developing. That doesn't mean the market has to go down. It just means the market may be close to price exhaustion. I would be very surprised to see significantly new price lows in the stock market any time soon given the cyclicality and rhythm of this analysis tool. In other words, as I wrote last week, we could easily have seen the market low for 2009. We shall see because intermediate term moves six to twelve months out are just about impossible for anyone to forecast.

Now on the weekly chart, look at how upward exhaustion was reached back in October of 2007 with the second highest weekly count reading in the last twenty years. This was when the bullishness of Wall Street professionals was substantially positive as we noted at the time. Wall Street was basking in its glory just when the market rhythm was telling us a very different and potentially ominous story. And, we see downward exhaustion occurred in the last weeks of 2008 and has been rising ever since. Hence, one of the rationales for my remarks into the end of 2008 and early 2009 that a rally was building under the market but would take some time to develop. This dovetails with some monetary analysis I also do, which was drawing the same conclusions and supplements quantitative analysis quite nicely.

And, finally, let's look at the monthly chart. Market exhaustion was clearly reached in 1999 well before the price peak in 2000 and just so happened to reach zero or downside exhaustion in 2002. The market rallied substantially off of late 2002 but then collapsed again into 2003 when the ultimate bottom was set. We then saw a substantial monthly count in the October 2007 market peak. The same count reading we saw in 2000. Frankly, I never thought the market could again reach the same levels as were achieved in 2000. And as I have said before, were this a normal cycle, I am confident the May 2006 peak would have been the market peak. But the data doesn't lie. I will say in hindsight that there is substantial data pointing to inflows into the U.S. from China and other emerging markets that fueled the run from the July 2006 lows to the peak in 2007. Is that really a surprise? These countries have absolutely no experience handling excess reserves or investing. So, they ended up being bag holders buying American stocks at what will likely be a multi-decade peak.

So, here's the way to interpret this data. Monthly trumps weekly which trumps daily which trumps intraday. In other words, a daily extreme reading could work itself off while a monthly count is marginally impacted just as we saw from the March low. So, as an example, when Nouriel Roubini, a bearish economist called Jim Cramer a dunce or whatever it was, for calling the market low in early March, yet he was doing so as we were clearly seeing a rising count on the weekly readings and the daily count was moving higher. In other words, Roubini relied on rhetoric rather than analysis so he spoke at exactly the wrong time. Cramer now gains confidence that he is right yet his call was also nothing more than rhetoric because again Cramer was very bullish when stocks were at their peak in 2007 when the data was pointing to imminent price exhaustion. My point is that banter is completely useless as an investing tool. Being lucky doesn't involve a lot of critical thinking and has no basis in truth. Markets don't care what Roubini, Cramer, Jesus Christ, you, me or anyone else thinks.

Realize there is no magic that states a market has to reach an oversold level of 0 or 10. It's simply a natural rhythm that seems to occur across all time frames. And, in fact, we saw counts substantially lower at short term peaks since the 2007 top. But regardless of count number, the counts often started lower before a price drop and started higher before a price rise, forewarning of potential price exhaustion points. That's sort of nice now isn't it. :)

Extreme readings occur weekly on an intraday analysis, they may occur a handful of times a year on a daily analysis but may occur every few years or even decades on weekly or monthly analysis. This obviously makes sense because of the macro factors at work in markets that are very different over the decades. It just so happened that we saw very extreme readings in 1999 and 2007. But, "just so happens" is really indicative of a far more substantial reality. That Wall Street was able to shove monthly readings to such extremes on the up side is a sign of the power of these advances. Power that I believe macro factors will mitigate for decades to come. In other words, It could be ten or twenty years before we ever see monthly counts of 10 or more.

If one is to consider we could eventually reach an exhaustion level of zero on the month data as this cycle winds down, and there are ample fundamental reasons to believe we may reach some absurdly low level, and even hit zero months before a price bottom, we will likely have more large price drops at some point in the future. But, from a weekly perspective, it would also be possible to see price advance substantially higher into the 1000-1100 region of the S&P before we take any such plunge. An upside price band I do believe we highlighted early in the year at a time no one thought such a move possible. That may or may not happen. But, the possibility does exist were we to see the market reach count extremes of 10-ish on the weekly data.

It's a mad, mad, mad world.
posted by TimingLogic at 7:57 AM