If you read interview with Paul Tudor Jones in Market Wizards by Jack Schwager, you may remember that Paul predicted and handsomely profited from 1987 crash by using what he called an "analog" model, which was simply a comparison of 1987 chart pattern leading to the crash with a chart pattern of 1929. This begs a question - do analog models work? Can a trader extract consistent profits from the markets by simply data mining chart patterns?
My personal view is that consistently relying on data mining of the chart patterns will not produce any meaningful results due to spurious correlation. However, there are isolated periods in history when price patterns did look unbelievably similar. Just take a look at long-term charts of boom and bust of Nikkei (before and after 1989), Nasdaq (before and after 2000) and Gold (before and after 1981). I believe that a key to differentiating between spurious correlation and real historical similarities is the analysis of the underlying macro economic conditions. In most manias increase in asset prices were justified by rapidly improving fundamentals in the initial stages, and accelerated rise in prices in light of deteriorating fundamentals in later stages of the bubble, that inevitably lead to crash. In later stages of the bubbles prices are normally driven by the unsustainable feedback loop when fundamentals are ignored and the only justification of higher prices are the rising prices. In addition, this feedback loop is reinforced by excess media attention to rising prices that brings new money (energy) from naive investors who did not participate in the initial stages of the rise and now feel left behind. The later stages of the manias and mini-manias can be identified by interesting price patters, when markets rise in a straight - line fashion with practically no pullbacks.
Looking at the current situation, I believe we are going through the last stages (and most likely already entered a correction) of the mini-mania in US equity markets, and are at a danger of a sharp correction in the upcoming weeks. If I were to use an analog model for the current market rise, I think 2007 top provides the best similarity from the price pattern point of view. Let's examine Nasdaq 100 chart pattern of 2007 leading to a sharp correction and then multi-year decline. The last leg of the bull run that started in mid August of 2007 was more or less in a straight line fashion with no meaningful multi-day pullbacks. The rise lasted for 53 days and exceeded prior high by ~ 8.7% - see chart below.
The current picture of Nasdaq 100 price action looks somewhat similar - the duration from the low point in the beginning of February until the high point of 2,060 was 54 days and the index exceeded the prior high by ~ 8.4% - see chart below.
On the macroeconomic level I see the following similarities:
- In 2007 market was rising despite of the fact that mortgage crisis was already unfolding
- Currently, markets are rising in face of the clear signs of unfolding sovereign debt crisis and possibly severe currency crisis, that will take years to work through
- In 2007 liquidity started leaving the system as the banking sector was getting squeezed by the deteriorating mortgage loans
- Currently, liquidity is leaving the system as well as Fed already shut down all emergency lending programs and is probably months away from announcing the intention to raise Fed funds rates.
Take this unscientific, price action based, non quantified "analysis" with a grain of salt, but I think there is a decent chance of a sharp break in the equity markets in the upcoming weeks.