My own opinion is that, regardless of the usefulness of previous cycle-bottom charts, articles such as this one are full of problems. The media and blogs this year are full of articles similar styles in style and content, so I do not mean to unreasonably this one in particular; it is merely the latest one to annoy me. That said, let's look at a few of the problems contained therein.
- Comparing dissimilar things - This is the old cliche about comparing apples and oranges. Notice how the author compares two different indices. S&P 500 and the Dow Jones Industrial Average. The first is a select group of 500 companies across a diverse set of business sectors, with a bias toward stability, large size and national scope (US). The second is a basket of 30 large industrial companies. The companies are not chosen according to a rigid algorithm, except that they are supposed to serve as good indicators of the financial health of the sector(s) from which they have been selected. Not only that, the companies in these and other indices change over time. There are few holdovers in these indices over the 70 year interval of the comparison in the article. Further, the industries represented are quite different (e.g. Microsoft).
- Time frames - The Great Depression began after the market crash of 1929, not long after the market peaked, and continued for well over a decade. In the present recession, the market peaked in 2007 (3 years ago this summer) and the economy, while still quite fragile, is showing strong signs of recovery. Of course one could argue that this recession will continue for many years to come, making the comparison more meaningful, but even a casual glance around the world shows that the two situations are not at all alike. Even if our recovery stalls for a while, it takes a lot of belief to say that the market charts can be so glibly laid atop each other.
- Selection bias - For a true believer in this sort of comparison exercise, there is a need for much data mining and chart-fu to come up with the perfect comparison to exemplify the original premise. That is, the search only ends when kinda, sorta, more-or-less similarly shaped charts are found; the multitude of charts which do not fit the premise are rejected and never presented. The author then has to engage in a lengthy and bewildering rationalization to justify why that one chart comparison is the correct choice.
- Technical vs Fundamental analysis - A chart comparison is pure technical analysis, where the chart alone is used as predictive indicator, and to therefore guide buy and sell decisions and money management. Except that in the types of articles we're looking at today, the chart itself is not considered sufficient by the author. They must delve into a range of fundamental economic indicators far outside the realm of the charts being compared to select those that buttress the premise. This is not only selection bias, but an admission that the author does not believe that the chart has any predictive power on its own merits, whatever those may be. No, they must then launch into a similarly disjointed set of fundamental analysis arguments. Presumably, they also believe that fundamentals alone are insufficient to prove the premise. In general, following multiple lines of evidence to prove a premise is appropriate and laudable, except that in this case the fundamentals of the 1930s and the 2000s are so different -- political, industrial, global connections, and industry mix among other factors -- that reading the convoluted arguments in these articles can twist your brain into a knot.
- Statistical significance - Said simply, the number of market peaks and valleys for which we have good data and where the economic landscapes are at all comparable is small. The population is so small that there is little confidence in the expected outcome of any one event sampled from that population, including the present recession. You might very well do better by flipping a coin.
No comments:
Post a Comment