- Some forecasters are using Dow Theory, a popular market-timing indicator, as justification for a bearish outlook today.
- While the theory is corroborated by academic study, there are excellent reasons not to follow it.
- Dow indexes are poorly constructed and non-representative of the broader stock market. Additionally, Dow Theory is a pattern without a logical or fundamental justification—a potentially very dangerous thing for investors.
A popular market theory is flashing warning signals of a bear attack, but we don't think it's time to pack up and flee the bull market picnic just yet.
Dow Theory is a long-held and popular market-timing indicator, but likely it's most famous appearance was in August 1998's Journal of Finance. The published study featured what was at the time sophisticated artificial intelligence software used for pattern detection. Eventually, the software found Dow Theory to be a viable market pattern. And why not? Its long-term track record seems pretty good. From 1930-1997, Dow Theory beat buy-and-hold strategies by an annual average of 4.4%.
To satisfy Dow Theory's bearish requirements, both the Dow Jones Industrial Average (DJIA) and the Dow Jones Transportation Average (DJTA) must undergo a "significant correction" from new highs. In the subsequent rally, at least one of the indexes must fail to beat its pre-correction highs, and from there both averages must drop to pre-correction lows. When this happens, a long-term downtrend for the broader stock market usually arrives.
Many believe that's precisely what we're seeing this year. Why then, shouldn't we listen to it? After all, Dow Theory seems to have a solid long-term track record and is corroborated by the Ivory Tower of the Academy.
For one thing, both the DJIA and DJTA are price-weighted—an inferior method of indexing. Market capitalization weighted indexes (like the S&P 500, for example) are much better. Moreover, Dow indexes are antiquated, poor representatives of the broad stock market and economy. The DJIA was invented by Charles Dow back in 1896, and is a price-weighted average of 30 stocks including the likes of General Electric, Disney, Exxon and Microsoft. Sure, those names are well-known, but attempting to represent the broader market with a mere 30 names is a bad idea. A big move from one representative company could skew the whole average (which frequently happens).
The DJTA was started way back in 1884, and is currently a blend of 20 transportation stocks in the US including airlines, railways, trucking, and delivery companies. This is an even poorer representative the US market! Maybe a hundred years ago the DJTA made sense—when the economy was dominated by industrial goods and railroads. But today, the US economy is highly diversified and largely service-based. So studying a transportation index with the intent of gauging the entire market simply doesn't make sense.
Additionally, recall that stocks are not serially correlated. Said a different way: Past performance tells you nothing about future performance. This has been proven time and again and is consistent with financial theory. Therefore, any sort of backward-looking data on its own is meaningless as a predictor of the future. Be it valuations like P/E ratios, charts, whatever—they ultimately don't predict a thing.
There are plenty of times Dow Theory has been wrong. For instance, had you followed it in late 2002 to early 2003 where the markets double-bottomed and then headed upward to begin the current bull market, well, you would've missed out on +30% returns in 2003 alone.
A lot of market theories out there appear valid at first blush but really aren't. Human brains are designed to find patterns, and when patterns are found we have a tendency to corroborate and assign meaning to them. But patterns by themselves don't necessarily imply meaning or validity. If the underlying logic doesn't jibe, then it's just as likely your theory is a random relationship.
For instance, maybe at noon each day you have lunch, and somewhere, someone you don't know and never heard of does the same thing. These two events on paper seem highly correlated. When one eats lunch at noon, the other tends to also! Yet, neither actually predicts the other—the two events are completely unrelated. If you decided to eat at 11 AM tomorrow, it would not predict your doppelganger would also eat at 11 AM.
Believing in highly correlated events that lack logical corroboration can lead to big investing trouble, causing investors to make decisions based on illegitimate relationships. Those tend to break down as often as they work. That makes basing your stock market outlook on the antiquated Dow just plain dopey.