A month ago (December 10, 2008), I started an informal technical analysis experiment:
Certain chart and volume patterns are supposed to foretell a stock's (commodity's, ETF's, etc) future price movement. Occasionally, if I come across such chart patterns I'll post about them. Then, I'll check back after a while to see how the predictions turn out.
Two stocks were selected. American Capital Agency (NASDAQ:AGNC) exhibited a double top chart pattern, which is supposed to be a a bearish signal. It closed at $19.13 per share that day. Drugstore.com (NASDAQ:DSCM) closed at $1.19 per share, and had a double bottom chart, which is supposed to be a bullish indicator.
AGNC's last close was $20.83, and it closed as high as $21.95 per share between the day it was selected and the time of writing (January 11, 2009). Shorting it until now would have been a losing trade. Nevertheless, on December 11, 2008, the stock closed at $17.93 a share (I doubt it was the result of my post, but anything is possible). It went up on December 12, and back down to $17.93 on December 15, so short sellers had some time to cover and make a profit. In this case it was 6.27% over the course of one or three trading days. It looks like, then, that the bearish signal was correct. That is, one could have made money shorting the stock on December 10.
Drugstore.com was supposed to go up, but from December 10 to December 15 it fell around 20% to $0.93 per share. Then it went up to $1.39 a share on December 17, for a 16.8% gain from its December 10 price. It traded down to a loss of over 10% by December 30, only to bounce up to $1.40 a share on January 5. So, one could have made money buying DSCM on December 10, although a much better entry was on the 17th.
It appears then, that both technical indicators were correct. But one can also say they were wrong. In the case of AGNC, one had to exit the trade early to make money. The pattern was a week long, but it was most profitable to exit after one or three days. In the case of DSCM, one had to hold on and suffer a significant paper loss, but the pattern was a longer one. Should this then be deemed inconclusive?
Success or failure is harder to determine than I expected. So here's a rule that I think is sensible. For future trials, the trade in question has to produce at least a 10% gain over a period of time not longer than the chart pattern took to develop. On these criteria, DSCM would be a successful trade while AGNC would not be.
So let's try it again, with the same chart formations. Liberty Media (LMDIA), which closed at $18.14 on 1/9/09 is displaying a double top on its chart. The pattern takes shape over about a week. We're looking for the stock to fall, as a double top is a bearish signal.
OceanFreight (NASDAQ:OCNF), which closed at $5.23 on 1/9/09 just completed a double bottom formation. The pattern's duration is 39 days. We'll be looking for OCNF to go up, as a double bottom is a bullish signal.
Let's see what happens with these stocks.
Technical analysis is a lot of things to a lot of people, so despite what I wrote to the contrary last time, it cannot be disproved. That is, it can always be claimed that I'm not doing it right, even with archetypal chart patterns. In general it is hard to disprove many things (ask Richard Dawkins). For example, you can't disprove that there is an invisible, intangible, weightless monster sitting on your monitor. But things can be proven (to a greater degree than they can be disproved, anyway), and I'm hoping the experiment will do that.
Disclosure: At the time of writing I did not own any securities mentioned above.