A Tech In The Fed
by John Sweeney
Hidden away in the Federal Reserve System are entire departments of economists researching nearly every aspect of US business. In late 2000, we picked up hints of an economist who had actually taken on the definition and measurement of the concepts of technical analysis. Some sleuthing brought up Carol Osler's name.
Osler is a senior economist in the Capital Markets Research Function of the Federal Reserve Bank of New York. Her major interest is in explaining short-run exchange rate movements. In particular, she has focused on technical trading strategies, largely because they are used so widely in the market even though they fail to fit standard academic models. She has also researched banking, corporate finance, equity prices, and the influence of speculative bubbles on real investment.
Osler received her bachelor's degree from Swarthmore College and her doctorate from Princeton University. Prior to joining the New York Fed, she taught at the Amos Tuck School of Business at Dartmouth College and the Kellogg School of Management at Northwestern University. After coming to New York, she also taught in the Columbia University Economics Department for a number of years.
STOCKS & COMMODITIES Interim Editor John Sweeney caught up with Osler on March 19, 2001, to explore her work in quantifying some of technical trading's hoariest maxims: support and resistance levels. And there was more.
ILLUSTRATION BY CARL GREEN
First of all, I don't know much about you, since Federal Reserve people aren't usually prominent publicly. Can you give us a description of your work?
Sure. My research primarily focuses on the exchange rate movements on a daily or intraday basis, and I'm interested in whether technical analysis works and, if it does, why.
How did you get into the technical analysis side of this?
Honestly, I backed into it. I initially thought that technical analysis was probably just craziness in the market. And I wanted to make the point that such craziness was there and that it mattered for prices. But when I tested the stuff, I found it was not crazy. It worked. So I'm now making a career of showing that it actually works and why.
I remember you started off with a study of head-and-shoulders patterns. Can you describe that study?
I looked at whether head-and-shoulders patterns are useful for predicting exchange rate movements, using daily exchange rates over about 20 years. The study, "Methodical Madness: The Head-And-Shoulders Pattern In Foreign Exchange," was written jointly with Kevin Chang. We found that head-and-shoulders patterns were pretty useful for forecasting exchange rate movements. But when we looked at it more closely, we found that you could actually do a lot better using much simpler technical tools, such as moving averages or filter rules. The head-and-shoulders strategy overlapped strongly with the others in terms of the positions they recommended, but the simpler technical rules were more frequently right. So we argued that the head-and-shoulders strategy, while potentially useful if there was nothing else available, was what we refer to as dominated by the much simpler rules.
In the abstract of another paper, you reported that head-and-shoulders trading is generally unprofitable.
Yes. In that paper, titled "Identifying Noise Traders: The Head-And-Shoulders Pattern in US Equities," I looked at the stock market and came to a very different conclusion about the usefulness of the head-and-shoulders pattern. I looked at daily stock prices for 100 different firms, which I chose at random. First of all, I asked: Are head-and-shoulders patterns profitable in the stock market? I found that on average, they absolutely are not.
So then I wondered whether there was active trading on this pattern that does not seem to be profitable in the equities markets. I was able to show statistically that there's a lot of extra trading volume right around the time my analysis says people should be taking positions. And the extra trading volume peaks on the exact day I think people should be taking positions, and then it tapers off over the next few days, just as I would have expected. So it sure looks like people are trading on head-and-shoulders patterns, but on average they are not getting much out of it in equities markets.
What do you suppose is the difference in the markets here? Stocks versus foreign exchange?
That's a really good question. A lot of people in the market think it might be a self-fulfilling prophecy; the more technical trading there is, the more it makes profitable patterns, so the more that people trade on it. It's my impression that there is more technical trading in the foreign exchange market, so it could be self-fulfilling. Somehow, I think there has to be more going on.
The foreign exchange market does tend to be dominated by institutional traders - a smaller number, say, 3,000?5,000 of them, versus the equity markets, which might be affected by a larger number of individuals. Could that be important?
It's hard to know why it would matter whether it was the trading of institutions versus the trading of individuals. I have identified one factor that differs between stock and FX markets that might help explain why technical analysis works better in FX - stop-loss orders are used very heavily in the FX market, but fairly infrequently in stock markets. This could matter because, as I show in another paper, stop-loss orders have a lot to do with the success of some standard technical strategies in the FX market.
...Continued in the June 2001 issue of Technical Analysis of STOCKS & COMMODITIES
The views expressed herein are not necessarily those of the Federal Reserve System or the Federal Reserve Bank of New York.
Excerpted from an article originally published in the June 2001 issue of Technical Analysis of STOCKS & COMMODITIES magazine. All rights reserved. © Copyright 2001, Technical Analysis, Inc.
Return to June 2001 Contents