¿Puede andrew lo, legitimar el análisis técnico?.

Can Andrew Lo Legitimize Technical Analysis?
Robert Huebscher January 27, 2009 When it comes to technical analysis, I am not just skeptical. I flat-out don’t believe in it. Technical analysis rests on the conviction that observing patterns in historical data can produce reliable indicators of when to buy and sell securities — that the past can predict the future. It claims you can profit byidentifying patterns such as head-and shoulders or double bottoms, or by utilizing Kondratiev waves or Fibonacci series. The rational scientist in me says this in nothing more than alchemy. So I was stunned to learn that Andrew Lo, the Harris & Harris Group Professor of Finance at MIT, has co-authored a book on the subject. In addition to directing MIT’s Laboratory for Financial Engineering, Lo is thefounder of the hedge fund AlphaSimplex Group, and he is recognized as an authority on risk management, quantitative analysis and investor behavior. Andrew Lo’s endorsement of technical analysis would be akin to a respected astrophysicist endorsing astrology. Jasmina Hasanhodzic, Lo’s co-author, presented their latest research at a meeting of the Boston chapter of the Quantitative Work Alliancefor Applied Finance, Education and Wisdom (QWAFAFEW) on January 20. I decided to go with an open mind, to see whether their research could impart a degree of legitimacy to what I consider a pseudoscience. I was underwhelmed, to put it mildly. Hasanhodzic began by recounting the background research for their book, which included interviews with 14 leading practitioners of technical analysis and anextensive study of its history. Technical analysis goes back to the Babylonians, who used it to study the prices of a half dozen commodities. Their research shows uses of technical analysis in the middle ages, Renaissance, and 19th century, before it was adopted on a more widespread basis for the study of financial markets. The evolution of technical analysis may be a legitimate topic of historicalinquiry, but so were the Salem witch trials. What the QWAFAFEW crowd really wanted

© Copyright 2009, Advisor Perspectives, Inc. All rights reserved.

to know, as did I, is whether this stuff works. More to the point, is there any evidence than an investor can make money using technical analysis? To answer this question, Hasanhodzic and Lo took the first steps toward automating technicalanalysis. They studied the prices of 25 randomly chosen NASD securities over a four-year period. Using pattern recognition technology, they identified when certain signals occurred in the data. These signals — a head-and-shoulders pattern, for instance — were those that were identified in the interviews they conducted, and, in their model, indicated when a security should be bought or sold. Theycompared the returns their statistical buy-and-sell cues produced to those of a “buy-and-hold” portfolio — owning all 25 securities over the four year period. They found that there was a statistically significant difference in the distribution of the returns generated through technical analysis compared to the returns of the buy-and-hold portfolio. The data show, according to Hasanhodzic and Lo, thattechnical analysis produces what the quantitative community calls “information value,” which is a prerequisite for creating a trading strategy from which profits can be derived. The next step in their research is to develop such a strategy. But a point raised by a QWAFAFEW audience member suggests Lo and Hasanhodzic may be getting ahead of themselves. The comparison of returns from technicalanalysis to the buy-and-hold strategy is not an apples-to-apples comparison. The buy-and-hold returns were generated over the entire four-year period. But the technical analysis returns were generated only after a signal was recognized – a period less than the entire four years. A meaningful comparison requires that returns be compared over identical time periods. And even if technical analysis can be...