The validity of technical analysis is still an unresolved issue with adherents and opponents on both sides. I may normally be expected to believe in the efficiency hypothesis given the fact that I am a U.S trained finance professor and academic world of finance puts its considerable weight in favor of this hypothesis with only a few exceptions. There is really no room for the technical analysis within the strict context of the efficiency hypothesis. All the information that technician or fundamentalist can hope to utilize for guessing the future course of stock prices is already reflected in the current price. The deviation of tomorrow’s price from today’s price is governed by pure chance. Stock prices and returns obey a normal distribution just like an unbiased coin thrown for heads and tails. The claims of technical analysis is then no more valid then the claims of astrology. This is what efficiency hypothesis says.
Indeed I was one of the sympathizers of the efficiency hypothesis initially. Science, however, is based on facts and observations. Rather than taking sides on the basis of what others say, I decided to investigate this issue myself. I started testing numerous technical indicators using actual stock data (stochastic, MACD, moving averages, etc) to see if their performance were indeed better than a simple “buy and hold” strategy. The results were mixed. The aforementioned indicators seemed to perform well for a certain period of time, but then turning into a futile exercise with no gains but a lot of commission costs. A common feature in most of them was the fact that they were simply trend following. Consequently, they were a bit late in signaling the buy and sell signals. This is an important shortcoming since the best daily gains are usually realized in the very early phases of an uptrend.
Then, I got acquainted with Japanese candlesticks through the book of Steve Nison, which is kind of a bible in this subject. At the first glance, candlesticks are a very unlikely candidate for an excellent technical indicator. After all we are talking about a method used by 18.century rice traders in Japan. Patterns only use opening, closing, high and low prices as inputs. No math, no formulas.
Despite this simplicity, their timing at the turning points of stock prices was much better than other trend following indicators. Candlestick patterns generate buy and sell signals just at the point when the trend is changing direction. Somehow they succeed to feel the pulse of the market better than the other indicators. One only needs to browse through the Chapter 7 of a book called Japanese Charting Explained written by Gregory L.Morris to find quantitative evidence of candlestick superiority over other technical indicators. Morris tests the performance of 68 different candlestick patterns on 2277 stocks from NYSE, AMEX and NASDAQ. He uses two different testing periods. The first period is the particularly volatile period from April 30, 2002 to December 31, 2004 with 675 market trading days. The second period is a much longer one spanning from November 29, 1991 to December 31, 2004 with 3300 market trading days. After showing the performance of each candlestick pattern on the basis of percentage of wins and average per cent gain, he then compares the candlestick performance with 14 widely utilized technical indicators (MACD, ROC, CCI, Stochastic, etc.). The results are an eye opener. Candlesticks do definitely perform better than all other indicators in the short term and assess the direction of prices better than others within a horizon of 5 days.
So far, so good, however there is a major problem. How is a small investor expected to tackle with all the intricacies of recognizing and identifying the patterns correctly? This is not a superficial problem. There are hundreds of different patterns to memorize. There are also some subtle questions such as if a very thin body (a small difference between opening and closing prices) is compatible with a Doji formation (a cross-like formation) or what we really mean when we talk about a long body?
The solution came from an unexpected source, the internet. Indonesian Bulls (www.indonesianbulls.com), which is just a branch of a group of sites dealing with the stock exchanges for different countries, generates buy, sell, buy if, and sell if signals on daily and weekly basis for all the stocks in the Jakarta stock market also providing a commentary. The site works with an automated system based on digital recognition of Japanese candlestick patterns. The site generates a cautionary buy if (or sell if) signal based on patterns before giving a definite buy and sell signal and defines a certain market based confirmation criteria to transform a buy if (or sell if) to a definite buy (or sell) signal thus showing due respect for the market dynamics of that day. The site also reports a two year history for all the stocks which transparently shows the results of trading, assuming that the trades are done by a small investor whose decisions are entirely based on site’s definite buy and sell signals. The reported results also allow a -0.2% slippage per trade to cover the commission costs and the possibility of missing the suggested buy or sell price, which is a reasonable assumption.
The site also has a stock ranking system ranging from five stars to one star. Apparently the stars reflect the conformity of the stock in question to the algorithm used to generate the signals. For example, a five star stock called Charoen Pi Ord (CPIN.JK) on December 12, 2010 was traded at 1880 IDR. This stock was traded at a price of 97 IDR two years ago, on January 13, 2009. The price of the stock increased almost 19 times since then. It was evidently a golden opportunity for a “buy and hold” strategy. However, even this spectacular performance looks quite modest, when it is compared with an investor’s performance that is assumed to trade under the guidance of the aforementioned site. A small investor following the site’s buy and sell signals would be able to increase an initial investment of 100 IDR to 7288 IDR in just two years. It means that 1 IDR invested in this stock is now worth 72.88 IDR and it shows an almost four times better performance than a simple “buy and hold” strategy, a theoretically impossible case according to the efficiency hypothesis.
Now be ready for the good news. The site operates in 34 countries including U.S.A, most of Europe, China, Japan, India, Australia, Russia, South America and others, apparently based on the same buy and signal system and with similar success stories. The samples below from different countries clearly show how signals generated by candlestick patterns can beat “buy and hold” strategy all over the world (by January 01, 2011):
Symbol Stock Name Country Buy & Hold (2yr.) Candlesticks (2yr.)
GGAL.BA GP FIN GALICIA Argentina 7.60 times 16.00 times
MEO.AX MEO AUSTRALIA Australia 2.23 times 49.87 times
KBCA.BR KBC ANCORA Belgium 0.82 times 13.68 times
TEKA4.SA TEKA PR Brazil 3.76 times 17.88 times
600111.SS RARE-EARTH A China 7.76 times 30.49 times
AVL AVALON VENTURES LTD Canada (TSX) 19.88 times 12.33 times
NOT NORONT RESOURCES LTD Canada (CDNX) 1.96 times 107.34 times
GES.CO GREENTECH ENERGY Denmark 0.85 times 10.45 times
RNO.PA RENAULT France 2.50 times 13.45 times
IFX.DE INFINEON TECHNOL N Germany 7.46 times 49.40 times
8200.HK SAU SAN TONG Hong Kong 2.32 times 125.85 times
RAJA.JK RUKUN RAHARJA Indonesia 5.37 times 43.69 times
F.MI FIAT Italy 2.84 times 10.35 times
8515 AIFUL Japan 0.41 times 46.38 times
003190.KS RNL BIO Korea 3.10 times 34.04 times
AIG.MX AMER INTL GROUP Mexico 1.71 times 10.60 times
ALMAL Al-Mal Investment Co KSCC Middle East 2.70 times 28.30 times
TOM2.AS TOMTOM N.V. Netherlands 2.39 times 30.53 times
GOGL.OL GOLDEN OCEAN Norway 2.01 times 39.05 times
MTX METOREX LIMITED South Africa 2.56 times 23.54 times
MTS.MC MITTAL STEEL Spain 1.77 times 10.16 times
EG0.SI JES INTERNATIONAL HOLD Singapore 2.91 times 33.32 times
PETRO.ST PETROGRAND Sweden 0.41 times 26.70 times
XTAN.SW XSTRATA Switzerland 3.43 times 13.39 times
2358.TW MAG TECH Taiwan 3.92 times 33.72 times
MAKTK Makina Takim Turkey 20.41 times 139.04 times
BKIR Bank of Ireland U.K. 1.42 times 168.11 times
APP AMERICAN APPAREL INC U.S. (AMEX) 0.53 times 51.12 times
MGM MGM MIRAGE INC U.S. (NYSE) 1.28 times 32.57 times
PWER POWER ONE INC U.S. (NASDAQ) 8.16 times 108.16 times
PCFG PACIFIC GOLD CP U.S. (OTCBB) 4.13 times 538.69 times
So what is the main point that needs to be highlighted? First, the efficiency hypothesis despite its theoretical elegance is incompatible with empirical facts. A simple “buy and hold” strategy can easily be beaten by a good technical system even after allowing for commissions. Second, there are friendly sites among the internet which can help the small investor to achieve much better results than a “buy and hold” strategy. So, there is no reason to be confined to the dull world of the efficiency hypothesis whose death warrant about technical analysis does not receive any support from the real world of the markets.