Basic Trading Systems

The goal of finding inefficiencies in the market is to devise trading systems which can profit from these inefficiencies. The profitability of these systems is compared to the buy and hold strategy as a benchmark. The first strategy tested by Sidney Alexander in 1961 is called a y% filter. If the price of a security moves up y%, buy and hold until it moves down y% then sell and short the security. When the price moves y% above the next low, cover the short position and buy. This strategy generates extremely small returns but can beat the buy and hold strategy when small percentage changes are used for the filter. The profitability of this system depends entirely on transaction costs and in most cases will not generate a profit due to commissions.

In 1962, Cootner proved that the buy and hold strategy is superior to a technical strategy based on the moving average. Loh examines some of the more recent trading strategies which utilize a combination of technical indicators to capture information contained in past prices. If the market is at least weak form efficient, these strategies should not be able to generate profits exceeding a buy and hold strategy. The moving average indicator compares the short term moving average with the long term moving average to give an indication if the security is undervalued or overvalued compared to previous tie periods. The stochastic oscillator is a common momentum indicator used by technical traders. When the ratio between the last closing price and the trading range is greater than the moving average of the same ratio in previous time periods, the security is emitting a buy signal. Trading occurs when either indicator emits a buy/sell signal but not when the signals conflict. Loh concludes than a combination of indicators produces higher returns than a single indicator strategy. This casts some doubt on comparisons between buy and hold and technical analysis because previous studies have only compared single indicator technical strategies to the buy and hold. Lo conclusions on the short-term dependency of stock returns show some evidence that technical trading strategies may work in the short-term, but are certain to fail in the long-term.

References:

Loh, Elaine, “A Proxy for Weak Form Efficiency Based on Confirming Indicators in Technical Analysis”, The Business Review, Cambridge, V.5:1, September 2006, 301-306.

Additional Efficiency Readings

Figlewski studied the effect of investors will different access to information. If an investor has better information than the market, wealth will be transferred from less informed investors to the better informed investors. In the long run, as wealth accumulates to better informed investors the trades of the informed investor will weigh more on the market price cause the market price to move toward a more efficient reflection of the available information. Due to the differences in volume, this process does not result in an efficient market rather, one that moves towards efficiency because the speed of wealth transfer is too slow to force the market to adapt before the next change information.

Chan and Ariff believe that the speed of price adjustment should be the measure of market efficiency. Their paper explains the use of intrinsic value changes and random trading noise to determine the variance in observed periodic returns. The regression is run for each day following the event being examined to find the day in which the estimated beta approaches 1. A beta close to 1 indicates that both intrinsic value and trading noise are explaining the observed periodic return. This method can also be used to determine how different sectors respond to information and test whether a market over or under reacts to specific types of information.

Fishman and Hagerty determine that insider trading causes market inefficiency due to the effect on information costs. Investors will obtain information if the cost of that information is less than the expected profit which the information can provide. As insider trading becomes more prevalent, the profit to each investor will diminish causing fewer investors to obtain information. While insider trading will lead to a more efficient price, the effect of investors leaving the market due to diminishing profits will lead to a less efficient price. Fishman and Hagerty ultimately conclude that a greater number of investors with less information will yield a more efficient price than fewer investors with better information.

In 1966, Victor Neiderhoffer and M.F.M. Osborne discovered that price directions are not random at the New York Stock Exchange as the random walk model would predict. Trades with changes in price direction are two to three times more likely than trades with the same price direction. This can be attributed to the accumulation of unexecuted limit orders which alternate price direction with the market orders. Fama concludes, “…the types of dependence uncovered do not imply market inefficiency.” (398) One aspect of this scenario which indicates inefficiency is the specialist’s book of unexecuted trades. This source of information rules out the possibility of strong form efficiency because specialists have access to information which the rest of the market does not.

Long-term dependency is the tendency for natural processes to display trends which seem to defy randomness due to their persistence. In 1991, Lo developed a test for long-term dependence based on the 1951 work of Harold Hearst. Hearst developed the rescaled range statistic (R/S) to measure dependency in time series data. Lo previously discovered that short-term dependency existed in stock returns and developed a method for measuring long-term dependency while taking into account the short-term. A series may be short range dependant if the series is strong-mixing. This means that the statistical dependence between two events diminishes as time increases between the events. Lo found that long-tern U.S. stock returns do not exhibit memory effects proving weak form efficiency in the long run.

The November 2001 collapse of Enron’s online trading system created an ideal situation to study information efficiency. At its height, EnronOnline accounted for 20% of the world’s electricity and natural gas commodity markets. The rest of the world used pricing information from Enron in nearly every other transaction. In 2004, Murry and Zhu conducted a test to determine if EnronOnline made the natural gas market more efficient and whether EnronOnline reduced price volatility. Murry and Zhu utilized rescaled range statistics to determine whether price data followed a random walk. Their regression model describes the natural log of the rescaled range statistic depending on the natural log of the number of observations in each subgroup. The estimated beta would provide the Hurst exponent which, if greater than 0.5 indicates a random series. The study concluded that while prices were random before and after EnronOnline, the Hurst exponent increased significantly while EnronOnline was operating. This would indicate that the natural gas market was more efficient while EnronOnline provided price data.

References:

Chan, Denis and Ariff, M., “Speed of Share Price Adjustment to Information”, Managerial Finance, V.28:8, 2002, 44-65.

Fama, Eugene F., “Efficient Capital Markets: A Review of Theory and Empirical Work”, The Journal of Finance, V.42:2, May 1970, 383-417.

Figlewski, Stephen, “Market Efficiency in a Market with Heterogeneous Information”, Journal of Political Economy, V.86:4, August 1978, 581-597.

Fishman, Michael J. and Hagerty, Kathleen M., “Insider Trading and the Efficiency of Stock Prices”, The RAND Journal of Economics, V.23:1, Spring 1992, 106-122.

Lo, Andrew W., “Long-term Memory in Stock Prices”, Econometrica, V.59:5, September 1991, 1279-1313.

Murry, Donald and Zhu, Zhen, “EnronOnline and Information Efficiency in the U.S. Natural Gas Market”, The Energy Journal, V.25:2, 2004, 57-74.

Definitions of Market Efficiency

In 1970, Eugene Fama defined market efficiency as “a market in which prices always fully reflect available information”. The concept of efficiency can be broken into three forms: weak, semi-strong, and strong. In weak form, current prices reflect past performance. Weak form efficiency implies that long term above average returns cannot be gained by studying past performance. The semi-strong form tests how quickly new information affects current prices. Strong form efficiency precludes the possibility of a small number of investors having access to information which other investors do not. In other words, insider trading cannot yield long term above average returns in a market which has strong form efficiency.

The first theory of market efficiency is explained by a random walk. The random walk is the combination of two concepts: first, sequential one period returns are independent of each other; second, one period returns have identical distributions. Early analysis of the random walk used the “fair game” assumption to explain that random bits of information caused random price changes because without information, tomorrows expected price is the same as today’s price.

Weak form efficiency may also be tested by evaluating the fair game assumption. By testing for first order serial covariance in historical prices, we can prove that future performance does not depend on past performance to the extent that historical prices can be used to gain and advantage over a simple buy and hold strategy. While the expected serial covariance is zero in an efficient market, one period returns may be related because the expected return on period t+1 depends on the return in period t. Fama’s test using Dow Jones stocks from 1957 to 1962 shows that serial covariance can only explain about .36% of one period price changes. Although this is statistically significant, it is insufficient to build a profitable trading system.

Semi-strong form efficiency is measured by finding the average regression residual of one period returns twelve months before and after the event which is thought to be the significant factor. This can be used to determine if the market is anticipating a change and how quickly information is incorporated into prices. The average residual can be substituted for the difference between the market and a specific security in order to measure its performance relative to the market. Other applications include testing the effect of exogenous variables on market performance and testing for implicit information exchange in a large stock offering by comparing the difference between periodic returns to the average periodic return.

Jensen tested strong form efficiency by measuring the degree to which a mutual fund produces a higher return than the expected return adjusted for the risk of each security contained in the portfolio. This will determine whether fund managers have access to information which the rest of the market does not. Even after the fund’s load and commissions are eliminated, Jensen found that mutual funds still perform slightly below the market line in the long run. Apparently, the investor stands to gain more from investing in the entire market than a group of selected stocks if costs are equal.

Submartingale

A submartingale is a random process where the next value is expected to be greater than or equal to the current value. This concept underlies the gambling strategy of recouping losses by doubling down on the next bet. This is an ineffective gambling strategy because a submartingale only exists when the game is fair and the odds favor the player. In reality gambling payoffs often favor the house causing the player’s payoffs to follow a supermartingale. In this situation, the expected payoff is less than or equal to the current payoff and in the long run the player loses all his money. According to weak form efficiency, expected returns follow a submartingale. The positive drift of expected returns is due to the effect of retained earnings and growth in the economy. This means that prices are expected to be greater than or equal to current prices causing expected returns to be positive.

Indonesia’s Labor Laws Deter Investment

On December 6^th, The Wall Street Journal carried an article about the effect from Indonesia’s labor laws on foreign direct investment. The law requires employers to give workers approximately 108 weeks of pay when laying off workers. Local governments are authorized to impose additional minimum pay increases. In one example, a jeans factory was forced to declare bankruptcy because downsizing was more expensive than shutting down the factory. According to the article, the main source of competition to Indonesia’s labor intensive economy is Vietnam rather than China. Vietnam’s more favorable labor laws and lower wages make an attractive substitute for Indonesian labor. This will cause the elasticity of labor to increase. Labor intensive industries such as textiles and agriculture attract mainly unskilled workers. This means that the supply of labor will be large leading to high labor elasticity. Since much of the manufacturing in Indonesia is for the garment industry, the finished goods are likely to be elastic due to the availability of substitutes from China and Latin America. The product elasticity will also cause the elasticity of labor to be more elastic.

Due to the high price of capital in Indonesia, the slope of the isocost curve will be fairly flat. The prevalence of labor intensive industries in Indonesia means that the isoquant will touch the isocost line closer to labor than to capital. The relatively elastic demand for labor means that a wage increase will yield a proportionately larger decrease in labor utilized. The firm’s marginal revenue product curve will also be relatively flat because productivity in Indonesia is lower than regional alternatives and the price of the goods manufactured in Indonesia is relatively low. The labor regulations in Indonesia effectively raise the wage rate because companies must give workers higher severance pay. Even though the cost is incurred when the output of the worker stops, the cost is directly related to labor and must be considered when making decisions on the quantity of labor to employ. If the labor laws were reformed, this would effectively lower the wage rate causing firms to employ more labor and capital. The increased capital usage will shift the MRP curve outward due to the output effect. The isocost line will also shift left increasing the amount of labor employed. The marginal cost curve will shift down increasing output.

Currently, Indonesia’s unemployment rate is 10%. Although other factors such as infrastructure, corruption, and complicated bureaucracy discourage foreign direct investment, clearly unemployment could be reduced by reforming labor laws. This would lower the cost of labor and increase productivity by allowing for more incentive pay. Increased productivity further lowers costs attracting more capital investment which in turn causes labor productivity to increase. This cycle will lower unemployment and stimulate growth creating a more stable social environment to attract business.