Directional-Change Event Trading Strategy: Profit-Maximizing Learning Strategy

Many investors seek a trading strategy in order to maximize their profit. In the light of this, this paper derived a new trading strategy (DCT1) based on the Zero-Intelligence Directional Change Trading Strategy ZI-DCT0, and found that the resulting strategy outperforms the original one. We enhanced the conventional ZI-DCT0 by learning the size and direction of periodic fixed patterns from the price history for EUR/USD currency pairs. To evaluate DCT1, experiments were carried out using the bid and ask prices for EUR/USD currency pairs from the OANDA trading platform over the year 2008. We compared the resulting profits from ZI-DCT0 and DCT1. The analysis revealed interesting results and evidence that the proposed DCT1 investment strategy can indeed generate effective electronic trad- ing investment returns for investors with a high rate of return. The results of this study can be used further to develop decision support systems and autonomous trading agent strategies for the FX market. Keywords-Trading strategies; Autonomous trading agent strategies; Pattern recognition; FX Market. I. INTRODUCTION Electronic trading strategies have become a hot topic in the field of financial markets, and numerous strategies have been developed. Investors are always looking for a trading strategy that maximizes their profits. The financial literature has featured a long debate on the effectiveness of the technical analysis of financial market time series (1-8). Some argue that prices are not predictable based on historical information, since all the relevant public information is mirrored in the prices. In contrast, recent studies (8,9) have uncovered empir- ical evidence of various price anomalies, and therefore have confirmed positive observed evidence on the effectiveness of technical analysis for analyzing financial price time series. Trend Following (TF) trading strategy is a widely used investment strategy due to the simplicity of the principle on which it is based and its effectiveness (10-13). TF adopts a rule-based investment strategy based on the directions of market price trends, where a trader takes advantages of the price trend on the assumption that the current price trend will continue in the same direction. Furthermore, the underlying assumption of TF is that a trader will follow the price trend with the assumption that some traders have market information prior to the general public which is reflected in the direction of the price trend (10). A TF trader places a buy order when the price is rising, while a sell order is placed when the price is falling. The financial literature reveals successful investments based on a TF trading strategy in stock markets (11), currency markets (14) and commodity futures' markets (13). Similar to the TF investment strategy is the Contrary Trading (CT) strategy with regard to the direction of the market price trend. A CT trading rule places a buy order in anticipation that the price will move in the opposite direction. For example, a trading rule may indicate a buy order opportunity when the price falls by 0.03% and afterwards places a sell order if the price rises by 0.06%. Despite the effectiveness of TF and CT investment strate- gies, comparatively few works have explored the application of learning in order to enhance TF and CF investment strategies. Aloud et al. (15) have constructed a trading strategy called ZI-DCT0 based on pooling two approaches: (i) the DC event approach (16) and (ii) TF and CT investment approaches. Trading in the financial markets is highly active at some times, but calm down at others which makes the flow of physical time discontinuous. For that reason using fixed time scales for studying the price changes in the market runs the risk of missing important price activities. The DC event approach captures periodic activities in the price time series to detect major periodic patterns based on the trader's expectations of the market. Given a fixed threshold size, the DC approach characterizes periodic price trend movements in the price time series, where any occurrence of a DC event represents a new intrinsic time unit, independent of the notion of physical time change. A comparable trading strategy to the ZI-DCT0 is introduced by Alfi et al. (17) in which a trader places an order if the price fluctuations exceed a defined threshold. The threshold is determined by the trader, and remains constant during the traders' trading period in the market. The main difference between ZI-DCT0 and the one introduced in (17), is that ZI-DCT0 considers the direction and the overshoot of price movements in the traders' trading activities. In the light of this, the work reported in this paper introduces a new trading strategy called Directional-Change Trading (DCT1) derived from ZI-DCT0, where ZI-DCT0 has been enhanced by the incorporation of a learning model fed by historical dataset, with the aim of determining the size and direction of periodic patterns in the price time series. As such, the DCT1 is able to recognize periodic patterns in a price time series such as DC events. This may be the key to providing effective decision support for traders in the financial markets. DCT1 applies a simple learning mechanism which avoids the complexity of artificial intelligent trading strategies and also the vagueness of zero-intelligence and Buy-and-Hold trading strategies in which traders trade randomly, subject to budget constraints.