Trading Algorithms Built with Directional Changes

摘要

Algorithm trading has become more and more important to financial markets. Most existing algorithms use time series as input. Instead of relying on physical time, Directional Changes (DC) focus on the price reversion events where the reversion reaches a certain magnitude, which is referred to as the threshold. In this paper, we propose two trading algorithms based on DC - TA1 and TA2. TA1 is also based on the Average Overshoot Length scaling law (AOL). An Overshoot refers to the event of price continuing to change in the current direction before the next reversion takes place. The AOL states that on average the Overshoot length is approximately equal to the threshold of DC. We have designed two DC based trading algorithms: TA1 takes advantage of the AOL and T2 takes profit with a more conservative criteria. By testing the algorithms with five stock market indices, the results suggest that in most scenarios, the algorithms are able to generate a positive outcome. The input arguments can be changed in order to change the performance of the algorithms, so TA1 and TA2 could be tailored to trade in different markets.