Characteristics and Forecast of High-frequency Trading

Abstract

High-frequency trading (HFT), which is a type of algorithmic trading, accounts for a significant percentage of trading volume in equity markets. Co-location, which is a low-latency service, enables high-speed transactions. For example, 70% of all orders traded on the Tokyo Stock Exchange use co-location. Because many HFT companies use co-location services, analyzing the characteristics of HFT can assist in understanding the market and identify improved trading strategies. Our study, which uses data from the Tokyo Stock Exchange (ranked as the third largest stock exchange in the world in terms of market capitalization) clarifies the following: 1) Most orders are filled or canceled without changing the price or the quantity. 2) If price rises due to an execution, the sell order significantly increases compared to the buy order, and the price gradually decreases. 3) After a price change, the price before the change returns within 300 s. 4) Based on the above findings, we analyze a simple trading strategy that can realize a profit with 90% reliability.