MARKET-MAKING STRATEGY IN THE SYSTEM OF ALGORITHMIC HIGH-FREQUENCY TRADING

Market maker is the most important participant of modern exchange trading, it provides the increase of market liquidity and reduces the difference between bid and ask (spread). The paper presents automatic market-making strategy for quoting of options and other kinds of financial instruments on electronic markets. Quotes are based on theoretical pricing which is a resource-intensive task. Presented algorithmic optimizations, in particular quotes caching and smoothing of underlying asset price oscillation, give the possibility up to four times boost for quote modify scenario on real market data. Mechanism of quotes caching precalculates quotes in certain diapason around current underlying price. If underlying price changes within the diapason, algorithm sends already filled message for quote modification, instead of new complex computation. Smoothing of underlying asset price oscillation prevents permanent moving of the diapason and reacts only on significant market moving. A size of caching diapason which provides optimal correlation between speed of quotes modification and resource consumption has been defined experimentally (40 elements). In case of quoting 36 options on Eurex Exchange an average delay between underlying price change and quote modification is 277 usec. The measurements were carried out on the Sun X4170 M3: CPU(s): 2xXeon 2.9GHz RAM: 128 GB server under Solaris 10 operating system. Obtained results correspond to modern market-making requirements. The developed strategy is used by big European banks and trading firms.