High Frequency Trading (HFT) is a subset of Algorithmic Trading (AT) in which timing, price and order execution are done without human intervention. HFT has grown to reach about 40 per cent of the total trades in India. It is likely to be the dominant mode for trading in the near future. In this context, there are some regulatory concerns, as HFT becomes prolific.
High frequency trading hunts for temporary pricing discrepancies in the market and trade quickly before it disappears. It monitors market fluctuations and then executes trades at the speed of light. The speed with which it does this, known as latency, is what drives the competition in this industry. This competition for speed is often compared with a technological arms race, with HFTs competing to access data microseconds faster than the rest. To facilitate such ultra-low latencies, most stock exchanges offer collocation services, which is the placement of HFT platforms very close to the exchange servers.
Constant up-gradation of technology and algorithms is crucial for HFT to survive, since most of the HFT algorithms have very low shelf life as markets are evolving very fast. They also face the risk of reverse-engineering (the controversial conviction and later acquittal of Sergey Aleynikov for allegedly stealing Goldman Sachs algorithm is a good example) and the risk of an algorithm turning out to be rogue.
With direct market access, a rogue algorithm has the potential to create havoc in the market. Yet, we do not have clear regulations defining who can develop, test, modify, and place an algorithm into production.
The debates on the benefits of HFT revolve around its impact on spreads, liquidity and market quality. From this point of view, HFT has the potential to improve the market quality if it can quickly incorporate information into prices. But, in reality, HFT has no interest in acquiring information about companies. It is not ‘investing’ in the conventional sense and, therefore seldom carries positions overnight.
Many advocates point out that competition in HFT can eliminate the price-arbitrage opportunities. But research shows that the very presence of continuous time trading creates mechanical arbitrage opportunities. For instance, an ETF and a futures contract that track the same underlying security may be perfectly correlated in the course of an hour. But, such correlations disappear in high-frequency time scales, thus providing mechanical arbitrage opportunities. Competition in HFT will not eliminate such arbitrage opportunities; instead it can only raise the latency (speed) bar for capturing them. One remedy is to move to discrete-time trading, where the trading day is divided into frequent but discrete time intervals of minuscule length.
Exchanges love HFT as it can increase trade volumes and its advocates reiterate the benefits HFT can have in improving liquidity. But, most likely HFT appears only where there is sufficient liquidity, and is not necessarily created by it.
Remedies and regulation
Exchanges will benefit from the volumes HFT can bring in. No wonder, NSE and BSE have been actively marketing their co-location infrastructure. However, they need to tread cautiously.
In 2012, SEBI instituted a fee for AT orders which have a high daily order-to-trade ratio (OTR). The objective was to check repeated management of orders without generating trading volume. In spite of this, the 2015 financial stability report (FSR) by the RBI finds that among the total cancelled orders, the share of AT orders is 90 per cent.
This is reminiscent of an ‘illegal’ HFT strategy such as layering, which creates an illusion of a significant investment happening in a stock to generate a price movement favourable to the trader. This is done by posting bids and immediately cancelling them at high frequencies, thereby building pressure on the stock price.
Along with the existing caps on OTRs, a cap or fees on excessive use of HFT methods should be considered.
To hedge any risks that can arise from the prolific growth of HFT, the exchanges should ensure that, there are: Limits on orders, positions and losses that can be made intra-day; a kill-switch (circuit breaker) to stop trading at different levels; and make orders rest on the exchange books for a discrete time, such as half-a-second.
(Anand Sasidharan is a Doctoral Student at IIM, Bangalore and Sankarshan Basu is a Professor)