Dr. Gangadhar Darbha is Executive Director and Head of Algorithmic Trading Strategies
Algorithmic Trading is attracting increasing attention among the market players in India ever since the regulator SEBI has permitted its use on exchanges. While the institutional players and brokerages are interested in it to exploit the new technology and trading strategies, retail investors, and hence regulators, are concerned about its implications to volatility, liquidity and efficiency of the securities markets. A public policy question in this context is whether introduction of Algorithmic Trading benefits institutional players against retail investors by creating technologically driven entry barriers to the latter. To understand this policy dilemma, we need to understand the basic tenets of Algorithmic Trading.
Algorithmic Trading can be defined generically as any type of computer-assisted quantitative model based trading activity which handles the timing, submission and management of trades and orders in the securities markets without human intervention. It is synonymous with “program trading”, “auto-trading”, “black-box trading”, “high-frequency trading” etc.
Hugely popular in the US and Europe, but still in its early stages in India, Algorithmic Trading was seen both as the driver of real-time arbitrage and market efficiency and as the cause of excess volatility and market crash. The proponents of Algorithmic Trading believe that, by increasing the speed with one can identify pricing inefficiencies and execute impact-cost minimizing trading strategies, it can lead to more efficient price discovery process that is less influenced by human behavioral biases. The opponents, on the other hand, argue that in so far as Algorithmic Trading systems use “common-knowledge” market data as the sole input, they, like technical trading systems, can make the market one-sided and unstable by generating homogenous trading strategies. Notwithstanding the apprehensions expressed about its adverse side-effects on market liquidity and volatility, Algorithmic Trading has had a great success globally over the last decade if one goes by the proportion of trades executed using Algorithms both by the buy- and sell-side firms across asset classes.
In the recent past, Indian Equity market is seen as an important opportunity for Algorithmic Trading due to several factors: (i) Well established and liquid stock exchanges, (ii) availability of liquid and exchange traded derivative products such as single stock-futures and options (iii) technologically sophisticated connectivity to stock exchanges, (iv) easy availability of relevant human-capital, (v) presence of global banks and brokerage houses, (vi) supportive legal and regulatory framework, and (vii) hitherto limited competition. In evaluating the potential benefits and risks of the Algorithmic Trading for Indian markets, we have to understand the generic structure of an Algorithmic Trading system, the kinds of risks that each of those components individually and collectively poses, and the risk management systems that market participants, stock exchanges and regulators need to create.
At a very generic level, any Algorithmic Trading System can be decomposed into three components: a (Quantitative) Model that analyses the incoming market data, a Trading Strategy that submits and manages the orders based on the Model, and a Technology system that connects the model and strategy to the market place.
The Model could be a set of pre-programmed simple decision rules or highly sophisticated mathematical or statistical models based on market data and / or private information; the most important “model risk” that one faces is the possibility that most models might generate the similar “trade signals” in response to an incoming market data thereby making the market one-sided. The quality of the model, therefore, is as much related to the scientific consistency and empirical validity against historical data as it is to the range of information that it processes. As we know from the “Weak-form Efficient Markets Hypothesis”, it is generally not possible to create a sustainably profitable Algorithmic Trading models that use only the publicly available historical market data. It is important, therefore, for Stock Exchanges and Regulators not to set model validation processes in such a way that market participants tend to develop “similar” models and produce homogenous trading signals.
Design of Trading Strategy constitutes setting up of decision rules that convert signals generated by the model into a buy/sell order placement and stop-loss limits based risk management system. Any Trading Strategy that combines a trader’s (qualitative) prior views with that of a (quantitative) model will significantly enhance the capability of any Algorithmic Trading System as it uses wider set of information than a pure technical model and can generate more stable posterior trade signals. It should also be noted, as more and more market participants use Algorithmic strategies, the liquidity and volatility of the underlying market will start changing and that in turn makes certain standardized Algorithms less useful. In designing Trading Strategies, one must also take into account the “feedback” effects of Algorithmic Trading on the market microstructure.
Latency-arbitrage is at the heart of designing of Technological Architecture of Algorithmic Trading system. High-speed processing of the order placement via co-location of servers in Stock Exchanges, for example, facilitates such latency-arbitrage. Management of operational risks associated with such high-frequency trading systems is an important and well-recognized aspect of Algorithmic Trading.
Stock Exchanges will have to realize that they are creating a “Trading-Highway” with significant externalities that could potentially change the market microstructure. The systemic considerations, therefore, will have to dominate the agenda than pure commercial considerations. Regulators will have to ensure that, much like car manufacturers are not allowed to control the use of Express-Highways, Market Participants and Stock Exchanges should not skew the development of market microstructure. Regulators will also have to ensure that they don’t lag behind the skills and capacity in monitoring the systems against gaming and other fraudulent practices, and / or un-coordinated but homogenous trading strategies making markets excessively volatile, public information sensitive and illiquid in terms of extreme news.
Dr. Gangadhar Darbha is Executive Director and Head of Algorithmic Trading Strategies at an Investment Bank in Mumbai; the views expressed are personal; [email protected]