Brian Ross of FIX Flyer talks to Buy- and Sell-side presenting the latest lessons on high frequency trading and algorithms from the Indian market.
India’s capital markets are experiencing increased interest from local and global firms and new rules are set to attract high frequency trading (HFT).
The capital markets regulator, the Securities and Exchange Board of India (SEBI), the exchanges, brokers and many investors are in favor of abolishing Securities Transaction Tax (STT). Eliminating STT will have a positive impact on market turnover, will help high frequency traders to be more profitable and, at the same time, narrow spreads should drive up trading volumes.
STT has been levied for all trades, domestic or foreign, on all transactions in either equities or derivatives markets since 2004. At the time, the purpose was to generate tax revenue and to protect market integrity by slowing down the pace of technological advancements of a few, well-funded players. Revenue generated by STT amounted to around USD 1.5bn in 2011.
It is widely expected that STT will be eliminated this spring, bringing new opportunities for HFT in one of the world’s biggest and fastest growingcapital markets.
To better understand the situation, we asked five panelists who are leading the charge in HFT in India, to share their insights with us.
You never forget your first algo. When you first got involved in algorithmic trading, what problem were you trying to solve? What was your decision process, and what technologies did you use?
Sanjay Rawal, Open Futures: We started off using algos for trading purposes and the first one we built was for a specific type of arbitrage that was getting difficult to run using manual input. We used third party software for the exchange connectivity and wrote our algo in C#.
Vishal Rana, IIFL Capital: My first experience with HFT was trying to create a straight-arb model on a real-time basis. Although it was a simple model, the most difficult thing was to clean the data. We got the data dumps and it took a lot of effort to clean it. Most of the coding was done using C++.
Rohit Dhundele, Edelweiss: At the onset of the project, the easiest yet most important task was gathering the business intelligence to be subsequently converted to algorithms. Some of the more intricate decisions were the selection of order, execution and risk management systems to ensure a stable back-bone to the platform. Other equally important criteria were a flexible programming environment and a friendly interface for users. To achieve these objectives, we had to decide whether to build or buy this technology.
At Edelweiss, we realized relatively quickly that there is a sweet spot between the two extremes of in-house vs. outsourced solutions. We have since been following this model – combining the best of both worlds, which has helped us deliver customized solutions within acceptable turnaround times, whilst still protecting our IP.
Sanjay Awasthi, Eastspring Investments (Singapore) Limited: In the Indian markets, propelled as they are by rapid information dissemination systems, anonymity becomes a key factor in determining efficient trading. It was this need for anonymity that propelled us towards algorithmic trading. Continued use and familiarity lead to further benefits by way of better execution control. Algorithmic trading has thus become an important part of our execution arsenal.
Chetan Pandya, Kotak Securities:
The first algo I worked upon and put in production was calendar rolls for derivatives. Our trading desk had huge positions to roll from the current month to the next and manual execution was leading to slippages and erroneous executions at times. Using the 2 legged order of NSE we created a simple algorithm which would roll the position at desired spread.
My first observation regarding algorithmic trading was to appreciate the difference between an individual trading manually versus a machine trading automatically. There are so many things that come naturally to a human being but needs to be told to the machine. Sometimes I wonder whether an algorithm can fully replace a human being ever. There are those nuances of the market and events that lead to erratic market behaviour that cannot be fully programmed for reaction.
Also, I had to ensure that there is no room for error when you are trading using an algo platform, primarily because of the sheer number of orders that it can process in a single second and also the inability to spot something going awry with the naked eye given the sheer speed. Hence, I had to also think of risk management capabilities of the Algorithmic platform while needing to ensure that risk management does not lead to inefficient execution due to latency.
In terms of technology, we were limited to applications that conformed to our market regulations. Once we had the base framework and architecture ready, we integrated it rapidly with our existing applications for order routing and downstream workflows.
