On May 5, the Indian benchmark Sensex and Nifty tanked, losing over 2 percent to wipe out entire gains made last year. The main reason: algorithmic trading.
Such mishaps have happened before, despite SEBI tightening its rules for algo-trades, as traders call them. This form of high-speed trading rose 12 percent on the Bombay Stock Exchange, to account for almost 30 percent of total trades. Its share is higher in the National Stock Exchange, with nearly 46 percent of trades happening on the platform.
What exactly is algo trade? This is a form of high speed trading — that Michael Lewis chronicled in his bestseller 'Flash Boys' — in which computers execute pre-programmed trade orders. It happens at tremendous speed, way quicker than what human traders are capable of, and is much more efficient. However, that can also result in large trades that can swing the market and result in significant losses for ordinary investors, like what happened on May 6. Most investors don't stand a chance of trading in stocks that are being traded algorithmically.
All major brokers now have options for algo trading. This has also spawned start-ups that hire engineers to offer similar trading facilities as the bigger guys. HuffPost spoke with with Kunal Nandwani, CEO & Founder, uTrade Solutions, a financial trading technology company that is engaged in algorithmic trading, about high speed trading, pitfalls in this kind of trading, and how small traders got priced out. Nandwani, who previously worked with Lehman Brothers, Nomura and BNP Paribas in London, raised $800,000 in seed funding for the firm that now employs 40 people and has 15 financial institutional clients.
1. What is algorithmic trading? How is it different from human traders executing orders?
Simply put, Algorithmic trading is a system of trading which facilitates transaction decision making in the financial markets using any simple or complex logics, based on user certain inputs. Guidelines of the Securities Exchange Board of India (SEBI) define algorithmic trading, popularly called ‘algo trading’ in trading parlance, as any order that is generated using automated execution logic in which the computer executes the pre-programmed trading instructions, accounting for a variety of variables such as timing, price and volume. It offers the benefits of speedier and more efficient order execution when compared to human traders executing orders, also mitigating execution cost as well as the scope for any manual errors and emotional trading decisions.
2. On May 6, four large orders in Nifty May futures, ranging between 7,000 and 39,000 contracts, were traded between 9:41am and 9:53am. The deals were worth Rs 1,470 crore and the stock market crashed wiping out all gains made the year before. All of these were algorithmic. Can you explain what might have happened?
As I understand, a large part of above trading volume can be adduced to foreign institutional investors (FIIs) reducing their exposure to Indian markets as a part of their plan to re-allocate funds in the subsequent week to the Chinese markets for IPOs of around 20 companies amounting to $483.19 billion, intensifying outflows from countries such as India. On the back of weak corporate earnings and uncertainty over certain taxation issues, FIIs chose to re-allocate their funds which is not unusual.
The execution of their sell orders were indeed through algorithms (for example, Time Weighted Average Price algorithms which slice the larger parent order into smaller child orders executed every few minutes in the market) causing the Sensex to tumble by 723 points. What has to be underscored is the point that this fall wasn’t the consequence of any dysfunctional algorithm and hence not an aberration. Had the trades been executed manually, they would have still had the same impact albeit the time taken would have been much longer to sell the positions. Algorithms did what they were designed to do, which was to ensure efficient sell order execution.
3. Is it the same as high-frequency trading involving 'dark pools'? This subject was covered by Michael Lewis covered in his book 'Flash Boys'.
High Frequency Trading (HFT) are algorithmic trades which entail very fast execution, typically more than 100 orders per second on average, with an objective of making small profits on minor market movements by buying and selling small or large volumes of security contracts.
Dark Pools are a means of off-exchange trading which facilitate block trading for institutional investors and are designed not to impact the market price with their large orders, leading to them realizing a more favorable price. Trade execution details are released after a time lag.
For example, If London Stock Exchange has best bid offer price on a stock as GBP 100-101, the dark pool like CHIX of Bats will cross the same stock at GBP 100.5 but never show what orders are available in the market. This way fund manager can place large orders in a dark pool, knowing that nobody will know about them, and they will get execution at mid of bid and ask.
4. So how is algo-trading different?
HFT is a subset of Algo trading. The difference largely as mentioned above is that when HFT is done involving dark pools, it constitutes an off-exchange trade wherein order book data is not displayed to market participants. What’s common is only the usage of cutting edge technology, used to make trading decisions.
"This costs between Rs. 7.5-15.0 lakh per annum, which most of the small brokers find unaffordable."
5. Right now this form of trading is more expensive or so is the perception. What do you think?
Yes, the perception is right. You have to pay more for co-location. Through this, brokers are allowed to place their servers within close proximity to an exchange’s trading engine, giving them faster access to price feed streamed from the exchange. This costs between Rs. 7.5-15.0 lakh per annum, which most of the small brokers find unaffordable. The benefit derived is low-latency i.e. time required for data flow between the exchange and the broker’s trading system being minimized. The end result is that traders get market information a few milliseconds beforehand which gives them an opportunity to capitalize upon to make profits.
6. As a company, how big do you think is the algorithmic trading market that you can address right now? How big might it be, realistically, in the next five years?
Backed largely by institutional investors, algo trading constitute 46% of the total volume on the National Stock Exchange (NSE) and over 30% on the Bombay Stock Exchange (BSE). We are still way behind markets in the US and Europe in which Algo trades constitute 75-80% and nearly 60% of the total trading volumes respectively. If Indian markets stay fundamentally strong to attract global capital over the next five years, we can see the proportion of algo trades equaling that of the US in the next five years.
7. Many people are not aware about the existence of this form of trading. What has been the response that you got when you set up your company? What is the response now?
Obsolescence, thanks to legacy systems, is what characterizes market trading platforms here in India. The retail trading community in India is guided by the principle of profitability. With awareness of platforms that would leverage the benefits of superior technology and data analytics to enable them to make better trading decisions, leading to greater profitability, we are optimistic that algo trading would see greater adoption. Given that it requires quantitative as well as technical acumen, at present its appetite is restricted to a niche segment of retail investors comprising IITians and other techies. Largely, proprietary traders, investment banks and financial institutions such as brokerage firms constitute a major proportion of our user base. The very fact that we are facilitating over $10 billion worth of daily trades is a vindication of the user adoption of our products from China to Chile.
8. Who are the core people you hire for this business and what are their main skills? Where do you hire from?
Being a fintech company that develops trading platforms which facilitate online trading for retail brokers, algo trading for prop traders and institutional investors and data analytics, the skills we seek in our potential recruits include a strong domain knowledge of the functioning of the financial markets along with software development skills. It is a combination of skills that is pretty rare to find. We hire from B-Schools and Engineering schools which have such talent – software engineers who go on to specialize in Finance.
"The main risk is the algorithms can misfire trades which were not intended as part of the logic."
9. How developed is the eco-system right now? Do you think Indian bourses will keep up in pace with more developed financial markets in adopting technologies such as high-speed trading?
The fintech space catering to the financial markets is highly fragmented and while you have many established IT behemoths also in the fray, it is players who are fleet footed, innovative, technologically superior and primed to cater to customer needs who would win market share. Indian bourses have grown fairly comprehensively over the last two decades to feature in top 7 exchanges worldwide (in terms of number of trades done on an exchange). We believe that in order to continue to attract global capital and stay competitive, they have to ramp up their technology. More than 80% of the Indian institutional trading flow (from both FIIs and Domestic Fund Managers) goes through algorithmic execution currently. The National Stock Exchange (NSE) has witnessed a 25% increase in active investors. While Retail participation stands at 2 crores (up 38% YoY), Institutional participation surged to a new high of $310 billion in FY'15 (up 56% YoY) with 465 FIIs invested in Indian markets.
10. What are the pitfalls of algorithmic trading?
Algo trading, albeit its benefits of market making, greater liquidity, efficient price discovery and efficient order execution, comes with its own share of risks. The main risk is the algorithms can misfire trades which were not intended as part of the logic. Hence, Algo traders need to accord priority to risk management by mitigating and managing risk first, and then focusing on profits. To avoid market crashes, they must observe sound Risk Management System (RMS) checks such as back testing algos against extreme events, kill switches (for cancelling all trades when things go wrong), enforcing compliance and tracking patterns that may lead to systemic risks.