The invisible hand in India’s stock market

Not many people would have heard about these firms. But these are all global majors doing business worth hundreds of millions of dollars in India in the highly-secretive world of algorithmic trading in stock markets.

Algorithmic trading—known as algos in market parlance—refers to a software designed to execute trading strategies. Algos can execute trades in a fraction of a second, which is just not possible to do manually. In the world of algos, the speed at which the trades are executed is of paramount importance. A split-second delay and the potential gains could be significantly reduced.

Simply put, a code is written to automatically trigger a buy or a sell trade when, for instance, the price of a certain share goes below or above a pre-defined level. Or, it could be designed to capture the small difference in the share price of a company on the BSE and National Stock Exchange (NSE)—known as arbitrage gains in market jargon.

While these are basic examples of what an algo can do, market participants say that more complex and sophisticated algos are now being used by the deep-pocketed global and even domestic institutional investors and brokerages.

Most of the large Indian brokerages, such as Edelweiss Securities, IIFL, Kotak Securities and Motilal Oswal Financial Services, among others, also offer algo-based trading facilities.

They service foreign and domestic institutional investors who prefer algos to execute strategy-based trades in India.

Market participants further add that while the global majors have set shop in India in the last 3-4 years and are seeing a consistent rise in trading, they are tight-lipped about their activities to avoid attracting undue regulatory or media attention.

“There are certain businesses where secrecy plays an important role and you can say that algo is definitely one of those," said the founder of an algo firm who requested anonymity. “Globally, regulators view algo as a very high-risk trading mechanism and, hence, algo firms do not like to talk much about their business activity. But it is a fact that global players use smart and sophisticated algos in India, which now offers a decently deep market to put their algos into play," he added.

It is believed that algo trading has now become quite popular in the Indian equity markets and nearly 70% of the daily turnover is attributed to algos.

That is a pretty large share considering the fact that algo trading was allowed in India only in 2008 (though market regulator Securities and Exchange Board of India (Sebi) has barred the use of algos for retail investors).

Orders from co-location services—wherein the broker keeps his server close to that of the exchange to reduce the latency in order execution—use algos, as do direct market access (DMA), where the investor can directly send the order to the exchange through the broker’s terminal but without any manual intervention at the broker’s end.

“Algo trading has gained a lot of traction in the Indian markets with around 60% to 70% of the daily turnover estimated to originate through algos," said Kunal Nandwani, co-founder and chief executive officer, uTrade Solutions, a fintech company that develops algo-based trading platforms.

“Algos have (also) become faster and more sophisticated over the years," he added.

The buy-side players like mutual funds often use algos to rebalance their portfolio that might run into thousands of crores. Algos will slice the order and keep executing it in parts so as to keep the impact cost—the cost of executing a trade— low while also minimising the probability of front-running that would attract the regulator’s ire.

Front running is an illegal way of trading in shares based on the insider knowledge of an institutional trade that could potentially influence the stock price on a particular day.

Algos have gained such immense popularity that, as per industry estimates, most of the large Indian brokerages now see over 80% of their institutional turnover emanating from algos.

According to market estimates, 15%-20% of the trades in algos are more complex and sophisticated.

The dangers

Around a year back, on 13 March 2020, the benchmark indices hit their 10% lower circuit limit and, as per regulatory norms, trading had to be halted for 45 minutes. Once trading resumed, the benchmarks staged an unprecedented rally to gain over 4,700 points. What caused such a rally?

Market players said that the intense buying which ensued after the huge fall was triggered, in part, by algos that were programmed to initiate buy or sell orders based on the S&P 500 levels. Incidentally, there was huge volatility in the US markets too, as just a day earlier, the US Federal Reserve had announced a trillion-dollar stimulus package.

Similarly, on 31 May 2019, in the span of less than 30 minutes, the benchmark Sensex witnessed a sharp and sudden 700-points fall, which, according to market players, was again due to some algo trades gone sour. There are many such instances—and even mishaps. In 2011, all the trades done in the equity derivatives segment of BSE during the special Muhurat trading session were cancelled due to an algo trading mishap.

Sebi has put in place certain checks for the algos that are being used in the Indian markets. One of the checks is what’s called order-to-trade ratio, which refers to the ratio of total number of orders being sent by the algos to the total number of orders executed. A lower order-to-trade ratio attracts penalty as it means that the triggers set in the algo are not in sync with the current market prices.

“The checks are important because the stakes are high. A faulty or rogue algo can cause huge disruption in the markets," said a regulatory official on condition of anonymity, while adding that the issue of regulating algo is still evolving and regulators globally were working together.

According to the Indian regulatory framework, any algo before being introduced in the market has to be approved by the stock exchanges and the Securities and Exchange Board of India (Sebi) as well. This check has been put in place to minimise the probability of a rogue algo disrupting the safety of the markets.

The danger is accentuated because the bulk of institutional trading is done through software codes and without any manual intervention. “The kind of algos that are used currently are quite diverse and, over the years, the software has become smarter even as the Indian markets are not as deep as the developed markets," said Hrishabh S., founder of a startup called Railofy and formerly the head of algo trading and direct market access at Edelweiss Securities.

“There are simple execution algos that institutional brokerages offer their clients. There is alpha-generating algos that hedge funds, arbitrageurs and proprietary desks use. Most of the large brokerages develop algos in-house while those without deep pockets depend on third-party vendors," he added.

The secrecy

While there are third-party vendors in India who sell algo software, most of the brokerages have developed in-house capabilities to develop the code and offer them to their clients. This also helps them keep their algo strategies a secret.

Incidentally, that is one reason why even though there are quite a few third-party domestic and global algo makers, they face stiff competition from brokerages that prefer in-house teams to develop and manage the algos.

Algos have a certain aura built around them. They are looked upon as highly sophisticated codes that can make millions or even billions in a fraction of a second. Such products do come with their share of secrecy. Not surprisingly, the trend of developing in-house algos is fast gaining pace. This is also one reason why it is nearly impossible to gauge the kind of algos in play in the Indian market or, for that manner, any market across the world.

While market participants are unanimous in their view that global institutional investors—global majors like CLSA, Citi, JP Morgan and Credit Suisse to name a few—use complex algos in India, there is no way to corroborate it.

“There is no public database of algos as such and no firm would talk about their algo strategies publicly," said the institutional sales head of a domestic brokerage that offers algo-based trading facilities to its clients.

“Hence, no one would know what are the kind of complex algos being used here. Developing an algo costs a lot and if the secrecy element is lost, then the advantage would also be lost as someone might just try to replicate the model. There are a few regulatory checks in place but that’s it. No one would know the kind of algo his or her competitor might be using," he said on condition of anonymity.

Algos are high-end sophisticated products and hence cost a lot to develop and manage. That is the reason why big brokerages are able to do it in-house while others are mostly dependent on third-party vendors.

“For a brokerage with a few institutional clients, the annual cost of developing an algo and managing it would run into crores. For the slightly larger ones, it could be still higher. Hence, only the bigger ones maintain an internal team for algos," said a brokerage official who wished not to be named.

Developing an algo requires a team of highly skilled persons. “One needs to hire people who understand both programming and trading. It is a highly specialised skill set and it is not uncommon to see mathematicians, statisticians, economists and even physicists with PhD degrees working on algo development. Not to forget the software engineers with a specialisation in machine learning, big data or even natural language processing (NLP)," said the person quoted above.

Algos require co-location facilities too and hence firms need to budget for that as well. According to market participants, the annual cost for co-location starts from around ₹6 lakh and could go up to ₹15 lakh or even more depending on the number of servers that need to be kept in the exchange premises.

The way ahead

Even as algos have gained immense popularity and acceptance in most markets globally, there is also a view that such trading comes with its own share of risks— hence only those with an appetite for sophisticated and complex products should be allowed to use algos.

In India, Sebi has barred the use of algos for retail investors. While a few years ago, there was a view that algos could be made accessible to retail investors as well and feedback was sought from market participants, nothing concrete came out of it.

A section of market participants are of the view that this disturbs the level-playing field that every investor is entitled to, as only one category of investor has been kept out.

“This is a big drawback for retail investors since all other categories of investors, including the proprietary desks of brokerages, are using algos to design trading strategies. It’s time that the regulators should look at allowing algos for retail investors as well while putting in place proper checks and balances. Algos are much bigger in the other developed markets and India is also going the same way. Hence, it becomes all the more important that a single set of investors is not left out," said Nandwani of uTrade Solutions.

Interestingly, market participants also add that even as the regulator has barred retail investors from doing algo trading, there is no explicit ban on retail investors creating and testing strategies in the back end and then executing it through a regular broking platform.

“There are services available that let retail investors create a strategy and even do back testing to test its efficiency. They can even deploy it in a dummy market environment that effectively mirrors the live market. This is, in a way, using algos but without executing the actual trade. So, the regulatory norms are not flouted even as retail investors get a taste of algos," said an algo developer on conditions of anonymity.

Ashish Rukhaiyar is a Mumbai-based journalist covering capital markets.