Algo trading: Can Sebi's draft rules curb fraud, ensure level playing field

Algorithmic trading refers to investing through computer programmes which can trade faster and more efficiently than a human being with limited intervention, connecting with broker platforms

Sachin P Mampatta Mumbai

6 min read Last Updated : Dec 23 2024 | 10:54 PM IST

It started with a profit of Rs 2,000 to Rs 3,000 for every trading session the first few days. Then came the offer for a lifetime subscription for Rs 53,000. And steady losses ever since.

A user posted the case on social media platform Reddit. They were unable to convince a friend that the company selling the algorithmic trading platform was a fraud. What followed was a string of comments from users who had suffered a similar fate. One user had lost Rs 40,000 in a fortnight. Another complained of paying Rs 60,000 to a provider and then being unable to contact them. A third lost Rs 225,000 in a single day.

Algorithmic trading refers to investing through computer programmes which can trade faster and more efficiently than a human being with limited intervention, connecting with broker platforms through an API (Application Programming Interface). Like any trader, the programme, too, would lose money if the underlying strategy of the trade was incorrect or had lost its edge in a changing market.

Technological challenges abound. But many naïve retail investors are sold algorithmic trading software as automatic technological wizardry, which always gives supernormal returns.

The Securities and Exchange Board of India (Sebi) has sought to bring to heel rogue third-party algorithmic trading providers who prey on investor hopes for easy money. The draft set of rules released on December 13 is likely to give clarity for third-party providers who were operating in a regulatory vacuum earlier, encouraging the growth of genuine vendors while cleaning up others, according to experts. What has also created interest all around is that the new rules allow for retail investors to design their own algorithms though they will have to get it approved before use.

The regulator has reportedly issued notices to multiple brokerages for continuing to associate with the section of algo providers who claimed unrealistic or assured returns. Recent regulator crackdowns involving top brokerages may also have prompted Sebi to put out clear regulations, said one person who was involved in the discussions.

The stock market regulator has invited comments on the draft circular till January 03, 2025.

“Sebi actually wanted these algos to put their own money,” said the person cited above.

The test was to take place for 90 days and then a report with the results of the algorithm’s performance was to be published. Other norms, including a provision for performance verification, have been adopted. The regulator did not immediately respond to questions about the regulations.

Grey area

The institutional space has been well-regulated for years while algorithmic trading for retail investors fell into a grey area, says Kunal Nandwani, founder and chief executive officer at algo provider uTrade Solutions. There has been a surge in the number of retail investors in the segment after the pandemic and this move provides much-needed clarity. Regulation will ensure vendors who do not have the necessary systems in place will move out.

“Algo providers giving API algos must be empanelled,” Nandwani says.

Nitish Shukla, vice-president at discount brokerage firm 5paisa, says approvals for individual algorithms can usually happen within 10 days, often in less than a week, though individual cases may vary. “There may be a cost involved. If there are multiple submissions, the investor would need to bear the expenses accordingly, but costs to get the approvals in place are not expected to be prohibitive,” he says. The amount involved is often less than what is required to buy a low-end smartphone, he adds.

The regulator had earlier sought to deter claims of high returns through a circular in September 2022. It had told brokerages to dissociate from platforms which make such claims. It had also come out with a consultation paper in 2021 on regulating algorithmic trading for retail investors.

Though algorithmic trading is said to have been in vogue since 2005, according to a 2013 note by the Motilal Oswal group the regulator had first formally provided a framework for regulating algorithmic trading in 2008 through a regulatory framework for providing Direct Market Access (DMA). This allowed institutions to connect directly to the stock exchange platform, though it would nominally pass through brokerage infrastructure.

Subsequently, institutional investors who made use of algorithmic trading were required to vet their algorithms before use. The share of non-algorithmic trading has declined to less than 20 per cent (Charts 1,2). And a similar regime is now being extended to retail investors.

Black and white boxes

The regulator in 2021 proposed that all API orders could be tagged as algo orders. But an API is only a means for different software applications to communicate or exchange data, and not all orders may reflect the kind of fast-paced trading Sebi seems to be looking to regulate. Its 2024 draft circular applies the label only to orders above a certain threshold per second, which is yet to be decided.

It has classified algos into two. The first category is “white box” algos, where the strategy is available and understandable to users. The second category, “black box” algos, has strategies which are not disclosed. The black box algo provider will have to register as a research analyst and maintain a report on the strategy, updating the report and authorities when there is a change. This is in addition to empanelling with the exchange as an algo provider.

Brokerages can only deal with empanelled algo providers who would be considered their agents, and handle any complaints against them. They will also require that access be granted through means that would be unique to a vendor and user.

This will likely prevent algo providers from acting as pseudo portfolio managers, according to an expert. It was technically possible for someone with an algorithm to trade on behalf of multiple people otherwise.

The person cited earlier, who was involved in regulatory discussions, says it would be difficult to track changes in strategy. This can create challenges for enforcing the rule that all authorities need to be informed of updates in underlying operations. Unless there is a major and obvious change in the behaviour of the system, one would largely have to go by the algo provider's words that there has not been a change.

The regulator in its board meeting on December 18 approved an agency to evaluate claims of risk and returns through a specialised Past Risk and Return Verification Agency (PaRRVA). “PaRRVA shall carry out the verification of risk-return metrics for Investment Advisors (IAs), Research Analysts (RAs) and Algorithmic Trading, and those persons permitted by the Board to offer these services,” it said.

All unrealistic claims can finally be verified.