Is history repeating itself, with retail investors starting to trade algos without a regulatory structure in place? Until recently, we believed that there were two markets operating within our stock exchanges. One is what Michael Lewis described as the Flash Boys who have their servers co-located in the server farms inside stock exchanges to get the fastest access to stock quotes so that their algorithms can find the tiniest profit opportunities and execute giant trades within milliseconds based on pre-defined parameters. Then there is the other market comprising retail investors and other institutional traders who don't need to get in and out of stocks as fast and soon as possible.
This is no longer true. Retail investors are quickly getting themselves turbo-charged with off-the-shelf algos that are being offered by a dozen-odd algo-writing platforms that are completely outside the purview of regulation. As many as 40 top brokerage firms are offering algo-based trading to clients and there is no clarity on the number of firms writing such algos.
A source inside the Securities and Exchange Board of India (SEBI) told me, “SEBI is aware that brokers offer algo trading to retail clients. It's only fair that algo should not be limited to institutions and big players. The current system is for brokers to take approval of exchange for all algos. So far, no need has been felt for separate framework for retail algos.”
The algo space should certainly be democratised to allow smart retail traders to use the system; but the view that it does not need to be regulated suggests that SEBI has not applied its mind correctly, or is being wilfully blind.
I also discover that both, the National Stock Exchange (NSE) and the Bombay Stock Exchange (BSE), are concerned about the lack of regulation, so are conscientious algo-writers; but the regulator has ignored the issue, despite letters and informal discussions with its officials.
Dharmik Thakker, founder of Algomaker, an algorithmic trading firm that writes code and creates trading strategies for retail clients, has been blowing the whistle on the absence of regulation by writing and attempting to engage with the regulator, without success.
How Does Retail Algo Work?
SEBI defines algorithmic trading as any order that is generated using automated execution logic. And API (application programme interface) allows an individual to access brokers’ trading platform without needing to log in manually. The algo writers, usually, show a back-tested report on trading strategies to sell the idea that you can make a lot of money on rapid intra-day trading.
Some algo makers have been advertising their products and charge about Rs500 a month to subscribe to an algo. They have also tied up with dozens of top brokerage firms to ensure direct integration with their platforms through a ‘bridge’. The retail client, thus, has a turbo-charged system to earn a living by day-trading. But nothing is sure-shot in the markets; many incur large losses as is evident from the anger that spills into social media platforms every time a new break or a tech glitch upsets the programmed logic.
Such discretionary algos, that claim to make money and generate returns, are sucking people into trading in the past couple of years. While some say that the algo market for retail investors is still small, with each broker having a few thousand clients, this is actually growing very fast. Dharmik says that, based on NSE data, that the current size of portfolio handled by API-based algo strategies is 20% of the trading volume.
Nithin Kamath founder of Zerodha, says his firm charges fees of Rs2,000 a month for an API access. But Zerodha restricts the use to those who really understand how to use them. “However, a person can pay for the API and integrate it himself with one of the algo writing platforms, which is outside our control. Zerodha does not allow direct integration from its platform,” he says. There are at least 40 firms that are offering these services.
What Is Wrong with Retail Algos Today?
Democratising access to the market by allowing tech-savvy retail investors to trade algos is a good idea, provided it is properly regulated and retail investors understand the risk. Today, there is not enough information about the size of the market or the kind of scrips in which algos are offered. Worse, they seem to fall foul of the stringent regulations that SEBI has imposed on RIAs (registered investment advisers) and RAs (research analysts) who take far greater responsibility for the advice they offer clients.
Significantly, with over 25 broker defaults at the NSE in the past two years, brokers are allowed to offer advice with no rules or restrictions while also holding the client’s Power of Attorney (PoA). This raises serious issues about risk assessment by SEBI and the Exchange. Worse, there is no clarity about where these retail algos are deployed. The anger among retail investors, and their claimed intra-day losses when there are sudden market movements or disruptions, probably indicate what is going on.
Need for a Regulatory Framework
As I said earlier, almost everybody, except SEBI, thinks that retail algos need to be regulated. Mr Kamath thinks algo trading strategies skirt SEBI’s strict regulations for RIAs and RAs. This is because an algo is essentially a computerised trading program that assures profits in certain situations. Or, as Nithin Kamath says, “it is an algo that is giving advice” instead of an individual and that ought to fall under SEBI regulation, with the attendant onerous record-keeping. Remember, SEBI has ruled that robo-advisories, under which automated investment advice would be doled out based on some customer parameters, are not legal.
An obvious comparison is with the rules that apply to large firms using algos through the co-location services of exchanges. These algos have to be approved by the exchanges and have a rather controversial past, as we have documented in our book Absolute Power
. In fact, SEBI was caught napping even then. It allowed NSE to start algo trading and offer co-location without official clearance in 2010 and finally issued “Broad Guidelines on Algorithmic Trading” only on 30 April 2012. In 2015, Moneylife
broke the NSE algo scam that led to sweeping changes in NSE’s top management.
It is extraordinary that SEBI is behind the curve once again, especially when all algo platforms in the international markets require some kind of regulatory licence. What makes the Indian situation surprising is that SEBI has imposed such stringent compliance and capital requirements for RIAs and RAs which, if applied to tiny algo platforms, would make them unviable. And, yet, we have Dharmik Thakker from the industry itself pleading for a clear regulatory framework.
His reason is clear. Sleeping regulators wake up when there is a big scam or the market collapses. They then turn into a super-cop and initiate drastic action that can drive companies out of business, while absolving themselves of the responsibility for failed regulation and supervision.
I wrote to the SEBI chairman, the chairman of its technical advisory committee (TAC), managing directors of the two national stock exchanges and the largest brokerage firm in the country. All except SEBI chairman responded and everybody agreed that it was a matter of concern. The TAC, however, felt that regulating retail algos would need to be looked at by SEBI’s risk advisory committee (RAC) and not the TAC.
Who Should Regulate Retail Algos?
Regulating algos used by retail investors is, indeed, complex. For starters, algo writers, who are offering programmed investment advice, are not registered with SEBI or the stock exchanges. Also, while SEBI could regulate algo writers who are marketing their products openly and offering direct integration through brokers, there could be thousands of individual programmers who can write algos and fly below the regulatory radar. “In a highly automated environment, it is not possible for a broker to know whether the trading strategy of a customer is coming from an algo platform or his own tech skill,” says Mr Kamath.
The point to understand, says Dharmik Thakker is that there is no difference between API-based algorithms and algorithms deployed in co-location services except the latency. If algos in co-location require approval, this should apply to retail algos too. He says that regulations should clearly define: What are the bare minimum requirements for algo service-providers in order to build trust in the system and avoid trading strategies that are based on ‘pump and dump’. He also says that the regulations need to cover algo writers as well as brokers who provide access by requiring them to get their algos approved by stock exchanges. The approval, he says, should cover their software, strategy, scrips and also order to trade ratios.
Mr Thakker believes that API-based algos can run on any scrip, including penny stocks, and to manipulate a stock using algos, all one needs is multiple user accounts on which any algorithmic strategy can be deployed. An exchange official, with a tech background, agrees that, unless checked, this can be used for systematic market manipulation and even tax evasion. Mr Thakker believes that even innocent investors can become pawns of such algos because end-users may be clueless about the exact trading strategies that they end up deploying. He has prepared a case-study by placing an order in ‘stock X’ with 40 user accounts that subscribed to his service, which shows how bulk orders through the algo could be easily used for market manipulation. The question is: Who will wake up a sleeping regulator?