Co-authored with Parker Karia and Pragya Garg, respectively senior associate and associate, Finsec Law Advisors.
Last month, the Securities and Exchange Board of India ("SEBI") proposed significant changes to its framework governing algo trading. To recap quickly, algorithmic trading, or algo trading, refers to any trading activity that automates trades, and does not require manual intervention to place any orders, or monitor prices.
There are two ways in which one can carry out algo trading. The straightforward method is to use the algorithms provided by the stock broker. The other route is through application programme interface (API), which enable electronic systems to connect with each other. Think of it as a data pipe which carries your algorithm. APIs enable the transmission of information, and as a result, a third party can create a code that will execute itself on the broker’s platform.
In the context of algo trading, third parties provide their algo on say, platform X, which is connected to the broker’s platform through an API. Thus, orders placed by the client on platform X get passed on to the broker. Now, while a broker can identify that an order is coming in through an API, it cannot verify that the order is an algo order.
In 2021, concerned with the rise of unregulated algos, SEBI proposed to treat all API orders as algo orders. This was a flawed departure from its mission to encourage innovative and digital solutions in the securities market, as the regulator’s proposal would have saddled connectivity between brokers and other sophisticated players linked to them for non-algo purposes. It appears that the proposal has been scrapped, and after extensive consultations with the industry, a more practical approach has been proposed. With respect to API orders, SEBI has suggested that an order per second (OPS) threshold be specified, and that all API orders above such threshold would be treated as algo orders.
Second, SEBI has proposed to bring “algo providers” (APs) within the regulatory ambit. These would be agents of stockbrokers, similar to the present “authorised person” or the erstwhile sub-broker concept. APs would also have to register with the stock exchange and get their algos approved by the exchange. This would ensure that the broker is responsible for customer grievances, and the redress mechanism deployed by SEBI would be available to AP clients. Individuals who design their own algos would also have to get them approved by the stock exchanges through their broker.
The regulator has also sought to categorise algos into white box and black box algos. White box algos, also known as execution algos, are those which execute orders based on fully transparent algorithms, where the logic, decision-making processes, and underlying rules are accessible and understandable to users and replicable. Black box algos are those whose logic is not known to the user and is not replicable. For providing black box algos, one would be required to register as a research analyst, and for each algo a research report would have to be maintained. In case of any change in the algo logic, it would have to be registered afresh, along with a new report.
The proposed framework places significant responsibilities on stockbrokers. They would have to put systems in place to detect, identify, and categorise all orders above the specified OPS threshold as algo orders. They would also have to ensure that they can distinguish between algo and non-algo orders. Further, brokers would no longer be permitted to offer open APIs, to ensure identification and traceability of the vendor and end user. Whether such restrictions are required in view of measures like the OPS threshold may require more thought. APIs have uses beyond algo trading. The proposed circular should not stand in the way of such use cases.
The stock exchanges would also have to do a fair bit of work. First, they must define the roles and responsibilities of brokers and empanelled vendors, and lay down the criteria and process of vendor empanelment. A turnaround time must be specified for registration of algos, including a fast-track registration for some such as white box algos. Further, the exchanges would have to deploy additional resources. They would be required to conduct post-trade monitoring of algo orders and trades, and put in place a standard operating procedure for algo testing. Further, they must have the ability to use a kill switch to stop malfunctioning algos. They would also have to supervise/inspect that stockbrokers have the ability to distinguish between algo and non-algo orders, as well as issue detailed operational modalities on the roles and responsibilities of stockbrokers and APs, including risk management systems for API orders.
The present proposal thus seems to be more carefully thought out, and seeks to adopt an approach that strikes a balance between the interests of stakeholders and SEBI’s concerns. In another positive move, in its last board meeting of 2024, SEBI’s board granted approval for the recognition of a “Past Risk and Return Verification Agency” (PaRRVA), which shall carry out the verification of risk-return metrics inter alia for algo trading.
There are still some points which may require rethinking as algo trading picks up, such as SEBI’s earlier question of whether an algo should be a facility provided by a research analyst, investment adviser, or a separate class of regulated entities altogether. With the proliferation of artificial intelligence ("AI"), roles, responsibilities, risks, and liabilities are getting redefined.
However, there are few items that may require consideration soon, if not now. For instance, algos may be designed by AI. Risks that may arise due to such instances must also be deliberated upon, along with the roles and responsibilities of the elements involved. While there may never be one right answer, and regulation of something as innovative as algos or AI-based algos is bound to create some unnecessary bureaucracy, it is important for the regulator to at least have a grip on something that could have systemic impact on the markets.
The article was originally published in the Financial Express and can be accessed here.
