Retail Algo Trading under SEBI’s Lens: The Unfinished Business of Black Box Regulation

[Akash and Rashi are students at Maharashtra National Law University, Mumbai. The following article is one of the winning entries in the 2nd Article Writing Competition 2025 organized by IDIA: Increasing Diversity by Increasing Access to Legal Education (Odisha Chapter).]

In the past two decades, the development of technology has taken over the financial markets by storm with the expansion of algorithmic trading (algo trading). With the circular dated 4 February 2025 (Circular) issued by the Securities and Exchange Board of India (SEBI), retail investors have come within the regulatory framework of algo trading. Previously, while the retail investors could utilize algos for trading, they were largely inaccessible and unregulated. By regulating the access of retail investors, SEBI has ensured a safer platform with reduced systemic risks when such investors rely on algos.

While the Circular is a positive step in encouraging retail investors to engage in algo trading, it is handicapped by certain ambiguities. This article aims to delineate these ambiguities specifically with respect to the regulations on black box algos. The first chapter focuses on the legal lacunae in the current framework on functioning of black box algos. The second chapter specifically brings forth the ambiguities within the Circular with respect to registration of black box algos.

Data Integrity and Black Box Regulation

The Circular distinguishes between 'white box' and 'black box' algo trading systems. White box algos are transparent, with accessible and auditable logic, while black box algos operate using proprietary, non-replicable logic that is undisclosed to brokers or regulators. SEBI in the Circular mandates stricter compliance for black box algos, including approval requirements, audit trails, and risk controls. Despite offering greater speed, accuracy, and the ability to process complex patterns, black box algos raise significant concerns regarding market stability and the potential for manipulation.

Black box algos work over multi-dimensional data stream input, autonomously executing trades based on complex, non-interpretable logic. The complexity of the data input in each such algo makes it non-replicable and logic incomprehensible. This creates a threat of market manipulation as the execution of trade is solely based on the input data. While the Circular creates an obligation on the research analysts (RAs) to maintain a research report on each such algo, there exists no regulation over the input data. This further raises concern over input of data that might be categorized as unpublished price sensitive information (UPSI). If such information is used as a data input, it can result in market manipulation as well as insider trading. Owing to the lack of regulation over the data input for black box algos, insider trading using such UPSI will further create difficulties in tracking and identifying insider trading.

A plausible solution to avoid such circumstances would be to require each black box algo to generate and retain real time logs of decisions, inputs and execution. This would enable audits and analysis of crashes, errors and manipulation on the basis of computer accurate data and without risk of human bias, manipulation or error. Additionally, the SEBI can also mandate and delineate  the nature of input data. The input data should only contain information about the company which is known to the public. It should be made sure that no internal sensitive information about the company should be part of the input data being fed. To enable this step, the SEBI should categorize and delineate the type of information that can act as input data for the logic of the black box algos which will act as a risk ameliorator. 

While there is a safeguard placed by placing an obligation on the algo provider of black box algos to register as a RA and maintain a research report, an RA is only required to confirm the maintenance of such research report to the exchange. In accordance with SEBI (Research Analysts) Regulations 2014 (Regulations) such report is inspected suo moto by SEBI, on receival of information or complaint. However, such limited inspection of research reports is not sufficient to regulate black box algos. A greater scrutiny with a structured inspection timeline of research reports pertaining to black box algos is necessary to minimize risks associated with black box trading. There should also be a mandatory obligation of SEBI to review these reports with respect to the strategies being used. This review has to be done before these strategies are being put to use in order to probe the aspects of market manipulation and insider trading. 

Furthermore, there exists ambiguity regarding whether the Regulations will be applicable as a whole or the relevant provisions pertaining to registration of RAs and Maintenance of Research Reports. Such confusion arises since the Circular in Paragraph V(a)(ii) only provides for “registration” of RAs and “maintenance” of research reports. Additionally, lack of clarity also stems from the fact that the Regulations impose a range of procedural and substantive requirements for registration of RAs, including stringent eligibility criteria, qualifications, and experience thresholds. These may not align with the practical skill sets or roles of many algo developers, especially technologically adept retail investors who may not meet the conventional benchmarks set for RAs. As a result, the regulatory framework, if applied wholesale, may inadvertently exclude a significant class of non-institutional algo providers from compliance, thereby stifling innovation and discouraging participation in algo trading. In this regard, the SEBI has to clarify the applicability of the Regulations to RAs for black box trading. Additionally, the eligibility criteria should also be waived off in order to create an invigorating environment for algo trading. 

Registration Conundrum

The black box algos can be a revelation for retail investors. However, the Circular has certain internal contradictions that create confusion over compliance and implementation. This regulatory ambiguity not only creates uncertainty but also hampers the adoption of black box algo trading by retail investors in India. The legislative vacuum has crept into the aspect of registration of algos. This can be said as the Circular doesn’t explicitly delineate as to whether both the black box and white box algo has to be registered. However, in Paragraph 5(V)(a)(ii), it delineates that if there is any change in logic in the black box algo it has to be registered as a fresh algo by the algo provider. The usage of the word ‘fresh algo’ depicts the intention of the Circular that prior to change in logic as well, black box algos have to be registered. In other words, if the Circular mandates registration of change in logic, it follows therefrom that the unchanged previous logic is also registered. 

On the other hand, the Circular in Paragraph 5(I)(c) specifies algos created by retail investors having crossed the specified order per second threshold should be registered. The clear intention of the Circular in this case is to demarcate the non-mandatory nature of registration of the algo if the specified order per second threshold has not been crossed. These literal interpretations of the paragraphs of the Circular create ambiguity. This can be said as the question then arises as to whether the black box algos having lesser order per second threshold should be registered or not. This is mainly because the Circular in Paragraph 5(I)(c) is clear in mentioning the word ‘algos’. The broad usage of the word algos includes both white box and black box algos. A bare perusal of Paragraph 5(I)(c) creates a scope of a certain portion of black box algos not getting registered which falls outside the scope of regulation by the exchange. 

It is also important to note that Paragraph 5(I)(c) of the Circular expressly states that retail investors can create ‘algos’. However, Paragraph 5(V)(a)(ii) refers only to algo providers in the context of creating black box algos. This inconsistency raises a critical question as to whether retail investors can be considered algo providers for the purpose of black box algo development and regulation. Since the term ‘algos’ in Paragraph 5(I)(c) is used broadly, without limiting it to white box or black box algos, it logically follows that retail investors may create both. Yet, the absence of the term ‘retail investor’ in the Paragraph 5(V)(a)(ii) dealing specifically with black box algos introduces regulatory obscurity. It is necessary to clarify whether the terms ‘algo provider’ and ‘retail investor’ are intended to be interchangeable under the Circular or whether distinct regulatory treatments are envisaged. This ambiguity has to be resolved considering the ramifications of the Circular. Black box algos which are confidential in nature should be clarified to be registered irrespective of the orders placed per second. This enhances the safety and also reduces the misuse, as the money trail comes under the regulation of the exchange.

Conclusion

It is necessary to have a robust system in place for complicated algos like that of black box algos. It is incumbent on the SEBI to make regulations in accordance with the non-linear functioning of black box algos. This will make the Circular more lucid. While the decision of SEBI to have a regulatory framework for retail investors is a step in the right direction, the Circular seems to be premature. This impetuous step has to be remedied by having intelligible clauses in the Circular which will act as a booster for algo trading.