.png)
Governing the Algorithm: Implications of SEBI's Proposal for Retail Algorithmic Trading: Part I
- Abhishek Sanjay and Tanya George
- Apr 1
- 9 min read
Updated: Apr 16
Abhishek Sanjay and Tanya George*
I. INTRODUCTION
Algorithmic Trading refers to computerized trading utilizing a set of pre-determined parameters in trading transactions. These algorithms convert large quantities of highly sophisticated data within microseconds, far outweighing the skill of human traders. In 2008, the Securities and Exchange Board of India (SEBI) opened the gates for Algorithmic Trading in India with the introduction of Direct Market Access (DMA), although limited solely to institutional investors. The post-2010 era has witnessed the popularization of algorithmic trading for retail investors, which has created a regulatory lacuna for platforms, with misleading claims and uncorroborated performance claims often creating major losses for retail investors. A report published by SEBI evidences that 97% of foreign investor profits and 96% of proprietary traders’ profits in F&O were attributed to algorithmic trading, clearly establishing an incline for profits for high-frequency trades.
Noting the growing demand for platforms offering algorithmic trading, and in line with jurisdictions allowing for open algo trades, SEBI recently issued a circular enabling the widespread participation of retail investors in algo trading to ensure scrutiny and transparency. The status quo and the existing regulation merely acts as a guardrail with retail algorithmic trading being limited to certified stock broker Application Programming Interfaces (API). However, as acknowledged by SEBI in the past, they are not entirely equipped to handle all instances of manipulative and disruptive practices brought on by algo-trades. Simply put, with the instantaneous growth of tech, methods of disruption through algo-trades now must first be identified and can only then be penalized, adding further deceleration to regulations that already function as ex-ante. The more recent circular released in February, 2025 hardly provides any tangible safeguards against the concerns raised by the present paper. Thus, while retail algorithmic trading also brings a new dynamic to the well-established and long-familiar trading approach, it brings within its wings cascades of potentially unfamiliar problems that the established trading system cannot handle. This article holds that the longstanding regulatory framework looks increasingly fragile in the modern market design, hailed by algorithmic trades.
While existing literature has largely overlooked the potential market-wide effects of opening algorithmic trading to retail investors, this paper seeks to bridge this gap by demonstrating how SEBI may have bit off more than they can chew, through empirical data analysis. Further, the solutions offered present a novel approach by shifting regulatory focus from inefficient reforms to structural market reforms that enhance fairness and efficiency while being tailored to address the necessity of reducing the disparity between small and large players in the algo trading field.
In Section II, the authors examine SEBI’s decision to open algorithmic trading for retail investors, laying the foundation for the paper’s central argument, which holds that while SEBI's initiative aims to democratize access to advanced trading technologies, it may inadvertently introduce significant risks due to inadequate oversight and the potential for market manipulation. Section III explores the broader implications of algorithmic trading on market dynamics, particularly concerning its impact on liquidity and volatility. The authors analyse the systemic risks posed by market fragmentation, short-termism, and manipulative practices like spoofing, which are exacerbated by the proliferation of algo-trades and argue that although algorithmic trading may initially enhance liquidity, its long-term effects often destabilize markets. In Section IV, the authors argue that the existing oversight mechanisms are inadequate to address the sophisticated nature of contemporary trading algorithms and advocate for the implementation of rigorous pre-deployment testing, real-time monitoring, and enhanced regulatory scrutiny to pre-emptively mitigate systemic risks. In Section V, the authors examine and propose systemic solutions attuned to the needs of the Indian market to regulate algorithmic trading, while ensuring that technological advancement remains adequately incentivized. The solutions proposed are tailor made for the retail investor, differentiated from institutional trading due to the starkness in technology and trade volume involved.
This article is divided into two parts to provide a comprehensive analysis of SEBI’s decision and its broader implications. The first part comprises Sections II and III, which focus on SEBI’s initiative and its potential risks, along with the systemic impact of algorithmic trading on market dynamics. The second part, consisting of Sections IV and V, delves into the limitations of current regulatory mechanisms and proposes tailored solutions for the Indian market to balance innovation with investor protection. Part II of the article can be accessed here.
II. SEBI'S RECOGNITION OF ALGORITHMIC TRADING
In a stride forward, SEBI, on December 9th, published a circular titled "Participation of Retail Investors in Algorithmic Trading", which provides a regulatory framework for algo trading and thereby hails the coming in of retail investors into a level playing algo trading field. SEBI’s new move seems to target illegal algorithmic traders and thereby introduces retail investors to be amenable to effective risk management. Further, while this officially premises the introduction of retail investors into the market, what it essentially brings forth is a change in nomenclature. Prior to the consultation paper, retail traders often used algorithms for trade, albeit unregulated and classified as “internet-based trades”. SEBI’s recent move allows them for a heightened span of control with their classification as algorithmic trades.
This move brings brokers within the regulatory purview by placing certain responsibilities on the broker, such as authenticating access through two-factor authentication and having OAuth-based authentication. The broker further becomes responsible for all the algorithms emanating from their APIs and redressal of investor disputes. Each exchange now bears the obligation of establishing a unique regulatory ID for each algorithm for auditing purposes. Additionally, brokers are now allowed to lawfully deploy their APIs to the public, once they obtain the permission of the stock exchange. However, all subsequent modifications to the algorithm are then subject to the approval of the exchange. This significantly hampers the effectiveness of one's strategy in a dynamic environment where every millisecond is crucial. As a result, the computational advantage provided by the algorithm becomes redundant, since it cannot be modified at a pace quick enough to meet ever-changing market synergies.
These stringent requirements, though essential for market integrity, could impose a heavier burden on retail investors compared to institutional investors, who typically possess more extensive resources and industrial knowledge to manage and work within compliance demands. Further, due to the disparity in resources when comparing retail investors to their institutional counterparts, institutional investors may also be able to speed up processes of modification and permission, thereby gaining the benefits of the market at a higher frequency than their less-equipped counterparts.
The framework further categorises algorithms as White Box Algos and Black Box Algos with white-containing algorithms that are transparent and replicable and Black indicating algos where the trading strategies are not disclosed but adhere to a higher level of scrutiny by maintaining detailed reports on the algorithm’s functionality. Although incrementally a step forward, this might bring in hassles in the long run as trading fundamentally requires a day-to-day checking of market tendencies, which in turn require tunings to the algo, which now cannot be done without registering the algo afresh in cases of Black Box algos. This would lead to significant hindrances on every minor scrutiny, which then becomes redundant at the point of a high-frequency trading system as this leads to their reduced competitiveness in the market. As shown by a study, even a seemingly benign 10-millisecond delay leads to significant decreases in returns for traders. Taking the example of Knight Capital Group, where it lost 440 million in just one hour, based on a human error in code, not allowing for minor changes in rewriting code speedily holds the risk of creating detrimental black swans in the stock market.
Speed has become a hallmark of market trading, with algorithms mining, processing, and delivering an order in milliseconds and microseconds. SEBI’s recognition of algo trading would be formulated with the Orders per Second Threshold, which holds that any order in excess of 120 orders per second would be tagged as algo orders, in addition to those already registered as algo orders. Similarly, brokers hold the authority to categorize all orders above this threshold as algo orders. This allows investors to absorb little economic risk in holding onto securities, in a sort of “hot potato” fashion. While this ensures that there is an efficient recognition of which trades are algorithmic, it fails to consider the heightened cancellation of orders in milliseconds, which would not, in turn, satisfy the order per second criterion but calibrates to create sporadic shifts in market trends and price movements, due to the market synchronizing rapidly to reflect incoming information.
Additionally, the stock exchange holds the ability to use kill switches for orders emanating from a particular algo ID. Simply put, these function as trip wires that cut investors off and cancel or withdraw existing orders based on pre-defined parameters. With algo trading’s widespread potential for disrupting markets, a kill switch may be deemed necessary. However, it has been argued that kill switches are an ill-equipped regulatory mechanism as technological advancement allows humans to write codes around the parameters of activation, evading its application. Furthermore, the activation of a kill switch for minor errors that can be internally modified by the coder may stifle his profitability as investors now stand to lose huge amounts in mere seconds. Applying the same to an investment firm holds the potential to deter future investors from working with them in the future, due to this regulatory delay. Therefore, how effective would it be to apply an alien red button to a first-of-a-kind, fast-paced and already fragmented market? This raises the question of whether a multi-layer kill switch would act as a better means of regulation by enabling traders to internally modify coding errors, which SEBI seems to have deemed unnecessary. Lastly, as vehemently argued by Steinfeld, enabling a mechanised automated program to implement kill switches would act as antithetical in the long run as these programs, being run on pre-set parameters, cannot account for unforeseen black swan events that require urgent intervention.
III. THE RIPPLE EFFECTS OF ALGORITHMIC TRADING ON MARKET STABILITY
While it may be argued that the advent of algorithmic trading has fostered increased competition by minimizing entry barriers for smaller firms, it has led to a state of market fragmentation as trading avenues multiply to accommodate different algorithmic strategies. One of the most notable shifts associated with algorithmic trading is its impact on trading behavior wherein traditional strategies such as buy-and-hold have been increasingly supplanted by HFTs. The focus has shifted towards micro-level price adjustments and arbitrage opportunities, sidelining long-term investment strategies and potentially fostering short-termism.
A significant argument in favor of algorithmic trading is its purported role in enhancing market liquidity; however, empirical evidence says otherwise. While its entry into markets has improved liquidity metrics such as reduced bid-ask spreads and enhanced trade execution in the short term, the long-term implications reveal a grim outcome. Markets with high algorithmic activity often experience a saturation point where the initial liquidity benefits taper off. Beyond this threshold, the strategies employed by algorithms, such as statistical arbitrage, can detach market prices from their underlying fundamentals, introducing inefficiencies and distortions. In periods characterized by high information asymmetry, such as dividend announcements, algorithmic trading often amplifies trade volumes, offering a semblance of enhanced liquidity. However, this liquidity is frequently consumed rather than provided, as algorithms prioritize exploiting momentary price discrepancies over stabilizing the market. It is this very myth of liquidity which makes such algorithms as dangerous as it is. The increase in HFTs through SEBI’s move would only exacerbate the same, especially in a nascent market such as India.
Algorithmic trading’s ability to integrate information into prices at unmatched speeds has undeniably contributed to market efficiency. However, studies have shown that the same speed and volume of trades, amplify market volatility, particularly when multiple algorithms respond to market shifts in unison. Algorithms designed to outpace competitors often widen bid-ask spreads or withdraw from the market altogether during black swan events, compounding liquidity issues and intensifying market turbulence. Following the resultant uncertainty, large trading entities, including high-frequency trading firms, frequently scale back their positions to mitigate risk. This withdrawal, however, adds further downward pressure on already declining markets. As markets spiral lower, stop-loss mechanisms are triggered in a cascading effect, creating a self-reinforcing cycle of decline. A prolonged market downturn, driven by these feedback loops, erodes investor confidence and sends recessionary signals that resonate beyond the financial sector.
A major factor adding to the market’s volatility is the practice of “Spoofing.” Spoofing involves placing large, deceptive orders that are never intended to be executed, creating a false impression of market activity. This tactic manipulates the behavior of other traders by fabricating the appearance of significant buying or selling interest and is a criminal offence in the U.S. under the 2010 Dodd-Frank Act. A stark example of the perils of spoofing is the Flash Crash of 2010. While the initial investigation by the Securities and Exchange Commission (SEC) and the Commodity Futures Trading Commission (CFTC) attributed the crash to a $4.1-billion trade by a Kansas-based mutual fund, later developments shed light on a different culprit. U.S. authorities, in 2015, charged Navinder Singh Sarao, a London-based trader, with market manipulation for using his algorithm to carry out spoof orders in the E-mini-S&P 500 contract, which later came to be known as the “NAV” scandal. Spoofing has also been recognized in India with SEBI, in March 2021, issuing a circular addressing the manipulation caused by excessive order modification & cancellation. SEBI's 2023 order against Nimi Enterprises, found the entity guilty under Section 12A of the SEBI Act read along with Regulations 3 & 4 of SEBI (Prohibition of Fraudulent and Unfair Trade Practices relating to Securities Market) Regulations, 2003 [PFUTP Regulations], marking the first instance where it formally used the term “spoofing” to describe such conduct. The NAV scandal is a textbook example of the perils associated with unregulated algorithmic trading, even with scarce capital.
Part II of this article examines the regulatory shortcomings that render the existing framework ill-equipped to manage the risks associated with retail algorithmic trading. Section IV assesses the limitations of current oversight mechanisms, with particular emphasis on their inability to preemptively address the complexities of automated trading strategies. It argues for the introduction of pre-deployment testing, real-time monitoring, and enhanced scrutiny to mitigate systemic vulnerabilities. Section V builds upon this foundation, advancing a regulatory framework that accounts for the specific characteristics of retail algo-trading in India.
*Abhishek Sanjay is a 2nd Year BA.LLB(Hons) at NALSAR University of Law. Tanya George is a 3rd Year BA.LLB(Hons) at MNLU, Mumbai.