Guide · Systematic and algo
Algorithmic trading in India: four layers, one word
The short answer
Algorithmic trading is any trade whose order is generated, or routed, by automated logic instead of a human hand. That single phrase covers four different activities sitting at four different layers: execution algos that work an already-made decision into the market, alpha algos that decide what to trade, high-frequency trading as a subset defined purely by latency, and retail automation through a broker API, which is what almost every Indian reader actually means. The layer you sit in decides which rules reach you, and the retail API layer is precisely the one the SEBI circular of 4 February 2025 was written to structure. Inside that layer, the part that decides whether you survive is not the strategy code. It is the risk layer, and it is a gate rather than an afterthought.
Most writing on this subject answers a question the reader did not ask. It lists languages and tools, then gestures at regulation as something "evolving", which tells you nothing you can act on. The regulation is not evolving in the abstract: it is a dated document with named duties, a delegated threshold and a specific accountability chain, and it decides what you are permitted to do. This guide fixes the taxonomy first, because the taxonomy is what tells you which box you are in. Then it reads the framework from the circular itself rather than from summaries of it, which turns out to matter, because the most widely repeated number about retail algo trading in India is not in the circular at all. Then it covers what a retail build actually involves. The uncomfortable part stays in view throughout: automation is an amplifier, and it amplifies a bad process faster than a good one.
What the words actually mean
The single biggest confusion in this field is that one word is doing four jobs. Getting the taxonomy right is not pedantry. It is the thing that tells you which regulatory box you sit in, which costs you face, and which of the arguments you read online were never about you in the first place.
An execution algo takes a decision that has already been made, buy this quantity, and works it into the market to reduce cost and impact. The classic form slices a large parent order into child orders scheduled against time or traded volume, so a fund buying a big line does not push the price against itself. Its logic is transparent and replicable, and it optimises how you trade, never what. An alpha algo, or strategy algo, does the opposite: it decides what to trade and when, encoding a hypothesis into rules that fire entries and exits. The two get bundled together in conversation and should not be, because an execution algo has no view and an alpha algo is nothing but a view. High-frequency trading is not a third species so much as execution taken to a latency extreme: very high order rates, holding periods measured in fractions of a second, and profitability that depends on sitting physically close to the matching engine. The fourth layer is the one Indian retail readers mean: a rule-based strategy running through a broker API, placing ordinary orders programmatically. It is not fast in the high-frequency sense, and its edge, if it has one, has nothing to do with latency.
The distinction is not academic, because each layer optimises a different problem. An execution algo exists inside a genuine trade-off: work an order too slowly and the price drifts away before you are filled, work it too fast and your own buying pushes the price up against you. Slicing schedules exist to balance timing risk against market impact, and they matter precisely when the order is large relative to what the instrument trades. A retail order of a few lots rarely moves the market, which is why execution algorithms buy an institution far more than they buy an individual, and why retail interest sits almost entirely on the alpha side: the rules that decide the trade, not the machinery that places it.
Four layers, four different rulebooks
Set the four side by side and the practical consequence becomes obvious. The infrastructure column is the honest one: it explains why three of these four conversations are simply not available to a retail account, no matter how good the code is. The rules column is the one that decides your obligations.
| Layer | What it optimises | Infrastructure it requires | Which rules reach you |
|---|---|---|---|
| Execution algos (order slicing) | How an already-made decision is worked into the market, balancing timing risk against market impact | An order large enough to move the price. Below that, slicing solves a problem you do not have. | The member's own algo permission from the exchange. The end client meets none of it directly. |
| High-frequency trading | Latency. Queue position measured in microseconds, holding periods in fractions of a second. | Colocation beside the matching engine, plus the engineering staff to keep it running | Exchange algo approval and colocation rules. A retail account is not in this tier at all. |
| Alpha algos, institutional | The research question: which rules carry a real, tested expectancy across a portfolio | A research desk, clean history, and a risk function independent of the person with the idea | Institutional access arrangements and the desk's own mandate and limits |
| Retail API automation | The same research question, on one account, with the execution reduced to an ordinary order | A broker API key, a static IP, and a machine that stays up during market hours | The February 2025 SEBI framework, in full. This is the layer it was written for. |
Three of these four layers are conversations about infrastructure you will never own. The fourth is a conversation about whether your idea is any good.
Read the table that way and something clarifying happens: most of what is written about algo trading is about the first two rows and is quietly irrelevant to the reader. The retail reader inherits none of the latency race and none of the slicing problem. What they inherit instead is the entire fourth row, and the fourth row is the only one with a rulebook addressed to them by name.
The tape is already majority automated
Before the rules, the scale, because the scale is what makes the rules make sense. Algorithmic order flow is no longer a minority of Indian trading. On figures attributed to NSE, algorithms crossed over to a majority of cash-market activity around 2024, reaching roughly 53 percent against about 14 percent as recently as 2010. Exchange data reported for the financial year 2026 to date puts the cash market near 54 percent and the equity derivatives segment near 67 percent. The consequence for anyone trading by hand is blunt: the counterparty to a large share of your orders is a machine, and the microstructure you trade in, the way spreads tighten and quotes flicker, is shaped by automated flow.
The reason institutional algos can operate at that intensity is colocation. Exchanges rent rack space inside or beside their own data centres, so a member's trading server sits metres from the matching engine rather than across a city, which removes the network latency that would otherwise dominate a high-frequency strategy. The tier is measurable: on figures attributed to NSE, colocation made up about 35.7 percent of cash-market turnover and roughly 62.1 percent of equity-derivatives turnover in 2024, against low single digits in 2010. Colocation is the price of admission to genuine high-frequency trading, and it is the clearest line between the institutional latency game and everything a retail trader does. A strategy running from a home connection or an ordinary cloud server is, by construction, not competing on speed.
Read the figures as direction, not decimals. Exchange participation shares are published per segment and per period, on definitions that differ between them, and they move month to month. The load-bearing fact here is not the second decimal place: it is that automation is now the majority of the tape and that colocation dominates the fastest slice. Figures cited here are attributed to NSE and were reported for 2024 and for the 2026 financial year to date. Treat any guide still calling algo trading a niche as a decade out of date, and check the exchange's current publications before quoting a number as today's.
The framework, read from the circular
Who is responsible for what
On 4 February 2025 SEBI issued circular SEBI/HO/MIRSD/MIRSD-PoD/P/CIR/2025/0000013, titled Safer participation of retail investors in Algorithmic trading. It does not ban retail algo trading. It replaces an ambiguous grey area, in which retail strategies quietly ran through broker APIs with no clear accountability, with a structure that names who answers for what. The exchanges then issued the operational modalities and FAQs that turn those principles into working rules.
The architecture rests on one organising idea, stated in the circular's own words: for the provision of algo trading through APIs, brokers shall be the principal while any algo provider or fintech vendor shall act as its agent. Accountability cannot leak out to an unregulated third party, because the regulated broker owns the pipe. Every algo order flowing through a broker API is tagged with a unique identifier provided by the stock exchange, so automated flow is distinguishable from manual flow and an audit trail exists all the way to the matching engine. API access itself is fenced: no open APIs, a unique vendor and client specific API key, a static IP whitelisted by the broker, OAuth-based authentication only with all other mechanisms discontinued, and two-factor authentication on access.
What ties the design together is traceability. Because every algo order carries an exchange-issued identifier and rides a whitelisted key from a known machine, the exchange and the broker can attribute automated flow to a specific strategy and a specific client, monitor it, and unwind it if it misbehaves. That is the point of the whole structure: not to make algo trading harder, but to make it attributable. It also reframes what an algo vendor is selling. Under this framework a provider is an empanelled agent operating under a regulated broker's licence, answerable for what it distributes, and its strategies are required to run on the broker's servers rather than its own.
What the circular actually says about your algo
Here is where reading the source pays, because the most quoted fact about retail algo trading in India is not quite what it is reported to be. The circular says that algos developed by tech-savvy retail investors themselves, using programming knowledge, shall also be registered with the Exchange, through their broker, only if they cross the specified order per second threshold. It then does something most summaries omit: it declines to say what that threshold is. A footnote records that the specified threshold shall be evolved by the Broker's Industry Standards Forum, under the aegis of the stock exchanges and in consultation with SEBI.
So the number does not live in the circular. It lives in the exchange-level implementation standards, where NSE's FAQ of 3 November 2025 refers to a threshold of 10 orders per second. The distinction is not trivia. It means the figure can change without SEBI issuing anything, and it means a page telling you "SEBI says 10 orders per second" has misidentified the source of the only number it gives you. The second omission is larger. Being under the threshold exempts you from registration, not from the framework: NSE's FAQ states that all orders received via API from clients are considered algo orders and require appropriate tagging, with a standardised tag used where the order rate sits within the threshold. You are tagged either way. You are registered only above the line.
| Requirement | Who bears it | Where it is written, and when |
|---|---|---|
| Broker is the principal; the algo provider is its agent | Broker, and the vendor | SEBI circular, 4 February 2025 |
| Every algo order carries a unique exchange identifier | Exchange issues it, broker applies it | SEBI circular, 4 February 2025 |
| Registration of a self-written algo above the order-per-second threshold | Retail trader, through the broker | SEBI circular, 4 February 2025. The threshold itself is delegated to the industry forum. |
| The threshold as a number, reported as 10 orders per second | Retail trader | Exchange implementation standards, not the circular. NSE FAQ, 3 November 2025. |
| Tagging below the threshold, using a standardised tag | Broker applies it to all API orders | NSE FAQ, 3 November 2025. Exemption from registration is not exemption from tagging. |
| Unique client-specific API key, whitelisted static IP, OAuth only, two-factor | Broker enforces it | SEBI circular, 4 February 2025 |
| A registered self-written algo may be used for family only | Retail trader | SEBI circular, 4 February 2025. Family means self, spouse, dependent children and dependent parents. |
| Provider strategies run on the broker's servers; the self-builder hosts their own | Provider, and the self-builder | NSE operational modalities, as summarised in the NSE FAQ of 3 November 2025 |
| Empanelment of algo providers with the exchanges | Algo provider | SEBI circular, 4 February 2025. Providers are not regulated by SEBI. |
| Research analyst registration and a research report per algo, for black box | Black box algo provider | SEBI circular, 4 February 2025 |
| Kill switch for orders from a particular algo ID | The exchange, not your code | SEBI circular, 4 February 2025, footnote |
| Framework applicable to all stock brokers | All stock brokers | SEBI extension circular, 30 September 2025, setting the date at 1 April 2026 |
Three further specifics are worth having, because they are checkable and they change what you build. First, the family rule: a registered self-written algo may be used by the investor for family, defined in the circular as self, spouse, dependent children and dependent parents, but not for other investors. Running your cousin's account, or a friend's, is outside it. Second, algo providers are not regulated by SEBI. The circular says so directly, and then requires them to be empanelled with the exchanges instead, with the broker performing due diligence before onboarding. A vendor advertising a "SEBI-approved algo" is describing a status that the circular does not create. The exception proves the rule: a black box provider must register as a Research Analyst, and that requirement exists because of the research report, not because the algo was blessed. Third, the hosting split: a provider's strategies must run on the broker's servers, with order messages originating there, while a tech-savvy investor hosts their own logic at their own end on a static IP.
Where this gets misread. Two errors dominate. The first is treating the order-rate threshold as a permission slip: it decides whether your algo must be registered, not whether the rules apply to you. The second is treating "empanelled" or "registered" as a quality signal. Neither status is an opinion about whether a strategy makes money, and the framework never claims otherwise. Its purpose is attribution: making sure a malfunctioning or manipulative algo can be traced to a source. A registered, tagged, fully compliant strategy with no edge is still a strategy with no edge, and the framework has nothing to say about that.
The dates, and why they keep moving
The calendar matters as much as the rules, mostly as a lesson in how much this regime moves. The February 2025 circular set its own provisions to apply from 1 August 2025, and required the implementation standards to be formulated by the Broker's Industry Standards Forum by 1 April 2025. That date slipped to 1 October 2025 by a circular dated 29 July 2025. Then SEBI circular SEBI/HO/MIRSD/MIRSD-PoD/P/CIR/2025/132, dated 30 September 2025, replaced a single date with a glide path: brokers with ready systems could go live from 1 October 2025; a broker had to apply to register at least one retail algo product and strategy by 31 October 2025, complete that registration by 30 November 2025, and take part in at least one mock session by 3 January 2026. Brokers missing those milestones were to be barred from onboarding new retail clients for API-based algo trading from 5 January 2026. The full framework, with the exchanges' operational modalities, became applicable to all stock brokers from 1 April 2026.
That is four dates for one rule in fourteen months, and the movement did not stop when the framework landed. SEBI revised the order-to-trade ratio framework for algorithmic orders with effect from 6 April 2026, exempting orders placed within a narrow band of the last traded price from the high order-to-trade penalty regime. That particular change is reported as having little direct effect on a retail algo trader's day, which is the honest thing to say about it, but it makes the general point better than any warning could: this is a live regime, adjusted repeatedly, and its details are not a thing to learn once.
Date every rule you rely on, and then check it. Everything on this page is stated as of 17 July 2026 and is drawn from the SEBI circulars of 4 February 2025 and 30 September 2025 and from NSE's FAQ of 3 November 2025, all linked in the Sources below. The framework applies to all stock brokers from 1 April 2026, and we found no later extension of that date, but rules in this area have been revised four times in fourteen months and the one number most readers want, the order-rate threshold, sits in exchange implementation standards that can change without a new SEBI circular. Before you act on any specific in this guide, confirm the current position on the SEBI website and with your own broker, who is the party that must actually apply it to your account.
The risk layer is the gate, not an afterthought
Strip away the marketing and a retail trading algo is a short, disciplined pipeline. Data feeds a signal module that applies the strategy rules; a valid signal passes to a risk layer that decides whether, and how large, an order may be; only if it clears does the order reach execution through the broker API and on to the exchange; and every step writes to logging so the system can be reconstructed. The engineering interest is not in the strategy code, which is usually the shortest file in the project. It is in the two ends that beginners skimp: the risk layer and the logging.
The risk layer is non-negotiable, and it is what separates a durable system from a fast way to lose. At minimum it enforces hard position caps and per-trade and daily loss limits, validates every order against available margin, and handles the errors that live systems actually throw: rejected orders, partial fills, a dropped data feed, a stale price. Above all it holds a kill switch: an automatic halt on a pre-defined condition that stops the strategy and pulls its resting orders. Here the wording of the circular is worth being precise about, because it is routinely stretched. SEBI's footnote defines the kill switch as an emergency function and the last level of defence against any Algorithm malfunction, expected to automatically trigger a halt on trading activity based on pre-defined conditions. But the kill switch the circular assigns is the exchange's, which is to retain the ability to use it for orders emanating from a particular algo ID. That is a backstop protecting the market from you. The kill switch in your own risk layer is a different object protecting you from your own software, and no regulation will build it for you.
| Stage | Purpose | Failure mode if neglected | The test that catches it |
|---|---|---|---|
| Data | Feed clean live and historical prices to the strategy, correctly time-stamped | Bad ticks and survivorship bias corrupt every decision downstream, invisibly | Reconcile your history against the exchange's own published data before you trust a backtest built on it |
| Signal | Apply the strategy rules to produce entries and exits | An untested rule fires confidently on noise, now at machine speed and without hesitation | Out-of-sample testing on data the rule was never fitted to, and a paper run before real size |
| Risk layer | Gate every order against position caps, loss limits and available margin | One runaway loop or fat-finger size drains the account faster than you can watch it | Deliberately feed it invalid orders and confirm each is refused, not merely warned about |
| Execution | Place, modify and cancel orders through the broker API | Unhandled rejections and partial fills leave phantom or doubled positions | Force the error paths: reject an order, kill the feed mid-fill, and see what the system believes it holds |
| Logging | Record every action and surface breaches in real time | A failure you cannot reconstruct is a failure you are going to repeat | Reconstruct yesterday's session from the log alone. If you cannot, the log is decoration. |
Automation scales the process, it does not create one
Automation is sold as a route to hands-off income. The economics say something duller. There are real, recurring costs before a single rupee of profit: data subscriptions, API access charges, a server or cloud instance that must stay up during market hours, and the ordinary transaction costs of Indian trading, brokerage, exchange charges, securities transaction tax, goods and services tax and stamp duty, which a backtest often quietly ignores. Then there is slippage, the gap between the price a signal assumed and the price the market actually gave, which widens in thin instruments and fast markets and is fully capable of turning a marginal backtest into a live loss. None of this is exotic. All of it is routinely left out of the story.
The deeper point is structural, and it is the one worth leaving with. Automation does not add edge; it scales whatever process you feed it. Wire a method with genuine, tested expectancy to a well-built system and you get that expectancy applied consistently, without the hesitation and rule-breaking that erode discretionary results. Wire a method with no real edge to the same system and you get the loss rate applied just as consistently, only faster and without the flinch that might have slowed it down. The machine is an amplifier with no opinion about the sign of what it amplifies.
Which makes the prerequisite work non-optional rather than merely advisable. You need the method built as a defined set of rules, and validated honestly, before automation is anything other than a liability. Two beliefs are worth killing on the way. Automation is not alpha: a bot is an execution engine and contains no view the strategy did not give it, so running a rule with no edge automatically does not create one, it only removes the friction that was slowing the bleeding. And speed is not the retail edge: the only participants for whom raw latency pays are colocated beside the matching engine, a tier measured at roughly 35.7 percent of NSE cash turnover in 2024, and a strategy on a home connection is not in that race and never will be. The base rate underneath all of this deserves stating plainly, because it is the number that should set the burden of proof: about 93% of individual traders in equity derivatives made net losses over FY22 to FY24, aggregate net losses exceeding ₹1.8 lakh crore (SEBI, September 2024). Automating a strategy does not move those odds. It decides how quickly and how faithfully your version of them plays out. That upstream discipline, the rules, the evidence and the risk budget, is exactly what the method we teach is built around, and it is the reason the code comes last here rather than first.
Where this fits, and where it does not
The honest limits of this guide are worth stating, because a page that claims to settle a live regulatory question is lying to you. This is a map of the layers and a close reading of a framework as it stood on 17 July 2026. It is not legal advice, it is not a compliance opinion, and it cannot tell you what your own broker will permit, which is a separate question with a different answer at each broker. The framework sets a floor; a representative Indian retail broker may sit well above it, restricting order rates, instruments or strategy types for reasons of its own, and it is entitled to, because it carries the accountability. The only authoritative answers to "may I do this" are the current SEBI circulars, the exchange's operational modalities, and your broker's own published terms.
What this guide can settle is the part that does not move. The taxonomy is stable: execution, alpha, latency and retail API automation will still be four different things after the next circular, and the layer you sit in will still decide which arguments are about you. The engineering is stable: a risk layer that refuses invalid orders and a log you can reconstruct a session from will be correct whatever the threshold becomes. And the economics are stable: automation will still be an amplifier, and the sign of what it amplifies will still be decided before any of it is written. Those three survive the regime. The dates and the thresholds do not, which is precisely why they are dated and sourced here rather than asserted, and why the last step of any plan built on this page should be to check them again.
What to do with this page. Use the taxonomy to work out which of the four layers you are actually in, which for almost every reader is the fourth. Use the requirement table to know what your broker is obliged to apply to you, then confirm the current specifics with SEBI and with that broker, since they are the ones who must implement it on your account. Use the pipeline to design the build, and treat the risk layer and the log as the first things you write rather than the last. And treat the last section as the order of work: the strategy, then the evidence, then the risk budget, and only then the API key.
Common Questions
Frequently Asked Questions
What is algorithmic trading?
+Algorithmic trading is any trade whose order is generated, or routed, by automated logic rather than a person clicking a button. SEBI's own working definition, carried since its 2012 guidelines, is orders generated using automated execution logic. The word covers four different activities that are worth separating. Execution algos take a decision already made and work the order into the market to reduce cost and impact. Alpha algos decide what to trade and when. High-frequency trading is execution taken to a latency extreme, defined by speed rather than by being a separate species. The fourth layer is what most Indian retail readers mean: a rule-based strategy placing ordinary orders through a broker API.
Is algorithmic trading legal for retail investors in India?
+Yes. Retail algo trading is permitted, and since the SEBI circular dated 4 February 2025 titled Safer participation of retail investors in Algorithmic trading it is formally structured rather than merely tolerated. The framework routes retail automation through a registered broker, requires exchange tagging of algo orders, and places defined duties on brokers, algo providers and exchanges. It is legal within that structure. Being legal is not the same as being safe or being suitable, and the framework itself makes no claim about outcomes. As of July 2026 this framework applies to all stock brokers, but rules in this area have moved repeatedly, so confirm the current position with SEBI and with your broker before relying on it.
Do I need approval to run my own algo in India?
+It depends on order rate, and the detail matters. The February 2025 circular says that algos developed by tech-savvy retail investors themselves shall be registered with the exchange, through their broker, only if they cross a specified order per second threshold. The circular does not state that number. It records that the threshold shall be evolved by the Broker's Industry Standards Forum, under the aegis of the stock exchanges and in consultation with SEBI. The number lives in the exchange-level implementation standards, where NSE's FAQ of 3 November 2025 refers to a threshold of 10 orders per second. Below the threshold you are exempt from registration, not from the framework: NSE states that all orders received via API from clients are considered algo orders and require appropriate tagging. Confirm the current threshold with your broker, since it can move without SEBI issuing a new circular.
What is colocation, and does it matter to a retail trader?
+Colocation is renting rack space inside, or immediately beside, the exchange data centre so that a trading server sits metres from the matching engine rather than across a city. The point is latency: removing the milliseconds a signal would otherwise spend travelling. It matters to a retail trader mainly as a boundary marker. On figures attributed to NSE, colocation accounted for about 35.7 percent of cash-market turnover and about 62.1 percent of equity-derivatives turnover in 2024. That is the tier where raw speed pays. A strategy running from a home connection or an ordinary cloud server is not in that race by construction, which is the useful conclusion: speed cannot be a retail edge, so the idea and the risk control have to be.
What share of Indian market volume is algorithmic?
+A majority, and it crossed over recently. On figures attributed to NSE, algorithmic trading reached about 53 percent of cash-market activity in 2024, overtaking manual trading for the first time, against roughly 14 percent in 2010. Exchange data reported for the financial year 2026 to date puts the cash market near 54 percent and equity derivatives near 67 percent. Read these as direction rather than as precise constants: participation shares are reported per segment and per period and they move. The load-bearing fact is not the decimal place. It is that a large share of the order flow you trade against is already automated, and that any guide still describing algo trading as a niche is a decade out of date.
Does automating a strategy make it profitable?
+No. An algorithm is an execution engine for a strategy, not a source of edge. It contains no view that the rules did not already give it. Automation removes hesitation and rule-breaking, which means it applies whatever expectancy you wrote, including a negative one, more consistently and more quickly than you would by hand. The base rate is worth stating plainly: about 93% of individual traders in equity derivatives made net losses over FY22 to FY24, aggregate net losses exceeding ₹1.8 lakh crore (SEBI, September 2024). Automation does not move those odds. It decides how fast and how faithfully your version of them plays out, which is why the edge has to exist before the code does.
What is a kill switch in algo trading?
+SEBI's February 2025 circular defines it in a footnote: the kill switch is an emergency function and the last level of defence against any Algorithm malfunction, and it is expected to automatically trigger a halt on trading activity based on pre-defined conditions. Read the circular carefully and the kill switch it describes belongs to the exchanges, which are to continue to have the ability to use it for orders emanating from a particular algo id. That is a regulatory backstop, not a feature of your code. The kill switch in your own risk layer is a separate object: an automatic halt on a condition you defined, which stops the strategy and pulls its resting orders. Both are worth having, and they are not the same thing. The exchange one protects the market from you. Yours protects you from your own software.
What is the difference between white box and black box algos?
+The February 2025 circular categorises algos into two kinds. White box algos, which it equates with execution algos, are those where the logic is disclosed and replicable, so the user can see and reproduce the decision rules. Black box algos are those where the logic is not known to the user and is not replicable. The distinction carries real weight, because opacity moves the risk onto the subscriber. For black box algos the provider must register as a Research Analyst, maintain a detailed research report for each such algo, and confirm to the exchanges that the report has been maintained. If the logic governing the algo changes, it must be registered as a fresh algo with a new report. In practice, if you cannot see the rules, you are buying an opinion rather than a tool.
Where the facts come from
Sources
- The framework itself. Securities and Exchange Board of India, circular SEBI/HO/MIRSD/MIRSD-PoD/P/CIR/2025/0000013 dated 4 February 2025, Safer participation of retail investors in Algorithmic trading. Source for the broker-as-principal and provider-as-agent model, exchange tagging of every algo order, the API key, static IP, OAuth and two-factor controls, the white box and black box categorisation with research analyst registration for black box providers, the family definition, the statement that algo providers are not regulated by SEBI, the kill switch footnote, and the original applicability date of 1 August 2025. sebi.gov.in
- The timeline. Securities and Exchange Board of India, circular SEBI/HO/MIRSD/MIRSD-PoD/P/CIR/2025/132 dated 30 September 2025, extending the implementation timeline and setting the glide path: go-live for ready brokers from 1 October 2025, registration milestones through October and November 2025, a mock session by 3 January 2026, and applicability to all stock brokers from 1 April 2026. The interim extension from 1 August to 1 October 2025 was made by a circular dated 29 July 2025. sebi.gov.in
- The operational detail, including the threshold number. National Stock Exchange of India, Frequently Asked Questions on Safer participation of Retail investors in Algorithmic trading, dated 3 November 2025. Source for the 10 orders per second threshold, for the statement that all orders received via API from clients are considered algo orders and require appropriate tagging, for strategies being hosted on brokers' servers while a tech-savvy client hosts their own logic on a static IP, and for static IP being required only for a tech-savvy investor using an API. nsearchives.nseindia.com
- Participation and colocation shares. Figures attributed to NSE and reported in October 2025: colocation at about 35.7 percent of cash-market turnover in 2024 against 3.1 percent in 2010, and about 62.1 percent of equity-derivatives turnover against 7.3 percent in 2010; algorithmic trading at about 53 percent of the cash market in 2024 against about 14 percent in 2010. Figures for the 2026 financial year to date, near 54 percent in the cash market and 67 percent in equity derivatives, are drawn from NSE's own market publications as reported. These are secondary reports of exchange data; check the exchange's current publications before quoting any of them as today's figure. in.investing.com
- The retail base rate. Securities and Exchange Board of India, study of profit and loss of individual traders dealing in the equity derivatives segment, September 2024, the source of the finding cited here on net losses across FY22 to FY24. sebi.gov.in
- On what we could not verify directly. The full text of the February 2025 circular was read from a stock exchange's published reproduction of it, and the order-to-trade ratio revision effective 6 April 2026 is taken from reporting dated February 2026 rather than from the circular itself. Where a figure could not be traced to a primary document, it is described here as reported or attributed rather than stated as fact, and the order-rate threshold is presented as an exchange implementation standard, which is what the SEBI circular says it is.
Educational note. This guide explains what algorithmic trading is and how the Indian regulatory framework treats it, as of 17 July 2026. It is not a recommendation to trade, to automate, or to buy or sell any security, it is not legal or compliance advice, and it is not investment advice. Regulatory specifics change: verify the current position with SEBI, the exchange and your broker before acting on any of it. Trading in leveraged products carries a high risk of loss. Bharath Shiksha is an educational publisher, not a SEBI-registered investment adviser or research analyst.