The figures tell a consistent story. Co-located servers' share of equity futures turnover rose from 33% in FY15 to 51% in FY26. In FY26, 99% of the gross profits of proprietary traders and foreign portfolio investors in equity derivatives came from entities using algos, while 87.7% of individual traders lost money. Retail participation in futures and options fell in the same year. The report uses published data from NSE and SEBI only; QuantPower has not run a study of its own, and none of the figures are QuantPower results.
Algo trading in India at a glance
| Measure | Figure | Source |
|---|---|---|
| Algo share of NSE equity derivatives turnover | About 67%, FY26 to November 2025 | NSE data |
| Algo share of NSE cash market turnover | 54%, FY26 to November 2025 | NSE data |
| Co-location share of equity futures turnover | 33% in FY15, 51% in FY26 | NSE corporate presentation, May 2026 |
| Share of prop and FPI gross derivatives profits from algo entities | 99%, FY26 | SEBI FY26 study |
| Individual traders with a net loss | 87.7%, FY26 | SEBI FY26 study |
| Aggregate net loss of individuals | ₹91,685 crore, FY26 | SEBI FY26 study |
| Retail algo framework | In force for every broker since 1 April 2026 | SEBI |
How this report was put together
Every figure comes from a public source, named beside it and listed in the references. Algo shares of turnover are NSE data as reported by NiftyTrader for FY25 and in QuantInsti's release of 16 January 2026 for FY26 to November 2025.12 Co-location shares are from NSE's corporate presentation of May 2026.3 Shares of turnover by client category are from NSE's Market Pulse.4 Profit, loss, participation and cost figures are from SEBI's studies of individual traders in equity derivatives for FY25 and FY26, published on 20 August 2026.5 Figures were checked on 20 October 2026. Where a figure is not published, such as the share of turnover from retail algos, the report says so.
A short history of algo trading in India
Algo trading in India began with institutions. In 2008 SEBI allowed direct market access, which let institutional clients send orders to the exchange through a broker's systems without a dealer keying them in.6 Exchanges then offered co-location, where a member's servers sit in the exchange's own data centre, cutting the time an order takes to arrive. In March 2012 SEBI published its first broad guidelines on algorithmic trading: exchanges had to put risk controls in place, and brokers needed the exchange's approval before offering algo trading.7
The next decade brought rapid growth and some controversy. Co-location became central to high-frequency trading, and questions about fair access to NSE's co-location systems led to a long investigation and a SEBI order against the exchange in 2019. In August 2016 SEBI published a discussion paper on strengthening the framework for algo trading and co-location. Meanwhile retail traders found their own route to automation as brokers opened APIs and third-party platforms connected to them.
That route had no specific rules until SEBI turned to it. A consultation paper on algo trading by retail investors in December 2021 set out the problem: the exchange could not tell which retail orders came from algos, who had built the strategy or where an order came from. SEBI's circular of 4 February 2025 answered it with a full framework: algo orders tagged with an exchange identifier, algo providers empanelled through brokers, black-box strategies tied to registered research analysts, and API access tied to registered static IPs.8 After two postponements, it came into force for every broker from 1 April 2026.9
Algo share of NSE turnover
| Measure | FY15 | FY20 | FY25 | FY26 |
|---|---|---|---|---|
| Algo share, equity derivatives | — | — | About 70% | About 67% (to November) |
| Algo share, cash market | — | — | 54% | 54% (to November) |
| Co-location share, equity futures | 33% | 34% | 50% | 51% |
| Co-location share, equity options (premium) | 46% | 49% | 53% | 53% |
| Co-location share, cash market | 22% | 32% | 37% | 39% |
The pattern is consistent across segments: the co-location share has risen in every segment since FY15, fastest in equity futures and the cash market. Equity options rose least because they were already heavily automated: market makers quote hundreds of strikes at once, which is only possible with automation.
How to read these figures
Two measures appear in the table and they are not the same.
- Algo share counts orders placed by algorithms, wherever the server sits: in the exchange's data centre, at a broker, or at a platform. It is the broader measure.
- Co-location share counts only orders from servers inside the exchange's data centre, which is where the fastest trading happens. It is a subset of algo trading.
Turnover in options can be measured on the premium paid or on the notional value of the contracts. Premium turnover is much smaller and better reflects the money changing hands; the co-location figures for options in this report are on premium. And the FY26 algo shares cover the year only to November 2025, so the full-year figures may differ.
Who the algo traders are
| Participant | Role in algo trading | Evidence |
|---|---|---|
| Proprietary trading firms | Market making, arbitrage and high-frequency strategies from co-located servers, mostly with their own capital | 50.7% of NSE equity options premium turnover and 31.9% of equity futures turnover in FY26 |
| Foreign portfolio investors | Systematic and arbitrage strategies, mostly through algos | With proprietary traders, 99% of their FY26 gross profits in equity derivatives came from algo entities |
| Domestic institutions | Execution algos for large orders | Share not separately published |
| Individuals | Rule-based strategies through broker APIs and platforms | Covered by SEBI's retail algo framework since 1 April 2026; share not separately published |
Co-location is the infrastructure behind most of this. Servers in NSE's data centre handled 51% of equity futures turnover and 53% of equity options premium turnover in FY26, according to NSE. Our guide to high-frequency trading in India explains how it works and why it is out of reach for individuals.
Why institutions dominate
Three advantages explain why proprietary firms and foreign investors account for most algo activity:
- Speed. Co-located servers receive prices and send orders in microseconds. Strategies that depend on being first, such as arbitrage between related contracts or quoting both sides of the market, need that speed.
- Cost. Large firms pay much lower costs per order than individuals, and market makers can earn the spread between buying and selling prices rather than paying it.
- Scale and data. Firms run many strategies across thousands of contracts, with research teams and data that individuals do not have.
Individual traders cannot match these advantages, and do not need to. The strategies open to them work on rules that hold over minutes, hours or days, where speed matters less than discipline and cost control. A retail strategy that tries to compete on speed, such as reacting to price changes within the same second, is competing on the institutions' ground and will usually lose there. A strategy that holds positions for an hour or a day, with rules tested over years of data, competes on different ground, where a few milliseconds make no difference and where careful sizing, costs and exits matter most.
Where the profits went: SEBI's FY26 study
SEBI's study of equity derivatives for FY25 and FY26, published on 20 August 2026, gives the clearest picture of who earns from algo trading.5
| Participant | FY26 outcome |
|---|---|
| Proprietary traders | Gross trading profit of about ₹44,000 crore |
| Foreign portfolio investors | Gross trading profit of about ₹14,000 crore |
| Individual traders | Net loss of ₹91,685 crore after costs; 87.7% lost money |
SEBI reported that 99% of the gross profits of FPIs and proprietary traders came from entities using algorithmic trading. Individuals paid about ₹25,000 crore in transaction costs in each of FY25 and FY26, and about ₹1 lakh crore across FY22 to FY26.
What the profit figures do and do not show
The figures show that, as groups, institutions using algos earned and individuals lost, and they show it consistently: SEBI's earlier studies of FY22 to FY24 found the same broad pattern. They do not show that using an algo makes an individual profitable: the institutions' advantages are speed, cost and scale, not the fact of automation itself. Nor do they split individual results between those who used algos and those who traded by hand, because that split was not published. What they do show is the size of the costs individuals pay: about ₹25,000 crore a year, a large share of the total loss. Any strategy, automated or not, has to earn more than its costs before it earns anything.
Retail participation in derivatives is falling
| Measure | FY25 | FY26 |
|---|---|---|
| Active individual F&O traders | 98.1 lakh to 106.2 lakh, depending on the measure | 78.6 lakh to 87.5 lakh |
| New individual F&O traders | 34.3 lakh | 20.8 lakh |
| Individual traders making a net loss | 90.9% | 87.7% |
| Aggregate net loss of individuals | ₹1,11,788 crore | ₹91,685 crore |
NSE's active client base across all segments fell about 7% in FY26, from 4.92 crore to 4.57 crore, according to NSE data reported by IANS in April 2026.10 The decline followed SEBI's October 2024 measures on index derivatives: larger contracts, one weekly expiry per exchange and upfront premium collection.11
Proprietary and individual shares of turnover
NSE's Market Pulse shows proprietary traders' share of equity options premium turnover rising from 49% in FY25 to 50.7% in FY26, while their share of equity futures turnover fell from 34% to 31.9%. Individuals' share of options premium turnover rose from 35% to 37.5%, and their share of cash market turnover was 33.4% in FY26, against 34% in FY25.4
The rise in individuals' share of options premium while their numbers fell suggests that the traders who stayed traded more, or traded larger, than before. The published data do not say which. Proprietary traders' rising share of options premium, alongside their falling share of futures, fits the picture of firms concentrating on option market making and arbitrage, where automation and speed matter most. Both trends are worth following in the FY27 figures, when the effect of a full year of the retail algo framework can be seen.
How SEBI's 2024 measures changed the market
SEBI's circular of 1 October 2024 set out measures for index derivatives that came into force in phases.11 They shaped the FY26 figures in this report:
| Measure | From | What it changed |
|---|---|---|
| Larger minimum contract size for index derivatives | 20 November 2024 | Each lot became larger in rupee terms, raising the money needed to trade |
| One weekly expiry per exchange | 20 November 2024 | Weekly options ended for Bank Nifty, FinNifty and Midcap Select; Nifty 50 and Sensex kept them |
| Extra 2% exposure margin on short index options on expiry day | 20 November 2024 | Selling options on expiry day became more expensive in margin |
| Upfront collection of option premium | 1 February 2025 | Buyers had to have the full premium in the account before buying |
| No calendar-spread margin benefit on expiry day | 1 February 2025 | Positions across expiries lost their offset on the expiry day |
| Intraday monitoring of position limits | 1 April 2025 | Exchanges checked positions during the day, not only at the close |
The fall in new and active individual traders in FY26 came after these measures. The published data do not separate their effect from other causes, such as market conditions, but the timing is close.
How retail algo access works under the framework
Under the 2025 framework, an algo order from an individual passes through a set chain:
- The trader chooses a strategy: their own rules, a white-box strategy from a provider, or a black-box strategy from a registered research analyst.
- The platform or provider, if one is involved, is empanelled with the exchanges through a broker and acts as the broker's agent.
- The broker receives the order through its API from a registered static IP, checks it, tags it with the exchange's Algo ID where the algo is registered, and is responsible for it.
- The exchange receives the tagged order, so every algo order can be traced to its strategy and its broker.
Two limits apply throughout. A trader's own code that sends fewer than 10 orders a second does not need separate registration; above that, it must be registered with the exchange through the broker. And brokers must be able to stop algo orders on a client's account at once, a kill switch held by the broker.
Common misreadings of algo statistics
- "Two-thirds of trading is algo, so algos make money." The share of turnover says who places orders, not who profits. Most of the turnover comes from market makers and arbitrage firms whose edge is speed and cost.
- "Individuals lose because they do not use algos." The data do not split individual results by method. Costs and strategy choice explain much of the loss whatever the method.
- "Algo share fell in FY26, so algo trading is shrinking." The FY26 figure covers the year only to November 2025, and the share moves with the mix of participants. The co-location share kept rising.
- "Retail algo turnover is a known figure." It is not yet published. The framework's identifiers should make it measurable from FY27.
What SEBI's 2025 framework changes
SEBI's circular of 4 February 2025, in full effect since 1 April 2026, set out a full framework for retail algo trading.8
- Algos offered to retail traders must be registered with the exchange and carry an Algo ID on every order.
- Algo providers must be empanelled with the exchanges and act as agents of the brokers; brokers are responsible for every algo order.
- Traders' own code that stays below the exchange threshold of 10 orders a second does not need separate algo registration.
- API orders need a registered static IP, two-factor login and a daily logout.
- Brokers may not advertise the past or expected returns of any algo.
The framework makes retail algo activity traceable, so future NSE and SEBI reports may be able to separate retail algo turnover from institutional algo turnover, which published data does not yet do. Our guide to SEBI's algo trading rules covers the framework.
Questions to ask an algo provider
| Question | What a good answer looks like |
|---|---|
| Is the platform empanelled with the exchanges through a broker? | Yes, with the broker named |
| Can I see every rule the strategy trades? | Yes, for a white-box strategy; otherwise, a SEBI-registered research analyst stands behind it |
| Does it show past or promised returns? | No: brokers may not advertise algo returns |
| Where do my orders come from? | My broker's official API, from a static IP registered with the broker |
| Who holds my money? | My broker; the platform never does |
| What stops a runaway strategy? | Account-level loss limits and a kill switch, at the platform and at the broker |
What to watch in the data from FY27
With every retail algo order tagged from April 2026, the coming reports may answer questions that the data cannot yet answer:
- How much of retail turnover comes from algos, and in which segments?
- Do individuals who use algos lose less, or more, than those who trade by hand, after costs?
- How many retail strategies are white-box rules and how many are black-box products from research analysts?
- Does the order-rate threshold of 10 a second shape how retail algos trade?
What individuals can learn from how institutions run algos
Individuals cannot copy institutions' speed, but they can copy their habits:
- Risk limits sit above every strategy. Trading firms set limits on loss, position size and order rate for the whole desk, enforced by software, not by a trader's mood on the day.
- Every strategy is tested before it trades, and its live results are compared with the test.
- Costs are measured per trade. A strategy that does not cover its costs is switched off, however clever it is.
- Strategies are monitored and retired. An edge that fades is noticed in the numbers and the strategy stops, rather than being kept for hope.
- Nothing depends on one person watching a screen. Systems stop themselves when something is wrong, and refuse orders rather than guess.
A retail trader can apply all five with ordinary tools: a daily loss limit, a backtest and a forward test, a cost calculation, a monthly review and a kill switch.
Terms used in this report
| Term | Meaning |
|---|---|
| Algo trading | Trading in which software places the orders by preset rules |
| Co-location | Servers placed in the exchange's own data centre, for the fastest access |
| High-frequency trading | Algo trading that places very many orders with very short holding times, usually from co-location |
| Proprietary trader | A firm trading its own money, not clients' |
| FPI | A foreign portfolio investor registered with SEBI |
| Premium turnover | The value of option premiums traded, as opposed to the notional value of the contracts |
| Algo ID | The identifier the exchange gives a registered algo, carried on each of its orders |
| Empanelled provider | An algo provider approved by the exchange, acting as a broker's agent |
| White-box strategy | A strategy whose logic is disclosed to the user and can be replicated |
| Black-box strategy | A strategy whose logic is not disclosed; needs a registered research analyst |
A worked example: what costs mean for a retail algo
This is a hypothetical example.
An individual runs an option-selling algo that places 8 orders a day, on 22 trading days a month: 176 orders. At a broker that charges ₹20 an executed order, brokerage alone is ₹3,520 a month, before GST, STT, exchange charges and stamp duty. Over a year that is about ₹42,000 in brokerage for one strategy at one lot.
If the strategy's average result is ₹300 a day before costs, about ₹6,600 a month, brokerage takes more than half of it before any tax or charge. The same strategy run less often, with fewer re-entries and adjustments, may keep more of its result even if it catches fewer moves. This is the arithmetic behind the ₹25,000 crore a year that individuals pay in costs: automation does not lower costs per order, and a strategy that trades often can spend its edge on them.
What this means for individual traders
- Speed is not the retail edge. Co-located firms react in microseconds; retail strategies work on rules that hold over minutes to days.
- Costs decide outcomes. Individuals paid about ₹25,000 crore in transaction costs in FY26. Fewer, better-planned orders and limit orders cut the bill.
- Rules reduce impulse trades. Automated strategies with fixed entries, exits and an account-level loss limit follow the same rules every time.
- Test first. A backtest on past data and a forward test on live prices show how a strategy behaves before real money is at risk.
- Check the provider. Under the 2025 framework, a strategy with hidden logic needs a registered research analyst behind it, and no broker may advertise an algo's returns.
- Size for the bad month. Decide the largest loss you would accept in a month before the strategy goes live, and set the account's limits to enforce it.
Running rule-based strategies on QuantPower
QuantPower is software for traders who want to run their own rules; our page on algo trading for retail traders describes what it does. It connects to 100+ Indian brokers through their official APIs, includes two static IPs with every account for the registered-IP rule, and runs algos whose every rule is set by the trader and visible to them. Account-level controls sit above every algo: Smart Exit squares off every position and stops the algos at a set loss or profit for the whole account, the daily loss limit refuses new orders once the day's loss reaches it, and a kill switch stops a segment until its next session. A live algo costs 200 tokens per 28 days, with orders inside it unlimited (1 token = ₹1, GST extra).
Where algo trading in India is going
Three trends are visible in the data. Institutional automation keeps growing: the co-location share has risen in every segment for a decade, and nothing in the rules is likely to reverse that. Retail participation has fallen after SEBI's 2024 measures, and the traders who remain trade more of the options premium than before. And retail automation has moved from an unregulated corner of the market to a framework with identifiers, empanelled providers and registered IPs. The next few years of data will show, for the first time, how retail algos actually perform under that framework. For individual traders, the lasting lesson of the figures in this report is not about technology. It is that costs, discipline and risk limits decide results more than the choice between manual and automated trading, and that any edge has to be measured after costs, on data the strategy was not built on, before money follows it.
Start with 500 free tokens
Connect your broker, build a strategy and test it on paper before a single live order goes out.
Frequently asked questions
What percentage of trading in India is algorithmic?
About two-thirds of NSE equity derivatives turnover and 54% of cash market turnover in FY26 to November 2025, according to NSE data.
Who makes money from algo trading in India?
SEBI's FY26 study found that proprietary traders earned about ₹44,000 crore and FPIs about ₹14,000 crore in gross profits in equity derivatives, with 99% of those profits coming from algo entities.
Is retail algo trading growing?
Retail F&O participation fell in FY26, with active individual traders down 18% on SEBI's main measure. Retail algo use is not separately published, but SEBI's 2025 framework now tags every retail algo order, so future data should show it.
Is algo trading legal for retail traders?
Yes, through a SEBI-registered broker, using algos registered with the exchange or your own code below the order-rate threshold.
Do I need to register my own algo with the exchange?
Not if it is your own code running on your own account and sends fewer than 10 orders a second; it still goes through your broker's API from a registered static IP. Above that rate, it must be registered with the exchange through your broker.
What is the difference between algo trading and co-location?
Algo trading is any trading where software places the orders. Co-location means the trader's servers sit inside the exchange's data centre, which makes orders arrive fastest. Most co-located trading is algorithmic, but most algo trading by individuals runs far from the exchange, through broker APIs.
Related guides
- Expiry day volume analysis: where index option turnover concentrates on expiry days
- What is algo trading?: how automated trading works in India
- Best broker for algo trading: how brokers compare for automated trading
References
- NSE data on algo share of turnover, FY25, as reported by NiftyTrader, 2 June 2025
- NSE data on algo share of turnover, FY26 to November 2025, as cited in QuantInsti's press release of 16 January 2026
- NSE, corporate presentation, May 2026: co-location share of turnover by segment, FY15 to FY26 (nseindia.com)
- NSE, Market Pulse, April 2025 and April 2026: client category shares of turnover (nseindia.com)
- SEBI, studies on profitability and trading behaviour of individual traders in the equity derivatives segment, FY25–FY26, press release, 20 August 2026 (sebi.gov.in); figures also reported by Business Today, 21 August 2026
- SEBI, circular on direct market access, 3 April 2008 (sebi.gov.in)
- SEBI, circular "Broad Guidelines on Algorithmic Trading", 30 March 2012 (sebi.gov.in)
- SEBI, circular on safer participation of retail investors in algorithmic trading, 4 February 2025 (sebi.gov.in)
- SEBI, circular extending the timeline for implementation, 30 September 2025, with the framework in force for every broker from 1 April 2026 (sebi.gov.in)
- IANS, "NSE's active investor base tanks 7 pc in FY26", 14 April 2026 (ianslive.in)
- SEBI, circular "Measures to strengthen Equity Index Derivatives framework for increased investor protection and market stability", 1 October 2024 (sebi.gov.in)