Broker Responsibility for Algos Routed Through Its Platform
Blog by Aliceblue News Desk
13 November 2026
8 minutes
You download an algo app, link your trading account, and orders start going out on their own. A week later, one trade looks wrong. Who do you call: the app, your broker, or nobody?
For algo orders that come through a broker’s API, SEBI treats the broker as the principal and the algo provider as its agent. So your broker answers for how the algo runs on its platform. What the strategy earns or loses is still yours. This guide explains where each side’s duty starts and stops.
Quick Answer
Under SEBI’s February 2025 circular, the broker is the principal for algo trading through APIs, and the algo provider acts as its agent. The broker keeps risk management on its side and handles complaints. Profit or loss from the strategy stays with the client.
Table of Contents
Table of Contents
What Is Broker Responsibility for Algos Routed Through Its Platform?
It is your broker’s accountability for every algo order that reaches the exchange through its systems. SEBI set the principle in a circular dated 4 February 2025. From 1 April 2026, the framework applies to all stock brokers.
A few terms, in plain words:
- Algo: a set of rules that decides when to buy or sell and sends the order automatically.
- API: the connection that lets software talk to your broker’s trading system.
- Algo provider: a company that offers algo strategies or tools.
- Empanelled: approved and listed by the exchange. NSE has published empanelment criteria for algo providers.
Think of a shop and its delivery partner. The customer holds the shop answerable, even though the partner carries the parcel. Your broker is the shop, and the provider is the partner.
Orders you place by hand in the app or website are outside these rules.
How Does It Work?
- The provider gets approved. An algo provider has to be empanelled with the exchange before it can serve retail clients through a broker.
- The broker onboards it. Sahi’s guide says brokers must do due diligence on each provider before onboarding.
- The algo is registered. For black box algos, where the user cannot see the logic, SEBI says the provider must register as a Research Analyst and keep a detailed research report for each algo.
- The strategy runs on the broker’s servers. NSE’s circular says all strategies run on the broker’s servers and order messages originate there. NSE says this keeps risk management and data control with the broker.
- API orders are tagged. NSE treats all orders received through a client API as algo orders that need tagging. That includes strategies within the 10 orders per second threshold.
- Access is tied to an IP address. A static IP is a fixed internet address that the broker can recognise. NSE says a client static IP is required only for a tech-savvy investor placing orders through an API. If an algo platform is hosted on the broker’s infrastructure, the broker’s server IP is used.
- The broker monitors and handles complaints. Alice Blue’s guide says the broker is responsible for monitoring API activity and handling grievances, with providers acting as its agents.
Example With Numbers
This example is made up to show how the split works.
Riya has ₹2,00,000 in her account and subscribes to an empanelled provider’s Nifty strategy through her broker. She risks 1% per trade, which is ₹2,000.
Case 1: the strategy loses. Three trades hit their stop loss in a row. That is 3 × ₹2,000 = ₹6,000, or 3% of her capital. Nothing broke. It is an ordinary loss, and NSE’s FAQ places profit or loss with the client.
Case 2: a self-written script goes wrong. Riya’s friend Arjun codes his own strategy on his broker’s API. A bug makes it send 50 orders instead of 5, each worth ₹10,000. That is ₹5,00,000 of orders instead of ₹50,000. Arjun’s logic is his own, so he carries the outcome, while NSE says risk management is the broker’s job. Arjun’s job is to set sensible limits before he starts.
Event | Where it lands |
Strategy loses ₹6,000 as designed | Riya |
Bug in Arjun’s own script | Arjun for the outcome, broker for risk management |
Complaint about an algo service | Broker |
Provider that is not empanelled | Should not have been onboarded |
Key Features
- Principal and agent. The broker is the principal, and the provider acts for it.
- Tagged API orders. Every API order carries a tag so it can be traced.
- Broker side hosting. Strategies run on the broker’s servers, with an exception for tech-savvy clients who host their own algo on a static IP.
- Static IP rules. A client static IP is needed only for tech-savvy investors using an API directly.
- Order rate threshold. NSE refers to a threshold of 10 orders per second, which is 600 orders a minute. Alice Blue’s guide says strategies at this level need registration through the broker.
- Fee disclosure. Sahi’s guide says revenue sharing between the provider and the broker must be disclosed in full. Revenue sharing means they split what you pay.
Advantages
- You get one place to complain. You deal with your broker and not an app with no clear address.
- Only approved providers can reach the exchange through a broker, so there is a basic filter before money is at stake. Tagged orders also leave a trail, which helps when something looks wrong.
- Risk management stays with the broker for strategies run on its servers. That adds a layer between a faulty strategy and the exchange. It is a safeguard, not a promise that nothing will go wrong.
Risks and Limitations
- The framework controls how algos run. It does not make them profitable. A strategy that worked in a quiet market can lose steadily when prices start swinging, and leverage in futures and options can magnify the loss.
- Black box strategies carry extra risk because you cannot see the rules. Check whether the provider is registered as a Research Analyst, as SEBI requires for black box algos.
- Behaviour causes harm too. Some traders raise position size after a good fortnight. Others stop looking at positions because the system is automatic. A planned ₹2,000 loss can grow to ₹20,000 if nobody is watching.
Tips for Beginners
- Ask your broker if the provider is empanelled before you pay.
- Pick strategies whose rules you can read and explain.
- Test with a small amount first.
- Fix a daily loss limit in rupees and stop when you hit it.
- Never share your API key or login details.
- Read the fee page for subscription charges, brokerage, and any revenue share.
- Check positions every day.
Broker, Provider and Client Compared
Who | What they handle | What they do not handle |
Broker | Provider checks, risk management, complaints | Your profit or loss |
Algo provider | The strategy and its code | Direct access to the exchange |
Client | Strategy choice, position size, daily checks | Exchange approvals |
White box and black box algos differ in how much you can see:
Point | White box | Black box |
Logic | Visible, can be copied | Hidden |
Example | Moving average crossover | Subscribed strategy with undisclosed rules |
Extra requirement | Registered with the exchange | Provider registers as Research Analyst and keeps a report for each algo |
Common Mistakes to Avoid
- Expecting the broker to cover losses. The broker handles the process and complaints. Market results stay with you, so size positions to a loss you can take.
- Using an app your broker does not support. Orders may be rejected, and you may have no clear route to complain. Stick to providers your broker lists.
- Sharing API credentials. Anyone holding them can place orders in your account. Keep them private.
- Skipping the charges. A low subscription can come with brokerage and a revenue share on top. Read all of it.
- Running a strategy you cannot explain. If you do not know why it buys or sells, you will not know when to turn it off.
Who Should Use Algos?
Algos suit active traders who already understand their strategy, margin, and risk. They also suit people who can code and want to build on a broker’s API. Developers can start at the Alice Blue developer portal.
They are a poor fit if you are brand new to markets, hoping for quick returns, or unable to explain a strategy in two sentences. Learn manual trading first, then automate what already works.
Conclusion
When an algo sends orders through a broker’s platform, the broker is the principal. It checks the provider, keeps risk management on its side, and handles complaints, while the provider acts as its agent. None of that protects you from a bad strategy, because profit and loss stay with the client. Before you switch an algo on, ask whether the provider is empanelled, read how the strategy works, check every charge, and begin with an amount you can afford to lose. Treat automation as a tool, not a promise.
Frequently Asked Questions
Blog by Aliceblue News Desk
