Quant & algo · The evidence
Is algo trading profitable in India?
The regulator has counted it. About nine in ten individuals who trade futures and options lose money. Almost all of the profit made by the firms on the other side is made by algorithms. Both are true at once, and the gap between them is the whole subject.
What SEBI counted
- Over the three years to March 2024, 93% of more than one crore individual traders in equity futures and options made a net loss. Together they lost over ₹1.8 lakh crore; the average loser lost about ₹2 lakh, costs included.
- Only about 1% of individual traders made more than ₹1 lakh after costs over those three years.
- In the year to March 2024, proprietary trading firms made gross trading profits of about ₹33,000 crore and foreign investors about ₹28,000 crore. 97% of the foreign investors' profit and 96% of the proprietary firms' profit came from algorithmic trading.
- In the year to March 2026, 87.7% of individual traders made a net loss, and their combined net loss was ₹91,685 crore. Options accounted for about nine-tenths of it.
These figures are SEBI's, from its studies published on 23 September 2024 and in August 2026. They cover equity futures and options on Indian exchanges.
What the numbers do not say
They do not say that running an algorithm makes money. The algorithms that earned those profits belong to firms with research teams, servers beside the exchange, low costs and deep capital.
Automation carries out a rule. If the rule has no edge, automation loses money more reliably than a person would, because it never hesitates and never skips a trade.
So the honest reading is narrower than either side of the argument likes. Trading by tested rules, run by machine, is how the profitable firms work. Buying or building "an algo" is not the same thing.
Where the profit actually comes from
- An edge. A rule that has made money after costs on data it was not built on. Most rules that are tested honestly have none.
- Low costs. Brokerage, taxes and charges are paid on every trade, win or lose. A firm pays far less per trade than an individual, and gets better prices.
- Execution without error. The system does what the rule says every day, knows what it holds, and notices when something has gone wrong.
- Surviving the bad month. Size small enough, and limits firm enough, that the worst stretch in the test can be lived through when it comes.
Remove any one of the four and the result drifts towards the 93%.
What automation is good for
It is consistent. A person skips the trade after three losses, doubles up after a win, and exits early on a frightening day. Software does none of that.
It can be measured. Every decision is logged, so what happened can be set beside what the test said should happen. That comparison is the most honest number anyone will ever have about a strategy.
It scales. One person can watch a few instruments. A system can watch hundreds, and act within a second on all of them.
None of this makes a rule good. It makes a rule knowable.
How to tell, before risking money
- Was the rule tested on a period it was not designed on, and did it hold?
- Are all costs and a cautious allowance for slippage in the result?
- How many separate periods does the result rest on? One long favourable stretch is one case.
- What was the deepest fall in the test, and could the account and the person stand it?
- Has it run on paper against live prices, and then at small size, with results close to the test?
The method is set out in backtesting Indian markets and, for options, in backtesting Nifty options strategies.
A fair expectation
A sound systematic strategy has losing months, and sometimes a losing year. Its edge shows over hundreds of trades, not over ten.
A steady monthly return from an algo, promised in advance, describes something that does not exist. Under SEBI's framework anyone supplying algos to investors must now be empanelled with the exchanges, and a seller of strategies whose logic is hidden must be a registered research analyst.
Aurevian builds systems to a client's own rules and tests them before they trade. A test that finds no edge is reported as that. If you want a rule examined this way, write to Aurevian.
Questions people ask
Is algo trading profitable in India?
For firms with tested strategies and low costs, yes: SEBI found that almost all of the profit foreign investors and proprietary firms made in equity derivatives in the year to March 2024 came from algorithmic trading. For individuals the record is poor, and automating a rule that has no edge does not change it.
What percentage of F&O traders lose money in India?
93% of individual traders over the three years to March 2024, and 87.7% in the year to March 2026, according to SEBI's studies.
Can a retail trader make money with algo trading?
Some do. It takes a rule with a real edge after costs, tested on data it was not built on, and a system that runs it without error. A ready-made algo sold with a promised return is not that.
How much can you earn from algo trading?
There is no typical figure, and no honest one can be promised. It depends on the strategy's edge, its costs, the capital and the risk taken, and a tested strategy still has losing months.
Is algo trading better than manual trading?
It is more consistent, and it can be measured against its own test. It follows the rule every time, which helps if the rule is good and hurts if it is not.
Keep reading
Research and information, not investment advice. Historical patterns describe the past and do not promise returns.