Ask a Fyers user why they wanted API access and the answer is usually data before orders. Live prices across an option chain, open interest by strike, candles to backtest against: that is the raw material of every open-interest table and every gamma panel. This guide looks at the fyers api from that data consumer’s seat, as the documentation stood in September 2026: what version 3 serves, what it costs, where the limits bite, and how much analytics work still stands between its packets and a decision.
The data surface, precisely
Fyers charges nothing for API access. You create an app on the API portal, register a redirect URL, and each trading day the hosted login hands back an auth code that is exchanged for an access token. That token lasts until early the next morning, so the login is a daily ritual, the same as at most brokers, and any pipeline has to treat it as a scheduled job rather than a habit.
On the REST side the fyers api serves quotes for up to 50 symbols per call, a market-depth endpoint with the order book, an option-chain endpoint that returns each strike with its open interest and the change in it, and a history endpoint for candles. The history windows matter for anyone backfilling: up to 100 days per request for the minute resolutions, which run from 1 to 240 minutes, up to 366 days per request for daily candles, and a 30-trading-day window for seconds resolution. Longer histories are stitched from several requests. Expired futures and options contracts can be pulled too, with Greeks and open interest, which is rare among Indian broker APIs and useful for anyone studying how a strike behaved into expiry.
Live data comes over a WebSocket data socket with two subscription types, symbol updates and depth updates, plus a lite mode that sends only last-price changes when that is all you need. One connection can track 200 symbols, which is enough for a full weekly chain of one index but not for several indices at once, so multi-index pipelines run several sockets.
The rate budget
The fyers api enforces 10 requests per second, 200 per minute and 100,000 per day, and the rule that hurts is the one about repeat offences: breach the per-minute limit more than three times in a day and the account is blocked for the rest of the day. A chain-wide candle backfill at 09:15 is exactly the burst that trips it. Pipelines that survive treat the per-minute budget as a design constraint, warm their history outside market hours, and put a limiter in front of every REST call.
What having the data does not give you
A socket with 200 symbols is an ingredient, not a dish. To turn the fyers api into the analytics an options trader actually reads, you still build chain assembly (holding a coherent snapshot as updates land out of order), open-interest baselines per strike so that a change means something, build-up classification from price and OI together, an implied-volatility solver and Greeks, and gamma exposure if you go one level deeper. Because per-strike open interest arrives through the option-chain endpoint rather than as a field on every tick, the pipeline also polls that endpoint on a timer and merges it with the stream, which is one more moving part and one more claim on the rate budget.
When the API is the right tool
Build on the fyers api when your edge is the custom computation: a proprietary indicator, a backtest that needs its own candle store, an execution system with its own eyes. The free access, the generous history windows and the expired-contract data make it one of the better foundations in India for that work, and the daily token is a small price.
When it isn’t: the analytics are already built
If what you want is the standard options-analytics stack, none of it requires your own chain assembler or Greeks engine. The live option chain with OI analysis, the change in OI by interval, the open interest chart, dealer positioning and the screeners are already running. You connect OIData with your own broker API key, market data streams straight from your own broker account into the analytics, and the chain assembly, OI baselines, Greeks and reconnect logic are already written. The division of labour that works: the fyers api for execution and anything proprietary, OIData for the market-reading layer that would otherwise take weeks to rebuild and maintain.
A worked example: a wall that is being abandoned
Take one analytic. “The 24,500 call wall is being unwound” needs the strike’s open interest at the previous close, its open interest now, the direction of price in the same window, and a comparison with the strikes either side. With the raw API that is a baseline table, a poll of the option-chain endpoint every minute inside the rate budget, and a classifier. The pre-built version is a shrinking bar on the OI Stats page with the change overlay switched on. That gap, days of work against a page you open, is the general shape of the build-or-use decision.
A practical starter pipeline
If you do build: a pre-open job that logs in, refreshes the symbol master and subscribes the day’s chain before 09:10; a socket consumer that keys state by symbol and reconnects on its own; an option-chain poller with a limiter that never exceeds the per-minute budget; a candle backfiller that runs after 16:00 in 100-day windows; and an alert on the login job, because the mornings it fails are the volatile ones you wanted the data for.
Fyers API data FAQ
Is the Fyers API free? Yes, API access carries no subscription as of September 2026; brokerage on trades is separate.
How much history does one request return? Up to 100 days for minute resolutions, 366 days for daily candles, and 30 trading days for seconds data.
How many symbols can one data socket carry? 200 symbols per connection, in symbol-update or depth-update mode, with a lite mode for last-price-only streams.
What are the rate limits? 10 requests a second, 200 a minute and 100,000 a day, and more than three per-minute breaches in a day blocks the account until the next day.
Can I get the analytics without writing code? Yes: connect with your own broker API key and the chain, OI, GEX and screener layers are already built.
Verdict
As a free data foundation for custom builds, the fyers api is generous: deep history windows, expired-contract data and a 200-symbol socket. Its rate rules are strict and its per-strike open interest is a polled endpoint, so the engineering it obligates is real. Decide which trader you are, and spend the effort accordingly.