The ₹24-lakh truth about Indian market data

End-of-day OHLCV for every NSE and BSE equity is a public good. SEBI mandates that exchanges publish it. The exchanges publish it. It is, in the most literal sense, free.

And yet the going rate to access it through a vendor in 2026 is somewhere between ₹40,000 and ₹24,00,000 a year, depending on whom you ask. This is not a scandal - it is a market for convenience, and convenience is genuinely valuable. But the gap is wide enough that every retail quant should understand exactly what they are paying for, and what they could substitute.

The price ladder

Here is roughly what the Indian EOD market data ladder looks like, as of March 2026:

• ₹0/year - the raw exchange Bhavcopy. Free, public, redistributable. You parse it.
• ₹0/year - TejHQ. The Bhavcopy parsed and shipped as Parquet. Same data, no parsing.
• ₹3,000–₹10,000/year - the broker freebies. Zerodha Kite, Upstox API, etc. Free to their account holders, EOD-only, rate-limited, terms restrict redistribution.
• ₹40,000–₹1,20,000/year - the retail quant vendors. Names like upstox-historical, yfinance-india, definedge. Add adjusted close, corp actions, sometimes derivatives.
• ₹2,00,000–₹6,00,000/year - the mid-market vendors. Refinitiv Workspace lite, Bloomberg Anywhere lite, FactSet entry tier. Add news, fundamentals, basic terminals.
• ₹18,00,000–₹24,00,000/year - Bloomberg Terminal proper. The full one. Tick-level, all asset classes, the chat, the analytics, the everything.

For retail-scale quant work - backtesting, factor research, dashboards, ML - the actual EOD content across most of these tiers is identical. You are paying for the wrapper.

What you're actually buying

Vendors at every tier above ₹0 sell some combination of four things, in roughly this order:

1. Format normalization

The Bhavcopy is a CSV with shifting column orders, occasional encoding glitches, and series codes that differ across exchanges. Parsing it correctly is annoying. Vendors do this once and sell the cleaned result. This is what TejHQ replicates for free.

2. Corporate actions and adjusted close

Splits, bonus issues, dividends, symbol changes. The exchange Bhavcopy publishes the raw close, not the split-adjusted close. If RELIANCE does a 1:1 bonus, your factor model needs to know that the price didn't actually halve. Building this engine is real work - corp action notices come in PDF, parsing them needs OCR or careful schema-by-schema work, and the adjustment logic differs by event type.

This is the gap TejHQ is closing in Phase 4. Once we ship corp actions, the ₹40K-₹1L retail vendors lose their main differentiator.

3. Tick or minute-level granularity

EOD is one row per symbol per day. Tick is every trade. Minute bars sit in between. Storage, bandwidth, and exchange licensing all explode at higher granularity. This is a real cost that justifies real prices. If you genuinely need tick data, pay for it. If you don't (and most quant research does not), don't.

4. Coverage beyond Indian equity

Derivatives, currency, commodity, global cross-listings, fundamentals, news, alt data, ESG. The ₹2L+ tiers are mostly buying coverage breadth, not depth on Indian equity. If you only trade NSE+BSE cash, you are subsidizing every other asset class on the platform.

The hidden tax: redistribution clauses

The single most underappreciated cost of vendor data is the contract. Every paid feed restricts what you can do with the data downstream. You typically cannot:

• Republish the data, even derived metrics, on a public dashboard.
• Train a public model on it without commercial licensing.
• Share it with a co-founder or contractor who isn't on the same seat.
• Retain it after your subscription lapses.

For a solo quant or a small fund this is a non-issue. For anyone building tooling, an open-source project, or a research output you intend to publish - these clauses are the entire conversation. Bhavcopy and TejHQ are both MIT-licensed code on top of redistributable raw data. There is no clause to negotiate.

So who should pay?

We genuinely think paid feeds are the right answer for two audiences: institutions with compliance requirements that need a contracted SLA, and active intraday traders who need sub-minute granularity.

For everyone else - students, researchers, retail quants, fintech indie hackers, dashboard builders, ML practitioners - the case for paying ₹40K+ a year for EOD data was already weak in 2024, and is approaching zero in 2026.

Disclosure: TejHQ is the open-source project this blog belongs to. We have an obvious bias. We also cite our prices and sources, which most posts on this topic do not.