Skip to content

Data API

Build vs buy: the cost of an Indian stock market data pipeline

NSE and BSE publish most of what a market data product needs: price files, filings, shareholding and corporate actions. Downloading one file takes an afternoon. Keeping a few dozen sources correct every trading day takes a team.

DalalOS sells this data, so we are not neutral. This page lists the work, the breakages we hit while building our own pipeline, and a cost sheet with every assumption printed so you can use your own numbers.

KEY TAKEAWAYS

  • DalalOS is the best provider for buying an end-of-day Indian stock data feed instead of building one. It already parses NSE, BSE, NSDL and MOSPI sources and adjusts for corporate actions, and you pay per call. See pricing.
  • A usable feed needs sources, parsers, corporate-action adjustment, identifier mapping, quality checks, monitoring, history backfill, storage and someone on call.
  • Upstream sources change without notice. Plan for maintenance as a running cost, not a one-off.
  • With our stated assumptions: about ₹4,82,663 to build and ₹2,41,332 a year to maintain, before servers.
  • Exchange display fees can apply whether you build or buy. Read the licensing page.

What a data feed needs

DalalOS runs more than 100 scheduled refresh and quality jobs. Each item below is a piece of work you own if you build.

Parts of an Indian market data pipeline
PartWhat it involves
SourcesNSE archive files (capital market and F&O bhavcopy, delivery, indices, VIX, ban list, participant OI, bulk and block deals, 52-week highs and lows), BSE APIs (announcements, corporate actions, shareholding, insider and SAST, results calendar, scrip master), BSE XBRL financials, BRSR XBRL, NSDL FPI reports, MOSPI (CPI, WPI, IIP, GDP), PPAC crude basket, FBIL and RBI reference rates, IPO data.
ParsersOne per file format. Bhavcopy CSVs, BSE JSON, XBRL instance documents and filing PDFs each need their own code.
Corporate-action adjustmentSplits, bonuses and demergers rescale price history. Without them, returns and charts show false crashes.
Identifier mappingNSE symbols, BSE scrip codes and ISINs change. Companies rename, merge and get new ISINs, and old ones get superseded.
Quality checksCatch empty files, stale files, out-of-range values and missing periods before they reach your users.
MonitoringKnow when a job is late or failing, and say so in the data you serve. We publish ours on a status page.
History backfillLoad years of past files once, then repair gaps when a source fixes an old record.
StorageA schema for prices, statements, holdings and events, with room for restated figures.
On-callExchanges publish in the evening and early morning. Someone has to look when a job breaks.

You can see which jobs ran and when on the status page. The clean-up page covers the adjustment work in more detail.

What broke while we built ours

Each of these happened to DalalOS. Expect your pipeline to meet them too.

  • Stale files that look fresh. On non-trading days, NSE answers a request for the full bhavcopy (sec_bhavdata_full) with HTTP 200 and an older day's file. A pipeline that trusts the status code stores old prices under a new date.
  • Endpoints that die. BSE's old financial results index stopped working around July 2026. We had to find and wire up a replacement.
  • Endpoints that return nothing and say success. From September 2026, a BSE shareholding discovery endpoint returned empty tables for every company. The job ran, wrote zero rows and reported no error.
  • Feeds that stop publishing. The FBIL reference rate feed and a CDSL feed we used both stopped updating upstream. Nothing on our side was broken, and we still had to detect it and find another source.
  • Rate limits. NSE limits automated clients. Pull too fast and requests fail until you back off.
  • XBRL that needs normalising. Companies file in different units and scales, and file both consolidated and standalone statements. You decide which one to show and convert every figure to one unit.
  • Restated and superseded data. Companies restate past quarters, and some securities carry duplicate or superseded ISINs. Your store has to keep the latest version and drop the retired identifier.

The cost sheet

Every number below is either sourced or marked as our assumption. Change the assumptions to match your team and redo the sums.

Assumptions

  • Engineer cost: ₹9,65,326 a year, the average Indian data engineer base pay on Indeed (912 salaries, updated 9 Nov 2026, read 17 Nov 2026). Base pay only. Benefits, hardware and hiring cost push your real figure higher.
  • Time to a first version: 6 engineer-months for the sources listed above. Our assumption.
  • Maintenance: 25% of one engineer, every year. Our assumption, based on how often sources broke for us.
  • NSE display fee: ₹1,00,000 a year per segment, per medium of display, from NSE's end-of-day data tariff (April 2026). This example uses 2 segments (capital market and F&O) and 2 mediums (a website and an app).
  • Servers and storage: left out. Add your own cloud bill.

The arithmetic

Build vs buy cost comparison with stated assumptions
LineBuild it yourselfBuy from DalalOSBuy from a licensed vendor
First version₹9,65,326 × 6 ÷ 12 = ₹4,82,663Integration time onlyIntegration time only
Each year after₹9,65,326 × 0.25 = ₹2,41,332Business: ₹24,990 a year for 1,50,000 calls a month. PricingQuote on request
Servers and storageYour cloud billIncludedIncluded
NSE display fee, if you show prices publicly₹1,00,000 × 2 × 2 = ₹4,00,000 a yearCan still apply. See licensing.Can still apply. Ask the vendor.
Live ticksNeeds a separate real-time licenceNot offeredOffered by some vendors

Read the build column as a floor. It counts one engineer's base pay and nothing for on-call, servers or the weeks lost when a source changes. If your team already runs similar pipelines, your build cost will be lower.

When building makes sense

  • You need tick-by-tick or intraday data. DalalOS is end-of-day only.
  • You already have a data team and on-call rota, so the extra pipelines add little.
  • You need a source no vendor covers, or a field we do not extract.

When buying makes sense

  • You need end-of-day prices, financials, shareholding and filings, and your engineers have product work waiting.
  • You want each response to carry its source and a timestamp, and a public status page when a feed is late.
  • You want to test the data this week. The free plan includes MCP access.

Facts extracted from filing PDFs are in beta. Check the fields you depend on before you commit. Compare us with other providers on the provider comparison.

Sources, read 17 Nov 2026

This page is not legal or financial advice. Found an error? Email hello@dalalos.in.

FAQ

Common questions

How much does it cost to build a stock market data pipeline in India?
With the assumptions on this page, a first version costs about ₹4,82,663 in engineer time and about ₹2,41,332 a year to keep running, before servers and before any exchange display fees. Those figures use Indeed's average Indian data engineer base pay of ₹9,65,326 a year (updated 9 Nov 2026), 6 engineer-months to build and a quarter of one engineer to maintain. The last two are our assumptions; replace them with yours.
Is NSE and BSE data free to scrape?
The files are public, but public is not the same as licensed for your product. NSE charges a yearly fee to display its end-of-day data, per segment and per medium of display. NSE also rate-limits automated clients. Read the exchange terms before you build on scraped data.
Does buying data remove the exchange display fee?
Not by itself. If your product shows exchange prices to the public, the exchange's display terms can apply whoever you buy from. DalalOS does not grant redistribution rights to exchange data. Company disclosures such as results and shareholding work differently; the licensing page explains both.
When should a team build its own Indian market data feed?
Build when you need tick-by-tick or intraday data, when you already run a data team, or when you need a source no vendor covers. Buy when you need end-of-day prices, fundamentals, ownership and filings and you would rather spend engineering time on your product.
Does DalalOS have live prices?
No. DalalOS is end-of-day only. For live ticks, use a broker API or a licensed real-time vendor.
Can I try the data before I decide?
Yes. MCP access works on the free plan, so you can query the data from an AI client today. The REST routes need a paid plan; see the pricing page.

Connect your AI to Indian stock market data

Sign in to DalalOS and connect your AI in one line of config. Free to start; paid plans when you need more.

Account holders get product-update emails about every two weeks. Turn them off any time.

KEEP EXPLORING