tejhq/indian-markets

End-of-day OHLCV for NSE and BSE listed equities. Built straight from the exchanges' official Bhavcopy. No broker, no auth, no scraping — just clean SEBI CMTS data.

Updated daily after Indian market close (~6:30 PM IST).

Coverage

Exchange	Series	Instruments	From
NSE	EQ/BE/BZ	~2,300 / day	2024-01-01
BSE	A/B/T	~2,200 / day	2024-01-01

NSE and BSE both moved to the SEBI CMTS bhavcopy format around late 2023. We start where the schema is uniform. Older data (pre-2024 legacy formats) is on the roadmap.

Schema

Field	Type	Notes
date	date	Trading date
symbol	string	Ticker (e.g. RELIANCE)
series	string	Exchange series code
isin	string	International securities ID
name	string	Full instrument name
open	float64	Open price (INR)
high	float64	High
low	float64	Low
close	float64	Close
last	float64	Last traded price
prev_close	float64	Previous close
volume	int64	Total traded volume (shares)
turnover	float64	Total traded value (INR)
trades	int64	Number of trades executed

Layout

Hive-partitioned parquet, one file per (exchange, date):

nse/year=2025/month=04/date=2025-04-30.parquet
bse/year=2025/month=04/date=2025-04-30.parquet

zstd-compressed. Partitions prune cleanly in DuckDB / pyarrow / datasets.

Quickstart

Polars

import polars as pl
from huggingface_hub import hf_hub_download

p = hf_hub_download(
    "tejhq/indian-markets",
    "nse/year=2025/month=04/date=2025-04-30.parquet",
    repo_type="dataset",
)
df = pl.read_parquet(p)
print(df.filter(pl.col("symbol") == "RELIANCE"))

DuckDB

SELECT date, symbol, close, volume
FROM read_parquet(
    'hf://datasets/tejhq/indian-markets/nse/**/*.parquet',
    hive_partitioning = 1
)
WHERE symbol = 'RELIANCE'
  AND date >= '2025-01-01'
ORDER BY date;

datasets

from datasets import load_dataset
ds = load_dataset("tejhq/indian-markets", data_files="nse/**/*.parquet")

Source & license

• Source: NSE and BSE official EOD Bhavcopy — public, free, redistributable
• Code: github.com/tejhq/tej-bazaar (MIT)
• Data: Exchange-published; redistribution as cleaned parquet is permitted. Verify exchange terms before commercial use.

Caveats

• No corporate actions yet (splits, bonuses, dividends). Adjusted close lands in a later phase.
• Same-day data appears ~6:30 PM IST after market close — fetching earlier returns nothing.
• Bhavcopy occasionally has anomalies (close outside [low, high], zero-volume rows). The pipeline filters those out before publishing.
• Symbol changes (renames, mergers, delistings) are not yet tracked across history.

Roadmap

• GitHub Actions cron — auto-publish daily at ~6:30 PM IST
• Pre-2024 legacy bhavcopy backfill
• Corporate actions + adjusted close
• Derived metrics (returns, 52w hi/lo, avg volume)
• REST API via tej-api — same data, low latency, with auth tiers

Full plan: ROADMAP.md.

---

Tej — sharp, fast, bright. Just like the data should be.

Part of TejHQ — dev-first financial data infra for India.