Pavan2pk/nse-options-intraday

NSE Options Historical Data

Historical Indian index-options data (NIFTY, BANKNIFTY, SENSEX) for backtesting, in a single canonical Parquet schema. Assembled from NSE F&O bhavcopy (daily EOD, 2001→present) and Upstox 1-minute intraday candles (2024→present).

Coverage

Track	Underlyings	Granularity	Range
upstox_intraday/	NIFTY, BANKNIFTY, SENSEX	1 minute	Oct 2024 → 2026 (BANKNIFTY from Mar 2024)
historical_daily/	NIFTY (2001→), BANKNIFTY (2005→)	1 day (EOD)	2001 → present

~2.3 GB total, 62 Parquet files.

Layout

Partitioned by track / underlying / year:

upstox_intraday/NIFTY/NIFTY_2025.parquet
historical_daily/BANKNIFTY/BANKNIFTY_2010.parquet

Files are named {UNDERLYING}_{YEAR}.parquet, so a date-ranged query only needs to open the relevant years.

Schema (16 columns)

Column	Type	Notes
date	string YYYY-MM-DD	Trading date
timestamp	timestamp (IST)	Candle start; daily rows = date at market close
underlying	string	NIFTY / BANKNIFTY / SENSEX
expiry	string	Contract expiry date
strike	float	Strike price
option_type	string	CE / PE
exercise_style	string	european / american
open high low close	float	OHLC
volume	float	Contracts traded
oi	float	Open interest (NaN for Upstox intraday)
settle_price	float	NaN for intraday
source	string	e.g. upstox_expired, nse_bhavcopy
granularity	string	1min / 1d

Loading

import pandas as pd
from huggingface_hub import hf_hub_download

path = hf_hub_download(
    repo_id="<your-username>/nse-options-intraday",
    filename="upstox_intraday/NIFTY/NIFTY_2025.parquet",
    repo_type="dataset",
)
df = pd.read_parquet(path)

Or grab the whole dataset:

from huggingface_hub import snapshot_download
local = snapshot_download(repo_id="<your-username>/nse-options-intraday", repo_type="dataset")

Notes & attribution

• Daily data derives from the NSE F&O bhavcopy (publicly published by NSE).
• Intraday data was collected via the Upstox historical API; redistribute in line with the source providers' terms.
• oi and settle_price are NaN on Upstox intraday rows (not returned by that API).