Intraday OI-wall option selling

Sell at-the-money options intraday on the Nifty-50 stocks pressed hardest against their open-interest "walls" — in at 9:20, flat by 15:20, naked — and at small size, on the most liquid names, net of every cost, you keep a real if modest edge. It is profitable in every reliable year of data, the selection is clean of look-ahead bias, and below you can verify it yourself: every trade, filterable, with the equity curve, drawdown, and Sharpe recomputing live as you slice it.

The hypothesis, and the exact trade

Single-stock options open the day rich. Overnight, nobody knows where a stock gaps, so the 9:20 premium is padded with gap uncertainty — and that padding bleeds out through the session as the range reveals itself. If you're short, you want to be short into that bleed and flat before the close.

We aim it at the open-interest walls — the strikes where call and put OI pile up, which tend to act as intraday resistance and support. Each day, off the prior session's settled OI:

• Sell calls on the stocks whose spot sits highest relative to their call wall — spot has run up into the strike where call OI is stacked (resistance overhead).
• Sell puts on the stocks whose spot sits lowest relative to their put wall — spot is sitting on the strike where put OI is stacked (support beneath).
• Sell the ATM strike (nearest tradable to the 9:20 spot), one lot per leg.
• Enter 9:20, exit 15:20, every day, naked, no overnight gap risk.

The book below is the top-5 names a side — the most extreme, most-liquid candidates — net of every cost. That's the configuration that holds up, and the dashboard shows you exactly why.

The backtest — explore every trade

This is all 14,196 legs, 2019–2026, at 0.5% per-side slippage. Slice by year, side, stock, or winners-vs-losers — the equity curve, drawdown, year table, and every metric recompute live from whatever subset you choose. The trades table carries the observed option LTPs, the slippage and fees charged, the underlying at entry and exit, and where spot sat relative to its OI walls.

The fill-rate column flags which years are trustworthy. Hit 2023+ (reliable) to drop the thin early years and see the edge on clean data alone; toggle Calls vs Puts to see which side carries it.

What to trust here, and what to discount

The selection is clean of look-ahead. Stored entry/exit match the 9:20 open and 15:20 open of that day's 5-minute bars exactly — same session, different bars, no same-bar leak. The stock ranking uses the prior day's settled OI (known before the open); the strike uses the 9:20 spot (known at entry). Nothing reaches into the future.

Trust 2023 onward; discount the early years. A leg only counts if it had a real print at both 9:20 and 15:20. Pre-2023 that fill rate runs 62–85% — the illiquid contracts that didn't trade cleanly at our timestamps simply drop out, so those years quietly count only the liquid legs and read better than reality. From 2023 the fill rate is 93%+, so that's the window to believe (the dashboard greys the thin years and flags them ⚠). On that clean window the strategy is positive every year.

It's a small-book trade, and slippage is the swing factor. 9:20 is the widest-spread minute of the day. The dashboard runs at 0.5% per side; push it to a realistic 0.75% and the edge thins but stays positive at small size — and turns negative if you scale up:

Names ranked 6–25 are less liquid and pay more spread than edge, so a big book bleeds out under realistic slippage. Stay small.

What you actually keep

Even in the winning configuration, most of your gross goes to execution before you keep a rupee:

GROSS premium captured:   large and positive
− slippage (the big one):  ~half the gross
− STT/brokerage/txn/GST:   ~15% of gross
= NET:                     a real but thin residue

This is a thin-margin, high-turnover trade. If you run it, it lives or dies on your fills.

What this is not

• Not a scalable machine. Past ~5 names a side, spread cost overwhelms the edge; a 25-name book goes negative under realistic slippage.
• Pre-2023 is unreliable — 62–85% fill means those years only counted the liquid legs. Believe the 2023+ window.
• Naked tail risk is real — single-day drawdowns of ₹9 L+ on one-lot books; rout-type days still bite.
• Margin/ROM would be a 15%-of-notional proxy, not a SPAN computation, so we don't headline it.
• Not advice. This is a research log, not a trade plan. Quantcents does not publish trade levels.

What to take away

A small (3–5 names a side), most-liquid-names intraday ATM short on the OI walls is genuinely profitable — clean of look-ahead, positive in every reliable year (2023–2025), Sharpe roughly 1–2.6 a year net of modeled costs, and still positive under a stressed 0.75% slippage. The full 2019–2026 curve is up at N=5.

It's a tight-execution trade, not a free one: it lives on your fills, and it breaks the moment you try to scale it. A real edge — small, liquid, and unglamorous. Most real ones are.

Sources & method

• Up to 48 Nifty-50 constituents, ATM near-month equity options, 5-minute intraday bars at 9:20 / 15:20, 2019–2026 (ICICI Breeze / NSE). Pre-2023 chain coverage ~37/49 names; fill rate reported per year in the dashboard.
• Costs: Zerodha equity-F&O schedule — date-aware options STT (0.0625% → 0.1% → 0.15%), ₹20/order brokerage, NSE txn 0.03553% of premium, 18% GST, SEBI, stamp; slippage modeled per side at 0.5% and stress-tested to 0.75%.
• Lot sizes from the NSE F&O scrip master (current, held constant — documented approximation). Walls = OI-weighted average call / put strike from prior-day settled OI. Survivorship: current constituents.