Options for Quants · Volatility Research

The Volatility Clock

Implied vol is not flat across the trading day. Read from the tape of every symbol, one pattern dominates: ATM IV bleeds into the close — everywhere, and a long straddle held through the last hour reliably loses. Two secondary rhythms and one non-event round it out.

By Options for Quants Series 5-min ATM IV Window 2026-03-23 → 07-31 Coverage 49–56 days / symbol Regime single (falling vol)

The dominant pattern

End-of-day crush

Last-hour IV falls on 67–76% of days. A long straddle held through it loses −2.7% to −6.2%, profitable only 24–40% of the time — significant on 7 of 9 symbols.

MCX evening

US open deflates

When US equities open (~18:45 IST) MCX IV falls, not spikes — GOLDM down 80% of days, crude 71–72%. The pre-open risk resolves. Not a long-vol entry.

A non-event

Europe open · flat

The NSE afternoon shows no systematic IV response to the European open. A useful null — there is nothing here to trade or to build a rule around.

NSE / BSE index vol

09:15 – 15:30 IST · normalised to each day's median

intraday IV shape (% vs day median) last-hour crush zone US-open zone (MCX) — zero = day's typical level

MCX commodity vol

09:00 – 23:30 IST · US session runs through the evening

What to do with it

proposed guardrails · shadow-first

Don't buy premium late

Block or discourage new long-premium entries in the last hour (all symbols). The IV crush plus theta overwhelms gamma — the held straddle loses on 60–76% of days.

A time-of-day entry filter

Skip the MCX US open for longs

18:45–19:15 IST is a vol-deflation window, not an expansion one. A long straddle into it loses on crude, gold and crude-mini. Wait for the deflation to pass or trade the other side.

Consistent with event-window analysis

The last hour favours the seller

The same crush that punishes buyers is a theta-capture window for short premium — subject to the gamma tail on the 24–40% of days price does move. Worth a shadow test.

Candidate — validate before deploying

Method. ATM IV from the canonical 5-min series (a 5-minute ATM-IV series), Brenner approximation, close-anchored DTE, front expiry. Each day is normalised to its own median so the intraday shape is comparable across vol levels. Event windows use per-day log-changes with a 2,000-sample bootstrap 95% CI and sign-consistency; the long-straddle P&L fixes the strike and expiry at window entry and marks it to the window's last print.

The crush is conservative. Intraday DTE shrink biases the IV approximation upward, so the true end-of-day crush is stronger than shown, not weaker.
One regime. Apr–Jul 2026 was a falling-vol regime; a rising regime could behave differently.
Gross of costs. Straddle P&L excludes round-trip cost, which only deepens the case against late long entries.
Shape is homogeneous. Day-shape clustering found one dominant profile per symbol plus event-day outliers — only crude shows genuine multi-type structure (3 day-types).

Bootstrap 95% confidence intervals · significance = confidence interval excludes zero

Disclaimer. This is a research note for informational and educational purposes only and does not constitute investment advice or a recommendation to buy or sell any security. Options and F&O trading involves significant risk of loss. The patterns described are drawn from a single, limited historical window (one falling-vol regime) and past behaviour is not indicative of future results. Nothing here accounts for transaction costs, slippage, or liquidity. Please read all scheme-related documents carefully and consult a qualified adviser before trading.