The selling side — a 9:20 straddle that passed twelve years, and a mirror seller positive in every one of eleven

It works on NIFTY, with the stop-loss doing all the work. Sell ATM CE and PE at 09:20 on the nearest expiry, buy back a leg at +25% of its premium on one-minute closes, square off 15:15, real charges, one lot: 2,119 sessions from 2015 to 2026 gave net +₹3.49L, maximum drawdown ₹52.7k, worst day −₹11.6k — COVID, elections and crash days all survived — 56% green days, profit-to-drawdown 6.6. With a 50% leg stop the result was similar. With no stop at all: +₹13.7k and a ₹1.73L drawdown, including −₹41.8k on election result day. A 36-cell grid (straddle or one-strike strangle × 09:20/09:30/10:00 × six stop levels) had every stopped cell positive and every unstopped cell toxic; earlier entry was strictly better. BankNifty passed more strongly (+₹4.37L, 61% green) but its weeklies no longer exist. Caveats on record: the edge is a weekly-era edge, flat years exist (2023, 2026), and a stop caps minutes, not instant gaps — one 2021 morning lost ₹42k in a single minute.

Take the pivot system's signals and, instead of buying the breakout, sell the opposite ATM option — a long signal sells the put, a short signal sells the call — with the same index-pivot trail as the stop and weekly rolls. NIFTY over eleven years: +₹10.9L, positive every year including the 2016–18 winters that bled the futures; worst trade −₹26k; worst year drawdown −₹80k. BankNifty over seven years: +₹11.6L, all positive, 2020 its best year because it sold the premiums the buyers were forced to buy. A parity check tied the three results together: buyer −₹6.4L plus seller +₹10.9L roughly equals the futures result less the two-leg friction. The mechanism is half-delta trend capture plus theta income minus friction, with the trail as a built-in stop.

The tail is uncapped. A limit-down gap against a short put is not in the sample and could plausibly cost ₹1.5–2L a lot with blown spreads, so honest sizing is ₹3L a lot rather than the ₹1.5L margin minimum, and crash-morning exit slippage exceeds the 0.1% model. Status that day: backtest-validated only; the next step was a paper month with margin simulation and gap-morning fill measurement, and no real money before a written gap-scenario plan.

The numbers are plausible and the algorithm is empty: holding NIFTY overnight every single night is a 67% win and about +3,393 points a year. Every close-to-open trigger lands between 64% and 72% — it is the overnight risk premium repackaged. The useful version we found: hold futures overnight only after an up day with the close above the 20-day average — +1,794 points a year (≈₹95k net at one lot) with a maximum drawdown of 626 points, flat through downtrend cascades, exit in the opening print rather than later. Options are the wrong vehicle for it: a weekly ATM call captures only about 20% of the overnight index move after theta; futures capture about 92%.

A lottery. Of nine usable events, seven lost holding to the close (each Budget −₹6.8k to −₹14.2k; even the 2019 election lost because exit polls had pre-gapped it) and the 2024 election jackpot of +₹59.4k single-handedly dragged the total to −₹9.9k. Event volatility is priced rich; neither buying nor selling it is clean. The one actionable piece: the straddle seller skips Budget day.

A real phenomenon, not a strategy. Using a twelve-year proxy for implied volatility built from our own premiums, buying the straddle in the cheapest 5% or 10% of days made ₹62–66k over eleven years; at the 20th percentile it was zero. The seller's edge vanishes at the floor but never turns into a buyer's edge — 50 to 85 trades, ₹5–6k a year, 42% win. The buying side of the ledger closed with that result; the proxy stayed as a dashboard gauge.