Most "0-DTE strategies" sell premium and pray. Day Trade does one thing well and lets the stop carry the edge — not a forecast. At 9:50 ET it sells a same-day SPY strangle just outside the money, and the instant price touches a short strike it buys that side back AT the strike — never before it, time value only, never intrinsic. After that first stop the winner-guard closes the other leg $1.50 before its strike, so a whipsaw is closed out before it can complete. Every rule below was settled on five years of real SPY 0-DTE option prices (2021–2026) — no Black-Scholes, no models — and priced at the fill you actually get.
The algorithm, start to finish
- 9:45 (optional) — buy the far wings first. A penny or two, deep out of the money, so the position is defined-risk before any short goes on. They expire worthless; leave them off to trade naked.
- 9:50 — sell the cell, one contract each side. C1/P1: short call a dollar above the ceiling of price, short put a dollar under the floor — a touch outside at-the-money, fewer whipsaw days than the tighter ATM cell.
- One side touched → stop AT the strike. The moment price touches a short strike, buy that leg back right there — AT the strike, never before it, time value only, never intrinsic. A one-sided touch is just a directional day: that side is done, the other rides.
- Reverses $1.50 → close the winner. If price turns and comes back within $1.50 of the other (winning) strike, close that leg too. You bank the gain and step out before the both-sided whipsaw completes. Now flat. This single move — the winner-guard — is the whole drawdown story.
- Travels the full $3 → one more chance. If price keeps going and reaches the opposite strike, re-sell the leg you were first stopped on — one lot, now $3 out-of-the-money and cheap. Guard it the same way: close it $1.50 before its strike, or let it expire. Once per day, never more.
- The touched leg stops AT the strike; everything else is guarded $1.50 before. The first stop fills AT the strike (time value only). The winner and the re-sold leg each come off $1.50 before their strike — never allowed to travel the last $1.50 into a stop.
- Hold survivors to the close. Anything still open is out-of-the-money, expires worthless, can never be assigned. The wings expire too. Nothing carries overnight.
- Size at the open — no doubling. One lot, two, five — the per-lot edge is identical. If you want more size you set it at entry, not half-way through the day. No stacking, no martingale, ever.
- Trades every weekday. Wednesday is the softest day (~57% win — still positive); leave it on, or skip it. Your call, not a hard rule.
- Sit out the known event days. On FOMC and major-event days the whipsaw rate roughly doubles (~57% vs ~26% on a normal day), and the worst losses cluster there — the single worst day in five years was an event day. These are known a day ahead, so the cleanest filter is simply to mark the day skipped before the open. It is the one place a calendar beats the market.
What to expect — one lot, five years, real 0-DTE prices (filled at the strike)
| Cell | Avg / day | Win days | Worst day | Big-loss days |
|---|---|---|---|---|
| C1/P1 + winner-guard | ~$20.5 | 65% | −$176 | 18 of 917 |
About 65% of days finish green. The worst single day on one lot is about −$176 — a few days of profit, fully recoverable — and only 18 days in five years lose more than $80. There is no fat tail: the loss on any day is capped at the strike, and the winner-guard clips the both-sided whipsaws before they complete. Figures are one lot on real Polygon 0-DTE prints; size is just a multiplier on top.
Why it’s “$20 a day,” not “$12”
The same trades price out very differently depending on how the stop fills. Model the buy-back at the 1-minute candle close and you get ~$12/day — but on a fast break that candle is already past the strike, so you’re buying back in-the-money (expensive). Fill at the strike — where the option is exactly at-the-money, time value only — and the same trades earn ~$20/day, with a smaller worst day too (−$176 vs −$280). The $12 was the modelling artifact; the $20 is the fill your synthetic stop actually gets.
Why stop AT the strike
At its strike an option is exactly at-the-money: zero intrinsic value — all time value. Buy it back there and you pay only that small time value. Wait even a little and price goes in-the-money, and now every dollar it moves is a dollar of intrinsic you owe. The strike is the one exit where you have paid no intrinsic at all — and because the touch is the close, you never hold an in-the-money leg, which is exactly why surviving legs ride safely to expiry.
Where the money is made — and the toll
| Day type | Share of days | Role |
|---|---|---|
| Directional (one side touched) | 67% | the engine — you stop the touched leg and keep the other |
| Untouched (neither side) | 6% | quiet wins — keep the full credit |
| Whipsaw (both sides touched) | 27% | the toll — capped at the strike, clipped by the winner-guard |
The directional days are the engine; the whipsaw days are the toll. Most whipsaws complete early (before 10:30) when time value is fattest — that is exactly why the winner-guard, which steps out on the first $1.50 reversal, helps most. We tested six different ways to dodge or cut the whipsaw, and capping it at the strike plus guarding the winner beat every one of them.
What we tested and threw away
| Idea | Why it was rejected |
|---|---|
| Re-center (re-sell at-the-money on the touched strike) | Looked like ~$27/day—but one reversal turned a naked ATM short deep in-the-money for a −$1,727 single day. A tail bomb. |
| Double / scale the winner to 3 lots | Lifts the average, but the loss size doubles — big-loss days jump 12→78. More income, fatter tail. |
| Trailing-roll the winner up the trend | The $1.50 stop is too tight for a trend — you exit on noise. Income fell to ~$9/day. |
| Midday time-exit (close everything at 12:30/1:00) | Cost about a third of income and barely moved the drawdown — exiting forfeits the afternoon decay. |
| Trend-close (close both on a >1% move by 10:30) | The big-move days are your best days ($60/day held). Closing them lost money. |
| Skip the whole day with a chop / moving-average filter | No predictive power — the quiet days are actually the safest, the volatile ones carry both the winners and the whipsaws. |
The one lesson: every income-add (double, re-center, scale, roll) juices the average but fattens the tail — the same volatile days carry your winners and your whipsaws, so you cannot cut one without killing the other. Capping the loss at the strike, guarding the winner at $1.50, and sizing at the open is the clean edge.
How we know it’s real — not a curve fit
- Real prices, not models. Every fill is an actual Polygon 0-DTE option print. Black-Scholes overstated this strategy several times over, so we never used it.
- The fill is honest. The same trades were priced both at the candle close and at the strike; we report the strike fill, which is what a synthetic stop actually achieves.
- Positive across regimes — through a bear (2022), bulls, and a chop year — not tuned to one.
- No fat tail by construction. The loss on any day is capped at the strike; the worst day in five years is about −$176 on a lot.
Runs itself, on the live feed
- Real-time stop off the live trade stream — the touch is seen the instant it prints, with a freshness guard so the stop never acts on a stale price.
- Wings fire first. Optional 1–2¢ wings are bought ~5 minutes before entry (≈ 9:45) so the defined-risk legs are in the book before any short is sold — no race.
- Auto or alert-only, paper or live — strikes, entry time, size, and the skip-Wednesday switch are all configurable.
Take the wheel — manual control
You are never locked out of the trade. While it’s open, the desk hands you both legs:
- Re-sell a leg — fire that side again at its same fixed strike. This is the answer to “it tapped my strike, the stop took it off, and then it pulled back — I want it back on.” A re-sell re-arms that side, so the next touch is seen again, as many times as the day asks for.
- Buy back a leg — close just that one leg, not the whole position and not “done for the day.” The other side keeps running, untouched.
- Sitting vs away. Watching the tape? Stay in Alert and you get a ping on every touch to act on. Stepping out? Flip to Auto and the engine runs the winner-guard and re-sell rules for you.
What it can and can’t do
- Can harvest the volatility premium the option buyers pay, and cap the loss at the strike.
- Can run hands-free all morning and put nothing at risk overnight.
- Can be driven by hand — re-sell or buy back either leg at its fixed strike, any time.
- Can’t move the strikes — they’re fixed for the day; the manual buttons only act on the strikes already set.
- Can’t guarantee profit. A violent whipsaw still costs the capped toll — that is the price of the edge. Paper-trade it first and watch the live fills.
Comments