Not risk-off, but premium is not cheap either. Long calls have been a coin flip here. If you want upside: a call spread, stock — or take the other side and sell a put credit spread, which is where the real fills show the edge (see below).
Every day since May 2019 was classified with today's rule, then a 21-day ATM call and put were bought at the VIX-implied price and held to expiry. This is the history the light is standing on.
| State | Days | Share | SPY up | Avg 21d move | Realized ÷ implied | Call P&L | Call wins | Put P&L | Put wins |
|---|---|---|---|---|---|---|---|---|---|
| Green | 321 | 18% | 67% | +1% | 0.87× | +34% | 55% | -15% | 25% |
| Yellow | 948 | 52% | 73% | +2% | 0.63× | +9% | 50% | -52% | 16% |
| Red | 542 | 30% | 67% | +1% | 0.64× | +3% | 46% | -42% | 18% |
| All days | 1811 | 100% | 70% | +1% | 0.67× | +12% | 50% | -43% | 18% |
P&L is percent of premium paid. Realized ÷ implied below 1 means options were priced for a bigger move than happened. Black-Scholes with the VIX as volatility, no skew, no bid-ask; the put numbers are therefore optimistic, the call numbers slightly conservative.
Same 21-day options, split by calendar year. Green-state calls beat all-days calls in 2 of 5 years — the light does not predict direction. Puts bought in the Red state lost in 7 of 8 years — that rule holds.
| Year | Green calls | Yellow calls | Red calls | All calls | Red puts |
|---|---|---|---|---|---|
| 2019 | +30% | +53% | +79% | +49% | -88% |
| 2020 | -23% | +15% | -1% | +9% | +98% |
| 2021 | — | +2% | -12% | -4% | -82% |
| 2022 | — | -28% | -35% | -31% | -38% |
| 2023 | +54% | +10% | +47% | +33% | -67% |
| 2024 | +42% | +26% | +43% | +38% | -88% |
| 2025 | -9% | +17% | -15% | +5% | -4% |
| 2026 | — | +30% | -32% | +6% | -44% |
Same 21-day calls, risk-off days removed, grouped by the VIX at entry. The common rule of thumb is "calls below 18"; the 15–18 band has not paid, so the light turns green below 15.
| VIX at entry | Days | Realized ÷ implied | Call P&L | Call wins |
|---|---|---|---|---|
| < 13 | 108 | 0.88× | +62% | 68% |
| 13–15 | 213 | 0.87× | +20% | 49% |
| 15–18 ← now | 339 | 0.58× | +2% | 46% |
| 18–22 | 263 | 0.65× | -4% | 42% |
| 22–30 | 276 | 0.66× | +18% | 55% |
| ≥ 30 | 70 | 0.64× | +57% | 79% |
The table above is a model. This one is real: SPY ATM options bought at the actual close on each day from Sep 2024 to Jul 2026 (938 fills, nearest-Friday expiry ≈ 30 days, $5 strikes) and settled at expiry. "IV − VIX" is how far the option's own implied vol sat from the VIX; "real ÷ model" is what you paid versus what the model assumed. Two caveats: this window is only two years, and it contains one crash (spring 2025) — states that caught that rebound look better here than their long-run averages.
| State · option | Fills | IV − VIX (pts) | Real ÷ model price | Model P&L | Real P&L | Real wins |
|---|---|---|---|---|---|---|
| Green · calls | 46 | -1.6 | 0.88× | -50% | -45% | 24% |
| Yellow · calls | 277 | -1.9 | 0.89× | +14% | +28% | 52% |
| Red · calls | 146 | -2.5 | 0.88× | -24% | -14% | 38% |
| Green · puts | 46 | -3.9 | 0.73× | -34% | -10% | 30% |
| Yellow · puts | 277 | -4.4 | 0.76× | -74% | -66% | 12% |
| Red · puts | 146 | -4.9 | 0.76× | -19% | +7% | 31% |
Since long puts bleed premium, the seller collects it. Real fills, Sep 2024 – Jul 2026: sell the ATM put, buy the 5%-OTM put, ~30 days, hold to expiry. Return is on the capital at risk (max loss); "blowups" are spreads losing ≥95% of it. The edge lives where the gauge says RISK_ON — the short put is a long-market trade, and the gauge's real skill is vetoing it when the market is breaking. It held in both halves of the sample (+20% / +21%). Do not sell them in Red: a third of Red spreads blew up in the crash half.
| Bucket | Spreads | Credit / width | Return on risk | Wins | Blowups |
|---|---|---|---|---|---|
| All days | 469 | 22% | +7% | 78% | 8% |
| Green | 46 | 18% | -1% | 67% | 2% |
| Yellow | 277 | 22% | +17% | 87% | 4% |
| Red | 146 | 23% | -8% | 64% | 18% |
| Gauge RISK_ON (any VIX) | 181 | 21% | +20% | 92% | 0% |
VIX below 15 and the gauge not in risk-off. Realized moves come closest to what options price here (about 0.87× implied, versus 0.63× elsewhere), so this is the cheapest a directional call gets. It is a price signal, not a direction signal: if you have a directional view, this is when to express it with a long call; if you do not, the light gives you none. Use 21+ days to expiry — shorter dates lose in every state.
VIX at 15 or above, gauge not in risk-off. Realized moves run about 0.6× what options price, so a long call needs a real directional view to pay. Pay for less premium (call spread, stock) — or sell the overpriced side: put credit spreads here made +16% of capital at risk with an 87% win rate in real fills, +20% with 92% and zero blowups when the gauge read RISK_ON.
Gauge in risk-off, whatever the VIX. The only rule here that held in nearly every year: do not buy puts — the drop the gauge is reacting to has already happened and the puts are priced for it (the one exception was 2020). Calls have been dead money too. Cut delta or sell defined-risk premium instead of buying it.