A golden sweep is an unusually large, high-premium options sweep that meets aggressive directional criteria. The term marks the subset of sweeps big enough and urgent enough that they are unlikely to be routine hedging.
There is no official definition
This matters more than it sounds. "Golden sweep" is a market convention, not an exchange designation. No regulator defines it and no data feed flags it natively. Every platform applies its own thresholds, which typically combine:
- Aggression — filled at or above the ask.
- Premium — a large total dollar amount, often six figures or more.
- Urgency — near-dated expiration, commonly under 30 days.
- Size — contract count well above the strike's norm.
Because the thresholds vary, the same trade can be a golden sweep on one platform and an ordinary sweep on another. Do not compare counts across tools and expect them to agree.
What it suggests
The combination is genuinely informative. Someone paid up, in size, for short-dated exposure. Short-dated contracts decay fast and offer little room to be early — you do not buy them as a long-term hedge. The structure implies a view with a clock on it.
What it does not tell you
The reason. You see a print. You do not see whether it is an opening bet, a closing sale, one leg of a four-leg structure, or a hedge against an equity block that crossed somewhere you cannot see.
Whether they are right. Large, confident, and wrong is a well-populated category. Size is not accuracy.
Whether you can act on it. By the time the print is on your screen, the move it anticipated may already be partly priced. The premium you would pay to follow is not the premium they paid.
Using it
Golden sweeps are best used as a scanner, not a signal — a way to surface names where something unusual is happening so you can go look. The work starts there rather than ends there: check whether the flow stacks against existing dealer positioning or fades it, whether it repeats, and whether the chart agrees.
Following golden sweeps mechanically is a strategy with a long history of disappointing the people who try it. Using them to decide what to research is considerably more defensible.