What Is a Sweep in Order Flow?

Most trades happen at or near the top of the book: someone takes the best offer, the spread refills, and the price barely moves. A sweep is different in scale and intent. It does not stop at the best level; it plows through several, accepting worse and worse prices to complete the order now.

Picture a Yes contract with asks resting at 50, 51, and 52 cents. A normal buy might lift the 50s and stop. A sweep takes the 50s, the 51s, and the 52s in a rapid sequence, moving the price several cents in seconds. The defining features are speed and the number of levels cleared, not the raw dollar size alone.

This is also why the depth of the book underneath the sweep matters so much. A sweep through a thick book that quickly refills is one kind of event; a sweep that empties a thin book and leaves a gap behind it is another. The same number of contracts can produce a very different price move depending on what was resting there to absorb it, so the read is always the trade against the book it hit.

The logic behind treating a sweep as a signal is straightforward: paying up costs money, so a participant who sweeps is revealing that getting filled now is worth more to them than getting a slightly better price later. That often lines up with a view that the price is about to keep moving, whether because of information, a developing story, or pressure elsewhere.

A sweep is especially informative on Kalshi, where the public data carries no user identities. You cannot follow a wallet the way you can on Polymarket, so the shape of the flow is the read. A sweep says decisive money just took several levels, even though it never says who. The comparison of Polymarket versus Kalshi explains why the two venues force different approaches.

A sweep rarely tells the whole story by itself, and it reads best alongside the other order-flow signals. Book imbalance, one-sided resting depth with order concentration, tells you where standing pressure sits before the sweep. Steam, fast price movement on volume, tells you whether the move stuck and kept going after the sweep or faded back.

A sweep into a thin, imbalanced book that is then followed by steam in the same direction is a more coherent picture than an isolated sweep that immediately reverses. Reading the book underneath the sweep is the skill covered in the explainer on how to read an orderbook, and the two topics go hand in hand.

The order-flow view on SharpPredict is built to show these signals together rather than in isolation. Seeing a sweep next to the imbalance that preceded it and the steam that followed lets you judge whether an event was a decisive, sustained move or a one-off that the market absorbed and shrugged off. A signal read in company is far more useful than the same signal read alone.

A sweep is behavior, not identity, and it is not a verdict. It does not name the participant, it does not prove they are informed, and it does not guarantee the price will keep moving. Sweeps can be wrong, they can be forced by risk management rather than conviction, and they can reverse minutes later. Chasing every sweep would be a fast way to trade noise.

Treat a sweep as a prompt to look closer: what side, how many levels, what the book looked like before, and what happened after. SharpPredict surfaces the signal so you can do that research. It is not advice or a pick, and nothing in the app places, routes, or executes a trade.

Sweeps are one of the clearest ways to read a market that gives you nothing else to go on. On Kalshi, where there are no identities to follow, the flow is the story, and a sweep is one of its loudest sentences. It marks a moment when someone decided speed was worth paying for, and moments like that are exactly what a research view wants to catch and timestamp.

Used with judgment, sweeps help you build a sense of a market's rhythm: when it is quiet, when pressure arrives, and how it responds. That context is the real payoff, more than any single sweep. It is a lens for understanding market behavior, not a set of instructions, and it fits alongside book imbalance and steam to describe what a contract is actually doing in real time.