Big orders move price. They don't predict it.

Every order-flow course has the same slide: the big buyer comes in, price dips to fill it, then delivers. This drop is what that slide looks like when you run it on the whole tape.

Five questions, five measured answers. Does price come back against a big order first, then deliver? Do big sellers get the mirror? Does the order trade in its own favour immediately? What is different about the environment when the answers differ? And what, if anything, can a discretionary trader do with it?

What we counted

A sweep is one aggressor's fills grouped: consecutive same-side prints within one millisecond. Sierra Chart stamps a per-millisecond sequence number on the tape, so a market order that walks several resting orders and levels shows up as N records one microsecond apart — you can put the order back together. 3,683,401 sweeps were scored across 23 front-month NQ contracts, 2020-11-27 → 2026-08-14, 1,472 regular-hours sessions; every sweep of 25 lots and up is complete, smaller sizes are sampled as controls.

Two facts about size before any result. First, sweep size is stable in lots across the whole period — the 99.9th percentile of RTH sweeps sat between 41 and 50 lots in every one of the 23 contracts — which is not true of single-print size, so a fixed lot threshold is comparable across years here. Second, what “big” actually is: the top 0.1% of sweeps is 47 lots; 100–199-lot sweeps run about 32 a session, 200–499 about 6, and 500+ about one every three days.

Everything below is measured on the mid (half the quoted spread) and from the post-sweep book — so bid-ask bounce cannot manufacture a “dip,” and the sweep's own impact is never counted as continuation. Sizes are in the aggressor's direction: a positive number means price moved the way the big order was going.

1 · The moment it lands

Impact is the change in the mid from the book before the sweep to the book after it. It grows with size but sub-linearly, and it does not snap back. The level of the mid one, five and fifteen minutes later still sits at 98–104% of the impact for every tier from 100 lots up.

In points the book has thinned: a 100–199-lot sweep moved 8.5 ticks in 2021 and 13.4 in 2026. In basis points it has not — 1.50 bps then, 1.23 now, with 2022 the outlier at 2.22. Percent-normalise before you compare eras; NQ doubled.

2 · Then what? On average, nothing

Here is the answer to “do they come in and immediately trade in their favour.” Only by their own impact. Once the sweep is done, the average forward move is inside one point at every horizon to 30 minutes for every tier below 500 lots — and it starts with a give-back of one to two ticks in the first minute for the biggest sweeps, which is the book refilling behind them. Only 500+ lots lean toward continuation (+8 ticks at 5 minutes, +16 at 30) and with 446 events and standard errors of 6–11 ticks that is a hint, not a fact.

The individual outcome is noise around that zero. The 15-minute move after a 100+ lot sweep has a mean of +0.5 tick and a standard deviation of 175 ticks — 44 points. Any single big print tells you almost nothing about the sign of the next 15 minutes.

Buy and sell sweeps of the same size print the same impact (10.2 vs 10.9 ticks for 100–199 lots), the same give-back and the same flat drift. The mirror holds; neither side has an edge on this tape.

3 · Does it come back first?

For each sweep: which came first inside 15 minutes, a move of k ticks against the aggressor or k ticks in its favour — and given the sequence, where price sat at 15 minutes.

The dip is real. The delivery is not.

Read on the question as it is usually asked. For a 100–199-lot buy, the sequence dip, then deliver (one point each way) happens 46% of the time against 38% for deliver, then dip; for 200–499 lots it is 51% vs 31%. So yes — the more common order of events is that price comes back against the big buyer first, and it does so with a median delay of one second. That is not new sellers being sourced to fill the buyer. It is the book refilling the levels the order just cleared.

The second half of the story fails. After that dip, price is above the post-sweep level at 15 minutes 48% of the time (52% after a deliver-first). Small 1–3-lot prints show the same 48 / 52. Whichever way it went first is mildly where it still is; there is no delivery premium for the big order.

4 · What changes the aftermath — and what doesn't

Every split below is on 100+ lot sweeps in regular hours. Because big sweeps arrive after price has already moved their way (+8 points over the prior five minutes for 100–199 lots, +14 for 500+), each split carries a control: the same state on 10–49-lot sweeps. Where the big-order cell and its small-order control agree, size is not the cause.

a. What the sweep did to the book

The most stable pattern in the study. When 100+ lots barely move the mid (bottom quartile, about 3 ticks: the book was thick or a block absorbed them), the next minutes lean with the aggressor — continuation-first beats pullback-first (50% vs 44%) and the mean drift is +2.4 ticks at five minutes. When 100+ lots blow the mid 27 ticks through a thin book (top quartile), pullback-first dominates (56% vs 31%) and the mean drift is -2.1 ticks. Same signs in every year 2021–2026 for the sequence, five of six for the drift. The mechanism is transient impact: the more a sweep walked, the more the book refills behind it.

b. A fresh session extreme

A 100+ sweep that prints through the running session high or low (19% of them) gives back -4.1 ticks in five minutes (SE 1.1) with pullback-first 54% vs continuation-first 35%; interior sweeps run +1.0 (SE 0.5). Small orders at fresh extremes show -0.5, so this one is about the size. It washes out by 15 minutes (-2.4, SE 1.7). Its mirror — sweeps near but not through the extreme continue about six ticks at 15 minutes — is present at every size including 1–3 lots: that is the location ruler, not the order.

c. Book balance, on the year with depth

The desk has recorded the full NQ book every session since July 2025. For the 34,030 sweeps of 50+ lots in that window that join to a book snapshot within 20 seconds, split by resting size in the 10 points ahead of the sweep and behind it, z-scored within the day. Thin-ahead is what makes impact big (r = -0.24), so this is the depth-side view of pattern (a).

A big order that clears a thin far side while its own side is stacked tends to hold and extend; a big order into a book that is thin on both sides marks a wide, fragile tape. One year, one regime, and the desk's earlier test of this idea as a trade did not confirm an edge — read it as description, not as a rule.

d. Everything that does not change the direction

Time of day, the prior five-minute move, the vol regime, clustering, morphology, side and year all leave the forward mean inside 1.5 points, and the small-order control moves with them. What the environment does set is the width: after a big sweep in the first half hour or in the fast-vol tercile the next 15 minutes travel about ±30 points either way; midday or quiet, ±14–17.

What it has and hasn’t been through

How the desk uses it

• A big print is a liquidity fact, not a direction fact. Across 56,401 sweeps of 100+ lots the forward mean is zero at every horizon, in every year, on both sides. Chasing it and fading it are both coin flips minus the spread.
• Read the impact, not the size. A 100+ that moved the mid a tick or two was absorbed — the book is thick there and the tape leans on continuing. A 100+ that moved it five points found air — expect the immediate one-to-two-tick give-back and no follow-through premium. On the DOM this is the same read as thin-ahead vs thick-ahead.
• A big sweep that makes the new session high or low is not confirmation. Those lean back about a point over the next five minutes, more than small orders in the same spot. Small and short-lived — not a fade — but the print itself adds nothing to a breakout thesis.
• Use the environment to size the box, not to pick the side. Open and fast-vol: ±30 points of travel in the next 15 minutes. Midday and quiet: ±15.
• Calibrate “big” to the sweep, in lots. 100+ about 32 a session, 200+ about 6, 500+ one every three days. A 100-lot moves NQ about 13 ticks today and 8.5 in 2021, but the same ~1.5 bps.