A liquidity sweep (also called a stop hunt or liquidity grab) is a fast push through an obvious price level — above equal highs, below a clear swing low — that triggers the stop orders resting there, fills large opposing orders against them, and then reverses. It happens because stops cluster at predictable places, and a triggered stop is a guaranteed market order: exactly what a large trader needs to fill size without chasing. Nobody hunted your account personally; your stop was simply parked where everyone else's was. This post covers where resting liquidity sits and why price seeks it, how to distinguish a sweep from a genuine breakout, entry models after a sweep, stop placement that keeps you out of the pool, and an honest way to validate any of it.
Where Resting Liquidity Actually Sits
"Liquidity" here means resting orders — limits and stops waiting at specific prices. Stops concentrate at a short list of locations because every trading book teaches the same placement and every chart shows the same obvious levels:
- Above equal highs / below equal lows. Two or three swing highs at nearly the same price look like a "double top with a clear invalidation." Shorts stack buy stops just above it; breakout traders stack buy stops in the same spot. The mirror applies below equal lows.
- Beyond obvious swing points. The textbook rule is "stop below the swing low." When a swing low is visible on every timeframe, the area just under it holds thousands of sell stops.
- Round numbers. Prices ending in 00 or 50 attract stops and targets simply because humans anchor to them.
- Session reference levels. Prior-day high and low, overnight high and low, and the weekly open are marked on charts worldwide, so orders pile up just beyond them.
The uncomfortable implication: the more obvious and "safe" a stop location looks, the more company your stop has — and the more attractive that pool becomes.
Why Price Seeks Stops: The Auction Explanation
Futures trade in a continuous two-sided auction. Every buy needs a sell at the same price. That constraint is the entire explanation for stop hunts — no conspiracy required.
Suppose a fund wants to buy a few thousand ES contracts. Market-buying that size lifts every offer in the book and fills at progressively worse prices. A smarter fill: let price trade down into the pool of sell stops resting under a well-watched low. When those stops trigger, they become market sell orders — guaranteed, price-insensitive supply. The fund's resting bids absorb that selling, the initiating pressure disappears, and price snaps back. On the chart, all you see is a wick below the low and a sharp reclaim. Underneath, someone just bought size at a discount from forced sellers.
That's why a plain stop order is best understood as a promise to trade against yourself at the worst moment: it converts to a market order in the direction of the move, at the exact price where large players want counterparties.
No, Your Broker Is Not Hunting Your Stop
The myth that "brokers run your stops" doesn't survive contact with how listed futures work. NQ, ES, CL, GC and every other CME contract match on a central limit order book at the exchange. Your broker routes orders to that book; it does not take the other side of your trade, it cannot see the aggregate stop map of the market, and it cannot move a market that trades billions of dollars of notional per day to pick off a two-lot. (The myth has some historical roots in dealer-model markets — spot FX and CFD shops where your counterparty can literally be the platform — but that's a different market structure.)
What does exist: large participants who can read an obvious chart as well as you can, and who know that equal highs mean resting buy stops. Seeking that liquidity is rational auction behavior, not fraud. Deliberate order-book manipulation like spoofing is illegal and does get prosecuted — but you don't need a villain to explain sweeps. Ordinary liquidity-seeking explains them completely.
Sweep or Breakout? The Distinguishing Tells
Every sweep and every breakout starts identically: price trades through a level. The difference is what happens in the minutes after, and it comes down to acceptance versus rejection.
| Tell | Liquidity sweep | Genuine breakout |
|---|---|---|
| Time beyond the level | Seconds to a few minutes, then back inside | Price holds beyond the level and builds time there |
| Close location | Closes back inside the range; long wick through the level | Full-bodied closes beyond the level, ideally consecutive |
| Reclaim speed | Fast, aggressive snap back through the level | If price returns at all, it drifts — and a slow drift back is a failing break, not a sweep signal |
| Follow-through | Break direction dies immediately; next bars reverse | Pullback holds the broken level as support/resistance, then continues |
| Participation | Burst of volume at the extreme, then nothing behind the break | Sustained pressure in the break direction |
Two practical notes. First, close location is the cleanest single tell: a 5-minute candle that pokes above equal highs and closes back below them is rejection; three candles closing above is acceptance. Second, you never get certainty in real time — you're stacking probabilities, which is why the entry models below demand confirmation instead of guessing at the extreme.
Entry Models After a Sweep
Sweep and Reclaim
The base model, long side (invert for shorts):
- Price trades below a marked level (swing low, overnight low, equal lows) and triggers the pool.
- It closes back above the level within a few bars — the reclaim.
- On a lower timeframe, structure shifts: price breaks its most recent lower high, signaling that sellers who drove the flush are done. If market structure terms are fuzzy, read the guide to market structure first — the shift is the confirmation that separates this from catching a falling knife.
- Enter on the structure break or on the retest of the reclaimed level. Invalidation is the sweep low.
Sweep Into Confluence
The highest-quality version adds location. A sweep that terminates inside a higher-timeframe demand zone — an order block left by a prior institutional move — is a different trade from a sweep into thin air. If the reversal leg is impulsive enough to leave a fair value gap, the retrace into that gap is a defined second entry with a tighter stop. Sweep + order block + FVG + structure shift is the core sequence behind most ICT-style trading strategies; you don't need the full vocabulary to trade it, but the confluence logic is sound: more independent reasons for the level to hold, better expectancy per attempt.
Session Context
Sweeps concentrate where liquidity and volatility concentrate. The New York open regularly raids the overnight high or low within the first 30–60 minutes before the day's real move — one reason when you trade futures matters as much as what pattern you trade. A ruleset that only takes sweeps of overnight or prior-day levels during the opening 90 minutes is far more testable than "trade every wick."
Don't Be the Liquidity: Stop Placement
If stops cluster just beyond obvious levels, the corollary writes itself: don't put yours at the obvious pool.
- Beyond the sweep extreme, plus a buffer. After a sweep-and-reclaim entry, the sweep low is your invalidation — but placing the stop exactly at it invites a second, deeper probe. A modest buffer (on NQ, even 10 ticks is only 2.5 points, or $12.50 per contract) puts you behind the crowd instead of in it.
- Not just under the next textbook level. Moving a stop from one obvious location to another obvious location changes nothing. Ask where the next pool sits, and stay behind it or well inside it — not at it.
- If the honest stop is too big, size down — don't tighten. A structurally correct stop on NQ can be 40+ ticks ($200 per contract). The fix is trading MNQ instead of NQ at one-tenth the dollar risk, not shaving the stop until it sits back inside the liquidity pool. A futures calculator turns a stop distance into dollar risk per contract in seconds.
A Written, Testable Ruleset
Vague pattern recognition can't be tested. Here is an example spec — every number is a parameter to test, not a claim:
- Market/timeframe: NQ, 5-minute execution, 15-minute levels.
- Levels, marked before the session: prior-day high/low, overnight high/low, equal highs/lows visible on the 15-minute.
- Trigger: price trades at least 10 ticks beyond a marked level, then closes back inside within 3 bars.
- Confirmation: 5-minute structure break in the reversal direction within 6 bars of the reclaim.
- Entry: close of the confirmation bar, or the retrace into a reversal-leg FVG if one exists.
- Stop: 10 ticks beyond the sweep extreme.
- Target: the opposite side of the local range, minimum 2R or no trade.
- Filters: first 90 minutes of the New York session only; no entries within 10 minutes of scheduled high-impact news.
- Limit: maximum two attempts per level, then done.
Every clause is objective enough that two people replaying the same session should log the same trades. That property — not the specific numbers — is what makes it a strategy you can actually backtest.
The Honest Part: Hindsight Bias Is Brutal Here
Liquidity sweeps have a credibility problem, and it's worth naming. After the fact, every reversal looks like a sweep. Scroll any chart and your eye lands on the wicks that reversed beautifully; the equal highs that broke clean and ran for 200 points don't register, because a breakout doesn't look like a "failed sweep" in hindsight — it looks like nothing. That selection effect makes sweep content the most seductive genre of trading education: the marked-up examples are real, and still prove nothing about live-edge win rates.
Be equally suspicious of anyone quoting a success rate for sweeps in general. There is no credible public dataset on it, and the pattern's outcome depends heavily on the ruleset, session, and market — a number without those attached is decoration.
The only honest validation is forward-testing against data you haven't seen: mark your levels, play the session forward without the right edge visible, take every signal your written rules generate, and log the results. Fifty samples is a reasonable floor before the numbers mean anything, and what you're measuring is expectancy — win rate and risk-reward together, not a highlight reel of winners.
Practicing Sweeps Without Paying Tuition to the Market
Sweep trading is a pattern-recognition and discipline skill, and both compress well under repetition. Market replay is built for exactly this: in TestMax's futures backtesting, historical NQ and ES sessions stream bar by bar at 1x–50x with the right edge hidden, so a sweep looks exactly as ambiguous as it did live — you commit to sweep-or-breakout in real time, get simulated fills, and the per-setup analytics tell you after 50+ samples whether your ruleset has positive expectancy or just good marketing. A practical loop: mark levels premarket, replay the New York open, take only written-rule signals, review expectancy weekly, and change one parameter at a time.
The free plan includes futures replay with three months of data and no credit card, which is enough to run the 50-sample validation before you risk anything. Whatever platform you use — there's a full comparison in the best futures backtesting software roundup — the standard is the same: if your sweep rules only work when you can see the right edge, they don't work.