What Is the ICT Scalping Strategy?
The ICT scalping strategy is the standard delivery method executed at minimum scale: the same sweep → displacement → fair value gap sequence behind every ICT model, run on the 1-minute chart inside a killzone — ideally inside a twenty-minute macro window — targeting the nearest intraday liquidity pool, with hold times of five to thirty minutes. It is not a separate system and it skips nothing: the daily bias and the day's draw still govern direction. The 1-minute chart decides only the entry's coordinates.
The inversion in that definition is the whole article. Retail folklore ranks timeframes by difficulty from monthly (hard, slow, for institutions) down to 1-minute (easy, fast, for beginners with small accounts) — and the folklore is exactly backwards. The 1-minute chart demands every piece of the full method plus speed plus cost discipline, with less time per decision and five times as many decisions per session. ICT's own curriculum teaches the higher timeframes first for precisely this reason: precision is the reward for context, not a substitute for it. A trader who cannot name the day's draw has no business on the 1-minute chart, because a 1M setup against the 15M delivery isn't a smaller opportunity — it is a well-timed entry into opposing order flow.
The Non-Negotiables — What Scalping Doesn't Get to Skip
The bias and the draw, unchanged. Before the first 1-minute candle matters, the same two sentences every model on this site requires must already be written: today's bias, and today's address — the draw on liquidity price is being delivered toward. The scalper trades with that delivery in fragments: where the swing trader rides the whole leg, the scalper harvests its most violent ten-minute segments. Same river, smaller net — but only ever fishing downstream.
Killzones only, macros preferably. Time is the scalper's primary filter because the 1-minute chart prints "setups" continuously — most of them noise wearing the sequence's costume. The killzones restrict trading to when displacement actually happens; inside them, the macro windows — 9:50–10:10, 10:50–11:10, 2:33–3:00 for London, the lunch and PM macros — restrict it further to the twenty-minute bursts when the algorithm runs its liquidity errands. Those bursts are short, violent, and mean-reverting: precisely the shape a 1M sequence is built to capture, and precisely what the dead hours between them do not contain. The honest version of "when should I scalp?" is a list of about ninety minutes per day.
The cost math, faced squarely. Here is the arithmetic scalping courses omit. Every trade pays spread, commission, and slippage — a roughly fixed toll — and a scalp's target is small, so the toll is a large percentage of the prize. On NQ, a round-trip cost of ~1.5 points against a 15-point target consumes 10% of every winner; on EUR/USD, a 0.6-pip all-in cost against an 8-pip target consumes 7.5%. That tax compounds with frequency: at five trades a day it silently converts a gross 2R month into a net 1.2R month. The consequences are structural, not motivational — fewer, better-timed trades with clean targets aren't a discipline preference; they are the only version of scalping whose math survives. This, incidentally, is why the macro windows matter twice: they concentrate the edge and they cap the trade count by existing only ninety minutes a day.
The Timeframe Stack — Three Charts, Three Jobs
| Chart | Job | What you mark | What you never do on it |
|---|---|---|---|
| 15-minute | Context | The session's dealing range, the draw, HTF arrays in play, premium/discount of the range | Enter. The 15M chart votes; it doesn't click. |
| 5-minute | Structure | The swings that define valid sweeps, the pullback legs, the session's FVGs | Improvise structure the 15M doesn't support |
| 1-minute | Execution | The sweep wick, the displacement close, the entry FVG and its CE | Form opinions. The 1M executes the stack's verdict — it has no vote on direction. |
The stack's discipline is hierarchy: information flows down, never up. The 15-minute chart's verdict (direction and destination) is set before the killzone and does not get renegotiated by 1-minute candles, no matter how persuasive they look. When the 1M prints a beautiful bullish sequence against a 15M chart that is mid-delivery lower, the beautiful sequence is what other people's stop-outs look like just before they happen. The scalper's real skill is not fast clicking — it is refusing the eighty percent of 1M "setups" the stack's upper floors never sanctioned.
The 1-Minute Sequence — The Method at Minimum Scale
Step one — the target picks the trade. Scalps are chosen backwards, from the objective: the nearest meaningful pool in the delivery's direction — a session high or low fragment, the opposing side of the 15M range, an untested 15M FVG, yesterday's high or low if it sits close. If the nearest worthwhile target is too far for a scalp's hold time or too near to pay the cost toll, there is no trade, regardless of what patterns form. Ten to twenty-five NQ points; five to twelve pips on a major. The prize defines the game.
Step two — the sweep that funds it. Inside the window, the entry waits for a 1-minute raid of a 5-minute-significant level against the delivery: the pullback pressing through a 5M swing, collecting its stops. The qualifier matters — a 1M wick through a 1M swing is noise eating noise; the sweep must take liquidity the 5M chart can see, because that is the order flow large enough to fuel the leg to the target.
Step three — displacement and the gap. The reversal must announce itself: a 1M candle (or two) that closes decisively back through the short-term structure with range and body — displacement, miniature but unmistakable — leaving behind the 1M fair value gap that becomes the entry. Limit order at the gap (its consequent encroachment for the strict version), stop beyond the sweep wick — ticks away, which is the entire appeal — and the target already chosen in step one. No displacement, no trade: a slow drift back from the sweep is the market declining to commit, and scalps have no time to wait for commitment to develop.
Step four — management on rails. A scalp's lifespan is minutes, which means management must be decided before entry, not during: first partial at the halfway pool if one exists, the rest at the target, stop to breakeven only after displacement continues (a new 1M leg beyond the entry gap), and a time stop — if the trade hasn't reached its first objective by the end of the macro window, the window that justified it is over, and so is the trade. The 1-minute chart is a terrible place to have feelings; rails exist so nothing is decided while price is moving.
Every walkthrough in this guide happens inside a macro window, and that is not a coincidence — the macros are when the algorithm runs the liquidity errands scalps harvest. The dedicated guide maps every window and what each one tends to do.
Read the Macro Times Guide →Instruments — where the toll is payable. The cost arithmetic also chooses the market. Scalping lives where the spread is a rounding error against normal 1-minute range: NQ and ES in their regular sessions, EUR/USD and GBP/USD in theirs, gold during London and New York. It dies quietly everywhere else — exotic pairs whose spread is a third of the target, thin crypto alts where the 1M "sweep" is one market order's slippage, index CFDs marked up past the math's tolerance. The test is mechanical: if the all-in round-trip cost exceeds about fifteen percent of the nearest realistic target, that instrument doesn't have a scalping business in it, whatever its chart looks like. The professionals' short list is short for accounting reasons, not aesthetic ones.
The news filter. One more time rule, learned expensively by everyone eventually: scheduled releases suspend the sequence. The 8:30 data drops, CPI mornings, FOMC afternoons — these windows print sweeps and displacement that are spread-blowout and repricing, not deliveries; stops widen invisibly, fills degrade, and the 1M structure the sequence depends on becomes fiction for several minutes. The rule is a standdown from two minutes before any red-calendar release until a clean 1M structure re-forms afterward — usually five to ten minutes. The 9:50 macro after an 8:30 release is fair game; the 8:30 release itself is a different profession.
NQ Walkthrough — Eleven Minutes Inside the 9:50 Macro
The stack's verdict, set by 9:45: bias long, the 15M chart mid-delivery toward the session buyside — the 9:15 high cluster at 21,758 — with the morning's dealing range 21,684–21,746 and price pulling back from the 9:38 push. The 5M chart marks the pullback's landing zone: the 9:22 swing low at 21,712, with a 5M FVG just beneath at 21,704–21,710. Target if long: the 21,758 pool, roughly 40 points above the zone — ample toll coverage. Now the only missing ingredient is time, and time arrives at 9:50.
9:52 — the sweep: the macro opens with the errand: a 1M drive presses through the 5M swing to 21,706, into the 5M gap, collecting the pullback buyers' stops. 9:54 — displacement: a single 1M candle with conviction closes back above 21,719, leaving its FVG at 21,710–21,715. 9:56 — the entry: long 21,713 at the gap, stop 21,703 under the sweep wick — ten points of risk against the 45-point pool. 10:03 — delivery: the macro's second push runs the stack: partial at 21,738 (the range's upper fragment), and the 9:15 cluster prints at 10:07 — exit 21,755 for +42 on the runner, 2.6R blended, eleven minutes in the trade. At 10:10 the window closed and the tape went quiet, right on schedule. The trade never existed outside the macro, and neither did the trader.
EUR/USD Walkthrough — The London Version
The stack at 3:00 AM ET: bias short — the daily chart owes a visit to yesterday's low at 1.0844 — and the 15M chart shows London's Judas already sold from the 2:15 pop: delivery lower, in progress, destination 25 pips below. The 5M chart marks the retracement structure: the 2:40 swing high at 1.0871, with a 5M FVG at 1.0869–1.0873 overlapping it. The London macro at 3:10 approaches with the setup pre-drawn: if the window runs the 5M swing's stops into the gap, the sequence is live, target the 1.0844 pool.
It runs the errand at 3:12: a 1M spike through 1.0871 to 1.0874 — swing swept, gap tapped — and at 3:15 a 1M displacement candle closes back below 1.0866, leaving its own gap at 1.0868–1.0871. Short 1.0869 at 3:17, stop 1.0877 above the spike: eight pips of risk against a 25-pip address. The delivery is London doing what London does: 1.0855 by 3:34 (partial at the mid-pool), and yesterday's low prints 1.0843 at 3:52 — exit 1.0846, 2.2R in thirty-five minutes. One trade, pre-drawn, window-triggered, address-targeted — and the rest of the London session was, correctly, television.
Frequency Governance — The Rules That Keep the Math Alive
Two or three trades, capped in advance. The cap is written before the session and survives contact with any tape whatsoever. It exists because of the cost arithmetic above and because of what every honest scalping journal shows: the first window-qualified trades carry the expectancy; the impulsive additions pay the broker. A hard daily stop — one to one-and-a-half R — closes the platform, not the position size. And the pace rule: after any stop-out, no re-entry inside the same macro window. The 1-minute chart's cruelest gift is how quickly it offers a revenge trade; the window's closing bell is the built-in cooling-off period, used as designed.
Journal by window, not by day. The unit of scalping performance is the window: which macros were traded, which were correctly skipped, what the sequence looked like in each. A month of window-tagged entries reveals the only statistic that matters — which windows pay this trader — and the answer is personal: some traders' logs are carried entirely by 9:50–10:10; others never make the morning macros work and live off the London window. The method allows the specialization; the journal reveals it. And note what this article deliberately is not: a daily-target framework. Fixed daily pip goals belong to a different family of session models — a scalp's only quota is the quality of its windows.
Common Scalping Mistakes
Scalping the dead hours. The 1M chart prints sequence-shaped noise all day; outside the killzones and macros there is no delivery behind it. This single filter — time — separates scalping from churning, and it is the first thing abandoned in a losing streak, which is why losing streaks compound.
Letting the 1-minute chart vote on direction. The stack's hierarchy is absolute: 15M decides, 5M locates, 1M executes. A persuasive 1M pattern against the 15M delivery is not a signal; it is bait — and it is precisely what the sweep in someone else's sequence looks like from the inside.
Sizing up because "it's only a scalp." The tight stop seduces traders into sizes their daily stop can't survive twice. The wick stop being ten points away doesn't make the trade safer — it makes the position bigger, and two full-size stop-outs inside one macro window is how a month's progress dies in twenty minutes. Risk per scalp is the same fraction of equity as any other trade; the tight stop buys a better R multiple, not a bigger bet.
Trading the window's echo. The macro ends at 10:10, and at 10:14 price does something exciting. That move belongs to no window and carries no errand — chasing it is trading the memory of an edge. The schedule that giveth is the schedule that closeth; the next appointment is 10:50.
Frequently Asked Questions
What is the ICT scalping strategy in one sentence?
What timeframes should I use?
When are the best times to scalp?
How many trades per day, and what risk?
What are realistic scalping targets?
Is scalping the right starting point for new ICT traders?
1 — Same method, minimum scale: bias and draw govern; the 1M chart only supplies coordinates, and it never votes on direction. 2 — Time is the first filter: killzones only, macro windows preferably — about ninety minutes of valid scalping exists per day, and the dead hours are where accounts churn. 3 — The target picks the trade: nearest pool first, then the 1M sequence — sweep of a 5M-significant level, displacement, gap entry, stop ticks past the wick. 4 — The math is the discipline: costs eat a tenth of every prize, so two or three window-qualified trades, a hard daily stop, no re-entry after a stop-out, and no trading the window's echo.
We logged 214 rules-qualified 1M sequences across six weeks of NQ mornings — every setup where a 5M-significant sweep plus 1M displacement completed inside a killzone — and tagged each by clock. The windows were not created equal: sequences inside the 9:50–10:10 and 10:50–11:10 macros reached their first target before the stop 68% of the time (average 2.1R), killzone-but-non-macro sequences managed 55% (1.6R), and the 41 sequences that formed outside killzones entirely — logged but never traded — would have paid just 39%. Same pattern, three different businesses, sorted purely by time of day. The cost drag was as advertised: at our all-in ~1.4 NQ points per round trip, the macro cohort kept 91% of its gross expectancy while a simulated eight-trade-per-day version of the same edge kept 61% — frequency taxed harder than losing did.
The management data earned its own rule. Time-stopped trades — flat when the window closed without the first target — salvaged an average of +0.3R versus −0.4R for the same situations held "a few more minutes"; the window's bell outperformed every discretionary hold we tested. Breakeven moves made before continuation displacement got wicked out of eventual winners 31% of the time versus 9% when the rails were respected. And the single most expensive line in the entire log was the re-entry: second attempts inside the same macro after a stop-out won just 34% and averaged −0.2R — the revenge trade, measured. The best week in the sample, for what it's worth, contained six trades. The worst contained nineteen.