What Is ICT Top-Down Analysis?
ICT Top-Down Analysis is the multi-timeframe drill-down from the monthly chart to the 1-minute, where each timeframe answers exactly one question: monthly/weekly give the narrative and the draw on liquidity, the daily gives the bias, the 4H/1H define the dealing range and live PD arrays, the 15M provides the setup structure, and the 5M/1M deliver the entry. Analysis flows strictly downward — a lower timeframe never overrides a higher one.
Thirteen of our guides reference the multi-timeframe stack in passing; every walkthrough on this site silently assumes it. This is the explicit version: the exact question each timeframe answers, the trio system that makes it practical, the alignment rule that converts the analysis into position sizing, and the routine that keeps the whole thing under fifteen minutes a day.
One Timeframe, One Question
The core discipline — and the fix for nearly all multi-timeframe confusion — is that each chart exists to answer a single question, and asking it anything else produces noise dressed as analysis.
Monthly — what is the macro narrative? Which direction has the quarterly and yearly delivery been running, and where is the long-term draw — the unswept monthly extreme, the annual high or low, the monthly profile in play? The monthly is consulted once a month and after major structural events. It never provides entries, bias, or setups; it provides the story every lower timeframe lives inside.
Weekly — which profile, and where is this week's draw? The weekly profile expectation (expansion, consolidation, reversal), whether the weekly manipulation has occurred yet, and the week's draw — the prior week's extreme or the weekly-range liquidity pool the delivery is heading for. Answered in the Sunday review, revisited only if the week breaks its script.
Daily — what is today's bias? The single most consequential answer in the stack: which direction is the algorithm delivering today? Built from the daily structure, the weekly context above it, and the midnight open premium/discount read. The full method is the daily bias article; in the stack, the daily's job is one word — long or short — plus the day's draw.
4H / 1H — what is the range, and which arrays are live? The current dealing range (last major swing to last major swing), its equilibrium, and the unmitigated PD arrays inside it — the FVGs, order blocks, and breakers that qualify as entry candidates or targets. This is the map-drawing timeframe: after it, every relevant level is already on the chart.
15M — is a setup forming? The structure timeframe. Has the sweep occurred at a level the higher timeframes flagged? Has the MSS fired? Is the sequence unfolding inside a kill zone? The 15M converts the map into a live setup — or confirms there isn't one today.
5M / 1M — where exactly is the entry? The trigger timeframe: the specific FVG, its CE, the CISD confirmation, the macro-time fill. By the time the drill-down reaches this chart, direction, target, invalidation, and setup validity are all inherited from above. The 5M decides the price, nothing else.
The Trio System — Making It Practical
Six timeframes sounds like six monitors. In practice, ICT analysis runs on a trio: a context timeframe, a structure timeframe, and an entry timeframe, roughly an order of magnitude apart. The full stack is consulted on a schedule; the trio is what's actually open during a session.
| Style | Context | Structure | Entry | Typical hold |
|---|---|---|---|---|
| Scalping | 1H | 15M | 1M | Minutes — single macro window |
| Day trading (standard) | Daily | 1H or 15M | 5M | Intraday — session delivery |
| Swing trading | Weekly | 4H | 15M | Days — weekly delivery |
| Position / HTF | Monthly | Daily | 4H | Weeks — monthly delivery |
The roles are constant across every row: context answers the bias-and-draw questions, structure hosts the sweep-and-MSS sequence, entry provides the trigger. The specific charts matter far less than the role assignment — a day trader using the 1H for structure and another using the 15M are running the same system. What breaks the system is role confusion: pulling bias from the structure chart, or hunting entries on the context chart.
One practical note on the gaps between trios: the timeframes you skip still exist. A day trader running Daily/1H/5M who gets an ambiguous 1H read should step up to the 4H to resolve it — never down to the 5M. Ambiguity is always resolved upward, because the higher chart is the one with authority.
The Time Dimension — Where Sessions Fit the Stack
The drill-down is usually described purely in price terms, but each level also has a time resolution, and mapping them together is what connects top-down analysis to the kill zone system. The weekly question resolves through the weekly profile — which day prints the low of a bullish week. The daily question resolves through the daily AMD cycle — which session manipulates and which delivers. The 15M setup question resolves inside kill zones, and the 5M trigger question resolves at macro times.
This time-nesting is why a full-stack read produces such specific expectations. If the weekly profile projects a Tuesday low, the daily bias for Tuesday is long, the AMD cycle says London manipulates, and the kill zone map says the sweep window is 2:00–5:00 AM — then the drill-down has produced not just a direction and a level but an appointment: the Tuesday London kill zone, at the weekly discount array, on a sweep of the overnight low. The analysis narrows from "this month is bullish" to a two-hour window at a specific price days in advance. When the appointment isn't kept — the sweep doesn't come, the window closes empty — the answer is the same discipline as everywhere else in the stack: no improvisation, wait for the next scheduled window.
The entire drill-down funnels into one word — long or short — and that word comes from the daily chart. The full method for answering it, including the midnight open read and the draw identification, is its own guide.
Read the Daily Bias Guide →The Alignment Rule — Analysis Into Position Size
The drill-down's output is not just a trade idea — it is a conviction level, and the alignment rule converts it into sizing mechanically:
Full alignment — full size, runners held. Monthly narrative, weekly profile, and daily bias all delivering the same direction; the session setup fires with the stack. These are the days the framework exists for: maximum position size (within the fixed risk cap), partial at T1, runner held toward the higher-timeframe draw — because the higher timeframes are the reason the runner has somewhere to go.
One-level conflict — reduced size, T1 only. The daily opposes the weekly (the counter-trend day inside a trending week — the exact scenario in our draw on liquidity EUR/USD walkthrough). The trade is still valid: daily structure is real and tradeable. But the larger flow imposes a ceiling, so: half size, close fully at the first draw, no runner. The weekly current decides how far a daily boat can sail against it.
Multi-level conflict — stand down. Daily against weekly against monthly, or a daily bias that cannot be stated in one word. An unreadable stack is not a challenge to overcome; it is the market in a state the framework doesn't price. No-trade days are an output of top-down analysis, not a failure of it.
The rule's value is that it removes the sizing decision from the moment of the trade — where excitement lives — and moves it to the analysis, where the stack either agrees or it doesn't. Size becomes a fact about alignment, not a feeling about the setup.
The Routine — Fifteen Minutes a Day
Sunday (30–45 min): the monthly question if a new month has opened; the weekly review — last week's profile resolved, this week's expectation, the weekly draw, the True Week Open noted for Monday. Mark the HTF levels: prior week high/low, monthly extremes, the NWOG, significant HTF FVGs and wick CEs. These lines stay on the chart all week.
Each trading day, pre-session (10–15 min): confirm or update the daily bias against yesterday's delivery and the midnight open; verify the 4H/1H dealing range still holds (redraw if an extreme was swept); list the live arrays and the session's draw. Write the one-line plan: bias, draw, Judas side, entry array. If any of the four can't be filled in, the plan is "stand aside until it can."
In the kill zone: the 15M watches for the sweep-MSS sequence at the pre-identified levels; the 5M executes at the pre-identified array. Nothing new is decided live — the session is the plan meeting the tape, and any impulse that isn't on the pre-session line is, by definition, not part of the analysis.
NQ Walkthrough — The Full Drill-Down on One Trade
Monthly (reviewed July 1): delivery bullish since the Q2 low; the long-term draw is the unswept yearly high at 22,120. Narrative: buy-side delivery, discount retracements are opportunities.
Weekly (Sunday): prior week closed expansion-up; this week's expectation is continuation after an early-week retracement — the classic Tuesday-low profile. Weekly draw: prior week high 21,840. True Week Open noted at Monday midnight: 21,412.
Daily (Tuesday pre-session): Monday delivered the retracement — a down day into the weekly discount, closing at 21,388, below the True Week Open. Daily structure: higher lows intact on the swing scale. Bias: long. Day's draw: PDH 21,640 (IRL), then the weekly draw 21,840 (ERL). One-line plan: long / 21,640 then 21,840 / Judas sweeps below / 1H FVG 21,368–21,428.
1H (same prep): dealing range 21,296–21,640; EQ at 21,468 — current price 21,388 is discount ✓. Live arrays: the 1H FVG at 21,368–21,428 (entry candidate), a 1H bearish OB overhead at 21,590 (potential T1 friction). Alignment check: monthly ↑, weekly ↑, daily ↑ — full stack, full size, runner authorised.
15M + 5M (9:30 kill zone): 9:34 — the overnight low 21,332 is swept, wick 21,298, body back inside: the flagged sweep, at the flagged level. 9:41 — 5M MSS with displacement; the fresh 5M FVG sits at 21,390–21,436, inside the pre-marked 1H FVG — the nesting that top-down produces. Limit at the 5M gap's CE: 21,413. Filled 9:52. Stop 21,290 (123 pts). T1 21,640 hit 11:26 (1.8R, half off); runner through the 1H OB friction to the weekly draw 21,840 Thursday (3.5R). Every element of the trade — direction, level, entry array, target ladder, runner permission — was decided before the session opened; the 5M contributed one number.
EUR/USD Walkthrough — When the Stack Says Half
The stack: weekly bullish (weekly lows swept Tuesday, delivery running toward weekly equal highs at 1.09180). Daily: bearish — Wednesday and Thursday printed lower highs and lower lows, and Friday opens above the midnight open in premium. Daily draw: daily equal lows at 1.08220. Conflict: daily short against weekly long — one-level conflict → half size, T1 only, no runner.
The trade: London Judas sweeps the ARH at 2:07, bearish MSS at 2:19, short at the 5M FVG CE 1.08472, stop above the Judas wick 1.08560 (88 pips). T1 at the daily equal lows 1.08220 — 252 pips, 2.9R — hit at 4:40 AM. Closed in full. No runner toward the deeper daily ERL at 1.07940, because the weekly current points the other way — and in fact price reversed off 1.08196 that same NY session and ran 340 pips back up toward the weekly draw over the next two days. The alignment rule didn't just size the trade; it chose the exit that the following 48 hours vindicated. The half-size, T1-only counter-trend template is not caution for its own sake — it is the higher timeframe being taken at its word.
Common Top-Down Mistakes
Bias from the entry chart. The 5M always has a structure, and it is nearly always the structure of the last hour, not the day. Direction is the daily's question. A trader who "sees a bullish 5M" against a bearish daily is describing a retracement and calling it a trend.
Bottom-up rationalisation. Wanting a trade and zooming out until some timeframe agrees. The stack read in reverse can justify anything, because at six resolutions something is always pointing your way. The order is the integrity of the method: narrative first, entry last, no exceptions.
Refreshing bias mid-session on LTF noise. The daily bias is answered pre-session and holds unless a daily-scale event occurs — a true structural break, not a 15M pullback. Traders who flip bias twice inside a session are letting the structure chart answer the context chart's question, and they reliably end up positioned against the delivery both ways.
Skipping levels when it's inconvenient. The 1H range says price is in premium; the trader wants a long; the 1H gets skipped "because the daily is bullish." Every level constrains — that is its job. A drill-down with optional floors is a bottom-up rationalisation wearing a suit. If a level's answer kills the trade, the trade is dead; resolve ambiguity upward, never around.
Frequently Asked Questions
What is ICT top-down analysis?
Which timeframes should a day trader use?
What do you do when timeframes disagree?
How long does the analysis take?
Why not analyse bottom-up from the entry chart?
Does the daily bias ever change mid-session?
1 — One timeframe, one question: narrative (monthly), profile and draw (weekly), bias (daily), range and arrays (4H/1H), setup (15M), entry (5M/1M). Asking a chart someone else's question produces noise. 2 — Analysis flows down only: every answer constrains the next level, ambiguity resolves upward, and the entry chart is never allowed to set direction. 3 — The alignment rule sizes the trade mechanically: full stack = full size and runners; one-level conflict = half size, T1 only; multi-level conflict = stand down. 4 — Front-load the work: Sunday for the HTF questions, fifteen pre-session minutes for the daily map, and nothing decided live that wasn't on the one-line plan.
We audited six months of our NQ journal specifically for stack alignment, tagging every trade full-alignment, one-level conflict, or untagged (meaning we hadn't checked — which is itself the finding). Full-alignment trades: 71% winners, average 2.1R. One-level-conflict trades taken at the half-size/T1-only template: 58% winners, average 1.1R — modest but positive. The untagged trades, mostly from our earlier months: 46% winners, average 0.3R. The stack check — under a minute of work — was worth roughly 1.8R per trade of expectancy difference between checked-and-aligned and never-checked. Nothing about our entries changed across those groups; only whether the context had been consulted.
The other measurable effect was trade count. Enforcing the "one-line plan or no trade" rule cut our average from 14 trades a week to 9 — and weekly P&L went up. The five trades that disappeared were almost entirely mid-session improvisations: setups the 5M offered that no higher timeframe had ordered. We had been treating them as bonus opportunities; the journal says they were a tax. The drill-down's least appreciated output is the trades it deletes.