ICT Confluence Framework
Three layers is a setup. Five is A+. One is a suggestion. Most traders die in the gap between.
The Confluence Framework
ICT concepts are tools. The edge is in how you stack them — and which setups you skip.
Confluence
Also known as: Stacking, Setup grading, Layered confirmation
Confluence is the stacking of independent signals — each pointing in the same direction — before risking capital. A single ICT concept (an FVG by itself, an OB by itself, a sweep by itself) is a suggestion. Three or more aligned concepts is a setup. Five or more is an A+. The skill isn't finding patterns — it's refusing to act when fewer than three line up.
Imagine you're a witness in a courtroom. One person saying something happened is suggestive. Three people independently saying the same thing is evidence. Five people from different angles saying the same thing is conviction. Confluence is the trader's version of independent witnesses. A sweep alone could be noise. A sweep plus a displacement plus an FVG plus an MSS plus your HTF bias all pointing the same way? Now you have a case.
Prereqs: Market structure · Liquidity · Fair Value Gaps · Order Blocks · MSS / CHoCH
The institutional logic behind confluence: large participants need multiple inefficiencies to fill size without showing their hand. A single FVG might be retail-induced or coincidence. But when an FVG forms after a deliberate sweep of liquidity, in displacement out of a HTF level, in the direction of HTF bias, with LTF confirmation — that's the footprint of a real positioning. The confluence stack is your signal-to-noise filter. The trader's job is not to find more setups; it's to refuse anything below A+ on the days they have the discipline to walk away.
Industry-tested 5-layer stack. Don't act unless 3+ are present. A+ requires 5+.
- 1Layer 1 — HTF bias
Daily / 4H structure (HH+HL or LH+LL) plus draw-on-liquidity. Counter-bias setups never qualify as A+, full stop.
- 2Layer 2 — HTF zone or level
Premium / discount of the dealing range, or a HTF FVG / OB. The setup must occur inside meaningful HTF context, not in mid-range chop.
- 3Layer 3 — Liquidity sweep
A clean sweep of an obvious pool (equal highs/lows, prior session H/L, daily H/L). The sweep proves the stops are gone — the path is clear.
- 4Layer 4 — Displacement + FVG / OB
Body-driven move out of the swept area, leaving an inefficiency (FVG) or an institutional candle (OB). This is the footprint. Without displacement, layers 1-3 are decoration.
- 5Layer 5 — LTF confirmation (MSS / CHoCH)
On your entry timeframe, structure must shift in the trade direction. Body close above the most recent LH (long) or below the most recent HL (short). This is the trigger.
- 6Bonus — Time of day (kill zone)
Setups during London Open, NY Open, or NY PM kill zones have higher base rates than mid-Asia or post-NY-close attempts. Time isn't a layer per se but it is a quality multiplier.
| Action | Description | |
|---|---|---|
| A+ (5+ layers) | Full size, full conviction | HTF bias + HTF zone + sweep + displacement + LTF MSS, in a kill zone. Rare — maybe 1-2 per session. |
| A (4 layers) | Full size | Missing one component (often kill zone). Still tradable with full plan. |
| B (3 layers) | Half size, tighter stop | Marginal setup. Trade only if your daily R is unspent and quality of present layers is exceptional. |
| C (≤2 layers) | **Skip** | Single concept setups. The market gives many of these and most resolve as chop. Skipping them is the edge. |
Multi-Confluence A+ Setup
Move runs to opposing liquidity. Five confluences aligned: HTF bias + sweep + displacement + FVG + MSS.
A bullish A+ unfolding bar by bar. Watch the 8 phases: HTF bias → HTF level → sweep → displacement → FVG → return → MSS → target. Five layers of confluence aligned.
Setup: **Layer 1 (Bias):** Daily HH/HL. Bias bullish. **Layer 2 (Zone):** Price into 4H FVG that previously held as support. **Layer 3 (Sweep):** Asia low at 199.20 swept by 14 pips at London open. **Layer 4 (Displacement):** Strong 5m bullish marubozu out of the sweep, leaves a 5m FVG. **Layer 5 (MSS):** 1m close above the most recent LH at 199.45.
Entry: Buy on retest of 5m FVG at 199.34, after the 1m MSS confirmation.
Stop: 199.05 — below the sweep low + 8 pip buffer.
Target: 199.78 (Asia high) → 200.20 (1H FVG above).
Five independent confluences, in a kill zone, with HTF bias backing all of them. Stop is structure-defined, target is liquidity-defined.
Outcome: T1 +1.5R, trail to BE, T2 stopped at break-even after London close consolidation.
Setup: Trader sees: a 5m bearish OB and a 5m FVG to the upside. They label this 'two confluences' and prepare to short. **Layer test:** the FVG is the displacement out of the OB — these are the same event from two angles. **Bias:** Daily is bullish. **Sweep:** No clean sweep — just chop. **MSS:** None.
Entry: Skipped — only 1 unique layer present, and HTF bias contradicts.
This is the most common B-grade trap. The chart looks 'ICT' but the framework filter rejects it. The trader who skips this and waits for a real A+ outperforms the trader who trades it.
Outcome: Logged as 'avoided' in the journal. Skipped trades that would have lost are wins.
Mistake: Mistaking correlated confluences for independent confluences (e.g., 'I have an FVG, an OB, and a breaker block — that's 3 layers!').
Why it hurts: An FVG and the OB that created it are the same event from different angles — not three independent witnesses. Stacking redundant evidence inflates conviction without adding signal. You'll feel certain about setups that are actually B or C grade.
Fix: Layer test: each piece of confluence must come from a different category — bias, zone, liquidity, displacement, structure shift, time. Two FVGs from the same impulse = one layer, not two.
Adding a 6th, 7th, 8th layer doesn't keep improving your win rate. Diminishing returns set in fast. The realistic curve: 1 layer ~ random, 2 layers ~ slight edge, 3-4 layers ~ tradable edge, 5+ ~ professional-grade. Past 5, you're mostly building a case to yourself rather than a case to the market. Trade the 5-layer A+; don't wait for the perfect 8-layer setup that arrives once a month.
Before clicking buy/sell
- HTF bias is set in writing for todaymust
- Setup is in the same direction as HTF biasmust
- Setup occurs at a meaningful HTF level / zonemust
- Liquidity sweep visible (or in progress)
- Displacement candle confirms institutional intentmust
- FVG or OB present and unmitigated
- LTF MSS / CHoCH printed in entry directionmust
- I can name 3+ independent confluences out loudmust
- Setup is in or near a kill zone
- Stop and target predefined; R:R ≥ 1.5must
20-setup grading drill
~60 min
Open 20 historical setups across forex / gold / indices. For each, list the layers present (bias, zone, sweep, displacement, FVG/OB, MSS, kill zone). Grade each as A+/A/B/C. Score yourself: did you correctly identify the highest-grade setups in retrospect?
You can grade a setup in under 30 seconds and your B/C grades correlate with bad outcomes; A/A+ grades correlate with positive R outcomes ≥ 60% of the time.
Upgrade to unlock this block.
- ICT concepts are tools. Confluence is how you use them.
- Layer test: each confluence must come from a different category.
- 3+ independent layers = setup. 5+ = A+. ≤2 = skip.
- More layers ≠ unbounded edge. 5 is the practical sweet spot.
- Most traders use confluence to enter. Pros use it to *not* enter.