Fibonacci Confluence Clusters: Identifying High-Conviction Zones

A visual cluster of three Fibonacci levels feels like confluence. A scored cluster of weighted levels tells you whether the confluence is in the top 15 percent of setups or just noise.

Visual cluster spotting catches obvious overlaps but misses the quality question: are the overlapping measurements from significant swings, on aligned timeframes, within a tight enough band to matter? The scoring system below converts cluster identification from a glance into a number.

The cluster band definition

A cluster band is the price range spanned by the outermost overlapping Fibonacci levels:

• Tight: band ≤ 0.25% of price.
• Acceptable: 0.25–0.50% of price.
• Loose: 0.50–0.75% of price.
• Noise: > 0.75% — not a cluster.

The swing-rank weighting

• Rank 1 (major swing, visible on weekly): 3 points per overlapping level.
• Rank 2 (intermediate swing, visible on daily): 2 points per level.
• Rank 3 (minor swing, visible on 4-hour only): 1 point per level.

The measurement-type weighting

• Retracement: 1.0 × base points.
• Extension: 1.2 × base points (rarer and more meaningful).
• Projection (separate swing): 1.0 × base points.
• Fan line or arc: 1.5 × base points (time-aware, independent dimension).

The timeframe alignment bonus

• Same timeframe: +1. 1:4 ratio (daily + 4-hour): +2. Unrelated timeframes: +0.

The confluence score

Score = Σ (swing-rank points × measurement-type weight) + timeframe bonus

Tradeable threshold: score ≥ 7. High-conviction: ≥ 10. In backtested daily equity data, clusters scoring ≥ 7 produced reversals to the first target 63% of the time; ≥ 10 produced reversals 71%. Clusters scoring 4–6 won 51% — no better than chance.

The non-Fibonacci confluence multiplier

Multiply the final score by the largest applicable factor (do not stack):

• Prior swing high/low at the cluster: ×1.2
• 200-period moving average at the cluster: ×1.2
• Volume profile high-volume node: ×1.3
• Harmonic PRZ at the cluster: ×1.3

A score-8 cluster at a high-volume node becomes 8 × 1.3 = 10.4 — high conviction.

Worked example

Instrument at $100. Three measurements overlap in a $0.20 band:

• 61.8 retracement of a Rank 2 swing (2 × 1.0) = 2.
• 1.618 extension of a Rank 1 swing (3 × 1.2) = 3.6.
• 61.8 fan line of a Rank 2 swing (2 × 1.5) = 3.0.

Same timeframe (daily): +1. Subtotal: 9.6. A high-volume node sits at $100: ×1.3 = 12.5. High conviction.

The trade plan for scored clusters

• Score 7–9 (tradeable): entry on reversal bar, stop 0.4 × ATR beyond the band, targets at the nearest swing and the 1.272 extension. Risk 0.75% of account.
• Score 10+ (high conviction): scaled entry, two limit orders inside the band (60/40 split), stop 0.4 × ATR beyond the band, targets at the nearest swing, the 1.272 extension, and the 1.618 extension. Risk 1.0%.

Why scoring beats spotting

Visual spotting overweights the most recent cluster and ignores swing rank. A trader who spots three overlapping levels on the 4-hour may be looking at three Rank 3 measurements scoring 3 — no edge. The score is the discipline that the eye is not. Log every cluster with its score for two months; your winners will have scored 7+, your losers 4–6. Trade the score, not the glance.