Fibonacci confluence without double counting

Prerequisite: A repeatable multi-timeframe process

On a small screen, scroll the figure sideways or open it full size.

Learning objective

Distinguish a cluster of related levels from genuinely independent information.

Confluence means that selected references lie near one another: for example a retracement near an earlier swing area. Define near using a fixed price tolerance or volatility fraction. Without a limit on grids and anchors, adding enough levels almost guarantees that some will sit close together by coincidence.

Two Fibonacci grids from overlapping moves share many underlying prices. A moving average from the same candles also shares information. Do not multiply claimed probabilities as if these were independent events. Treat the combination as one defined rule and evaluate its outcomes against a simpler baseline.

State a maximum number of grids, anchor selection and a clustering tolerance before viewing future reactions. Study whether the cluster offers useful risk distance after costs, not merely whether a wick touches it. Include cases where the price cuts straight through every nearby level and preserve the original anchors.

Common mistakes

Avoid adding indicators until a trade feels certain. Clusters discovered retrospectively suffer selection bias even when the arithmetic is correct.

Practice

Choose at most two grids and one prior area. Document anchors and tolerance, then compare a cluster rule with the prior-area-only rule using identical samples and costs.

Sources & further reading

• Fidelity — Fibonacci retracement