Fibonacci Confluence

The densest cluster of retracement levels across recent swing legs — where 38.2/50/61.8% levels from different legs stack up, reported as the cluster mean price and its strength (member count).

Quick reference

Item	Value
Family	Fibonacci
Input type	Candle (uses high, low)
Output type	FibConfluenceOutput (price, strength)
Output range	price a market price; strength a count (>= 1)
Default parameters	none (swing threshold 5%, 6-pivot history, baked)
Warmup period	3 (two legs)
Interpretation	Strongest stacked support/resistance zone

Formula

for every leg among the last six pivots, take its 38.2/50/61.8% levels;
group levels within 3% (relative) of each other;
report the densest group:
  price    = mean of the group's levels
  strength = number of levels in the group

The 3% grouping tolerance is the shared LEVEL_TOLERANCE family constant. See crates/wickra-core/src/indicators/fib_confluence.rs.

Parameters

None. The swing threshold 0.05, the six-pivot history, the three ratios and the 3% cluster tolerance are baked-in constants. FibConfluence::new is infallible.

Inputs / Outputs

const _: fn(&mut wickra::FibConfluence, wickra::Candle) -> Option<wickra::FibConfluenceOutput> =
    <wickra::FibConfluence as wickra::Indicator>::update;

• Python. update((o,h,l,c,v,ts)) → (price, strength) or None; batch(high, low) → (n, 2) Matrix (NaN warmup).
• Node. update(high, low) → { price, strength } or null; batch(high, low) → flat number[] length n*2.
• WASM. update(high, low) → object (price/strength) or null.

Warmup

warmup_period() == 3. At least two legs (three pivots) are needed for levels to cluster; before that update returns None. Pinned by tests accessors_and_metadata and no_output_before_two_legs.

Edge cases

• Picks the densest cluster when levels from two legs overlap within tolerance (test picks_the_densest_cluster).
• reset clears all state (test reset_clears_state).
• Streaming equals batch (test batch_equals_streaming).

Examples

Rust

use wickra::{Candle, FibConfluence, Indicator};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Legs 200->100 and 100->160: their 38.2% levels (138.2 and ~137.08)
    // cluster within 3% → strength 2, mean ~137.64.
    let bars = [
        (199.8, 200.0, 199.8, 199.8),
        (100.0, 198.0, 100.0, 100.0), // confirm high @200
        (101.0, 160.0, 101.0, 101.0), // confirm low  @100
        (144.0, 158.4, 144.0, 144.0), // confirm high @160 (two legs)
    ];
    let mut fib = FibConfluence::new();
    let mut last = None;
    for (ts, (o, h, l, c)) in bars.iter().enumerate() {
        last = fib.update(Candle::new(*o, *h, *l, *c, 1.0, ts as i64)?);
    }
    let v = last.unwrap();
    println!("{:.2} {}", v.price, v.strength); // 137.64 2
    Ok(())
}

Python

import wickra as ta

bars = [
    (199.8, 200.0, 199.8, 199.8, 1.0, 0),
    (100.0, 198.0, 100.0, 100.0, 1.0, 1),
    (101.0, 160.0, 101.0, 101.0, 1.0, 2),
    (144.0, 158.4, 144.0, 144.0, 1.0, 3),
]
fib = ta.FibConfluence()
print([fib.update(b) for b in bars][-1])  # (137.64, 2.0)

Node

const wickra = require('wickra');
const fib = new wickra.FibConfluence();
const bars = [[200.0, 199.8], [198.0, 100.0], [160.0, 101.0], [158.4, 144.0]];
let last = null;
for (const [h, l] of bars) last = fib.update(h, l);
console.log(last.price.toFixed(2), last.strength); // 137.64 2

Streaming

fib = ta.FibConfluence()
for o, h, l, c, v, ts in candle_feed:
    cl = fib.update((o, h, l, c, v, ts))
    if cl is not None and cl[1] >= 2:
        pass  # multiple legs agree on cl[0] — a strong zone

Interpretation

1. Strength in agreement. A price where retracements from several legs coincide is a higher-conviction zone than any single fib level.
2. strength is conviction. A strength of 1 means no real confluence; 2+ means independent legs point at the same area.

Common pitfalls

• Strength 1 is weak. With non-overlapping legs the "densest" cluster is just one level — check strength before trusting the price.
• Window-bounded. Only the last six pivots feed the search, so very old levels drop out.

References

• Fischer, R. Fibonacci Applications and Strategies for Traders (1993); Bulkowski, T. Encyclopedia of Chart Patterns (2005).

See also

• FibRetracement, AutoFib, GoldenPocket.
• Indicators-Overview — full taxonomy.