The 10-source S/R confluence merger.

One support level is a guess. Two is a coincidence. Three within a tight band is a structural level the market is actually defending. The Beginner Track introduced "structural support" and "structural resistance" without saying where they come from. This lesson opens the box. The framework computes ten different candidate levels per ticker, then collapses them via 0.15× ATR confluence merging into the few that actually matter. The output is what feeds the R:R calculation, the entry trigger, the chandelier exit, and every "is this a level" decision the framework makes. Understanding the merger explains why the dashboard treats some bounces as setups and others as noise.

Why one source is never enough

Pick any single S/R source — say, the 50-day SMA. It draws a line on the chart. Sometimes price respects it; sometimes price slices through it like nothing. By itself it's barely better than a random horizontal. The reason isn't that the 50-day SMA is bad — it's that any single line on a chart is bad. Markets don't defend math; they defend places where money is parked. A line earns its keep when it overlaps with other lines, because overlap is the marker of multiple groups of participants having put orders at roughly the same price.

Confluence is the technical word for "multiple sources pointing at the same level." The framework's job isn't to compute one perfect level — it's to compute ten cheap candidates, identify clusters where three or four overlap, and treat those as the real S/R. A 20-day SMA at $234.10, a prior-week high at $234.40, a volume-profile HVN at $234.25, and a psychological round number at $235 — that cluster is the level. Trade against it. The other six sources that don't cluster are noise for this ticker on this day.

The ten sources

The framework computes these ten S/R candidates per ticker per audit cycle:

1. Pivot points — classic floor-trader pivot R1/R2/S1/S2 derived from prior session H/L/C.
2. Prior-day high / low — yesterday's range boundaries.
3. Prior-week high / low — weekly range boundaries (more durable than daily).
4. 20-day SMA — short-term mean.
5. 50-day SMA — medium-term mean. (Often where institutional rebalances ladder in.)
6. 200-day SMA — the regime-defining line. Above = bull regime, below = bear, in retail consensus.
7. VWAP — volume-weighted average price (session). The institutional "fair value" anchor.
8. Volume profile HVN / LVN — High-Volume Nodes and Low-Volume Nodes from a 60-day volume-by-price profile. HVNs are sticky levels; LVNs are gaps price flies through.
9. Gap edges — unfilled gap boundaries from the prior 60 sessions. Markets often retest gap edges before continuing.
10. Swept highs / lows + psychological round numbers — recent swept levels (where stops were hunted) and round-number levels ($100, $250, $500, etc.) where retail order flow concentrates.

Ten sources, computed cheaply, refreshed every audit. Each one has a known failure mode — pivot points are mechanical and don't know about news; SMAs lag; VWAP resets each session; round numbers only matter near them. None is reliable alone. Together they triangulate.

The 0.15× ATR confluence merge

Once the ten candidates are computed, the merger runs. The rule:

cluster_band = 0.15 × ATR(14)

For each pair of candidate levels (a, b):
    if |a − b| ≤ cluster_band:
        merge(a, b) into a single level at weighted average

Repeat until no more merges possible.

Each merged level carries a weight = number of sources contributing.

A cluster of three levels within a 0.15× ATR band becomes one merged level with weight 3. A standalone level (no merge) has weight 1. The weight is what separates "real S/R" from "one source's opinion."

Why 0.15× ATR? Because that's the tightness band where retail order books actually concentrate. ATR is the daily volatility unit; 0.15× of that is roughly the spread between two limit orders placed at "the same level" by two different traders. Any tighter and you under-merge (treating real clusters as separate levels). Any looser and you over-merge (collapsing genuinely distinct levels into a fake mega-cluster). 0.15× is the empirical sweet spot that the framework has used since v6.5.

Reading the merged output

After merging, every level has a weight (number of contributing sources). The framework treats them in tiers:

• Weight ≥ 4 — major confluence. Multiple independent sources agree this is a level. These are the structural S/R that drive R:R calculation, primary entries, and chandelier-exit anchors. Rare; usually one or two per chart at any given time.
• Weight 3 — strong confluence. Real level, tradeable as primary support/resistance. The framework's default for "structural support" and "structural resistance" in the R:R math is the strongest weight-3 cluster on each side of price.
• Weight 2 — minor confluence. Worth marking but not strong enough alone. Often lives in the noise band between major levels. The framework surfaces these as secondary levels but doesn't trade them as entries.
• Weight 1 — single-source level. Decoration. The framework doesn't expose weight-1 levels in any decision logic. They're computed but not surfaced.

The "primary support" and "primary resistance" the dashboard surfaces are the highest-weight cluster on each side of current price. R:R is computed off those. Entry triggers fire (or refuse) based on proximity to them. Chandelier exits anchor relative to them.

What "the level held" actually means

A common retail observation: "the 50-day held." Often it didn't — the 50-day was at $234.10, the bounce happened at $234.30, and the trader rounded the bounce to the 50-day because that was the nearest line they'd drawn. The framework's confluence read tells the truer story: the bounce happened at the merged cluster of 50-day + prior-week-high + HVN at $234.25 ± 0.15× ATR. The 50-day was contributing weight, but it wasn't the level. The cluster was.

This matters because the next bounce attempt won't necessarily hold at the same place. If price comes back to $234 and the 50-day has migrated to $232 (because the SMA window slid forward), the cluster has weakened from weight 3 to weight 2. The framework reads the weakening; the trader staring at "the 50-day" doesn't.

Why ten sources, not five or twenty

Five sources isn't enough — too many days produce no confluence at all, and the framework falls back to single-source levels which are noise. Twenty starts to over-merge: with twenty candidates spread across a chart's typical 4-6 ATR range, the 0.15× band catches false clusters that aren't really there.

Ten is empirical: enough sources that confluence happens reliably (most charts have 2-4 weight-3+ clusters at any time), few enough that the merge band actually matters. Each of the ten sources contributes a different *kind* of information — pivot points are mechanical, SMAs are smoothed price, VWAP is volume-weighted, gap edges are event-driven, round numbers are psychological. The ten sources are diverse on purpose; adding more would mostly add correlated information.

What the framework does

The Trap & Structure Coach surface receives the merged level set and uses it for:

• R:R calculation — primary support and resistance from the merger feed the structural R:R math (Lesson 4)
• Entry trigger validation — the entry-trigger criteria require the candidate to be at or near a weight-3+ cluster (Lesson 7's "context not patterns" rule)
• Chandelier exit anchor — when the chandelier-exit math runs, it cross-checks against merged resistance to avoid setting trail stops at hidden cluster levels
• Trap detection — when price closes at a weight-3+ level with the wrong volume signature (Lesson 17's hidden tape), the trap chip surfaces

The merger isn't a separate surface; it's the input layer for several others. Understanding it is what makes the dashboard's S/R reads make sense.

The real lesson

Single-source S/R is folklore. The 50-day SMA, the prior-week high, the round number at $250 — none of them by themselves earn the right to be called "structural." Confluence is what earns it. The framework computes ten cheap candidates, merges them via the 0.15× ATR band, and only treats weight-3+ clusters as the levels that drive its decisions. When the dashboard tells you a setup is at "structural support," it means a confluence of three or more independent sources — not one trader's guess about a moving average.

Related: L4 — R:R floor · L17 — Hidden tape · L18 — Sweep detection