Visuals and Logic

This page teaches the pane. Not the pane's settings — those live on Settings. The pane: what it is trying to report, in what order to read it, and what each of the four structure layers is actually saying about the blend.

If you take one thing away from this page, make it this: the blend is the headline, the slots are the evidence, the histogram is the speedometer, and the four structure features are framing questions. None of the structure features are answers. They are not independent confirmations. They are four ways of asking something about the same line.

The mental model — convergence and divergence, bounded

Classical MACD is a convergence-divergence read. A fast MA minus a slow MA tells you whether short-term momentum is above or below longer-term momentum, and by how much. Add a signal line on top of that and you get a second read — how the convergence or divergence is itself trending. Add a histogram and you get the speed of change between them.

All three belong to a single idea: is the near past agreeing with the slightly further past, and how is that agreement evolving?

Axiom's projection keeps that idea and drops a single thing: the raw magnitude. Raw MACDs from different timeframes, different instruments, and different MA families live in different units. Stacking them unscaled is a category error. The ATR-sigmoid normalization maps each slot's MACD, signal, and histogram triple into a bounded 0..100 pane where 50 is the zero point of the transformed series. On the MACD line, that means the fast-minus-slow gap is near zero. On the histogram, it means MACD equals signal. You lose the raw magnitude. You gain comparability across slots. For most traders running more than one timeframe, that is the trade worth making.

Zero and one hundred are asymptotes, not thresholds. They are what happens when the sigmoid's tails flatten. Pinning against a boundary is a property of the transformation under a given sensitivity and volatility regime, not evidence of intensity. That is the first misread this pane has to disarm and we revisit it below.

Reading order — always

Drill this sequence into your first few weeks of use. It is the difference between reading the pane and looking at it.

1. Histogram first

The blended histogram paints columns rooted at 50. Its color encodes two things — which side of the midline the column sits on, and whether that column is taller or shorter than the previous column.

The headline on the whole pane is still the histogram. In one glance you get direction (above or below 50) and thrust (bright or faded). You do not need to read anything else to know whether the convergence-divergence relationship is accelerating or decelerating.

A histogram that is crossing from bright green to faded green is the single most useful early warning on the pane. Thrust is falling. That does not obligate direction to flip; it just says the convergence story is losing momentum.

2. Slot lines next — but only when you need them

Slots are the evidence underneath the headline. With the defaults, slot lines are hidden on slots 1–3 so the pane is not visually crowded during a first session. When you un-hide them, each slot line is the slot's normalized K — full-toned when above its own D, faded when below. Colors are assigned per slot: teal, aqua, blue, orange, yellow (slot 01 through slot 05).

Use the slot lines when you need to ask:

• Which slot drove the blend to where it is? A single high-weighted slot moving hard can swing the blend even when the other slots are steady.

• Is a higher-timeframe slot stepping on HTF boundaries, or moving intra-bar? That tells you whether the HTF slot's On Bar Close? is ON or OFF.

• Is a hidden or zero-weighted slot still doing something interesting? The slot color flips when the slot's K crosses its D — even when the line itself is hidden, the per-slot alert is tracking that state.

Slots are the layer you open when the blend's behavior is suspicious. They are not something you need to read continuously.

3. Blend — position and fill

The thicker blended fast line and the thinner blended slow line run in the middle of the pane. The tinted fill between them is colored by their own relationship: lime when the fast is above the slow, red when below. The fill's extent is geometric — distance between the two lines — but the hue carries the state cue, not a magnitude cue. A fat fill is not "more confident"; it is just further apart.

Position relative to 50 and to the reference lines at 30 and 70 is what matters here. Those reference lines are not overbought or oversold triggers. They are visual bumpers for your eye. Nothing in the indicator's logic reacts to them.

4. Donchian channel

Steplines in silver wrap the blend at its own highest and lowest values over the Donchian length (default 20 bars). An optional basis stepline sits between them.

A Donchian upper touch tells you the blend is printing a fresh high in its own recent lookback. That is not a price breakout. That is not a bullish entry. The blend has simply reached its own range edge inside its own window. Sometimes that coincides with price doing something meaningful. Often it does not. The Donchian channel is framing, not a trigger.

Watch the channel widen as the blend prints fresh extremes and pinch as the blend quiets down. Both regimes are information. A very tight channel means the blend has been range-bound relative to itself — a posture that can persist, that can resolve either direction, and that is not predictive on its own.

5. BBWP column

BBWP columns at the bottom of the pane paint the current bar's percentile rank of the blend's own Bollinger band width, measured over the percentile-rank lookback (default 252 bars). The column is aqua when the percentile is at or above the threshold (default 50), blue when below.

This is a blend-derived read. It is not price volatility. A choppy, violent price chart can print a low BBWP column if the blended MACD line has been in a compressed width regime by its own history. A price chart that looks quiet can print a high BBWP column if the blend's width has been building. Every time you notice BBWP, remind yourself: this is the blend's width against its own past, not the market's volatility.

When BBWP compresses (low column), the blend has been tightening. When it expands (high column), the blend has been widening. Both are descriptions of the blend, and both are useful framing for how to weight the other features on the pane. Neither is an entry.

6. Divergence triangle — last, with the most caution

Lime up-triangles at the bottom of the pane (pane-y 1) mark confirmed bullish divergence. Red down-triangles at the top (pane-y 98) mark confirmed bearish divergence.

A confirmed bullish divergence is two confirmed price pivot lows, with the second pivot lower than the first, paired with two blended-MACD values at the same historical offset where the second value is higher than the first. Bearish is the mirror: two confirmed price pivot highs in higher-high order, blended-MACD values at the same offsets in lower-high order.

That is a useful description of a particular moment. It is not a directional call. The triangle is geometry, not a forecast. Divergence against price has been part of technical reading for decades, and it has been misused for about as long — which means most readers come to this tool with a pre-existing frame that needs naming out loud.

Its honest utility is narrow and specific: it is a moment to ask whether the slot stack underneath the blend agrees with the geometry or argues with it. A confirmed bullish triangle during a stack of bearish slots is a geometry note inside a continuation — one to register and mostly leave alone. A confirmed bullish triangle during a stack of slots that are already rolling over is context that agrees with evidence already in the pane. In neither case does the triangle replace the slot stack as the thing you are reading. In neither case does it supply a decision.

It is not an entry. Every page in this pack that mentions divergence carries that qualifier without softening it.

Plot On Pivot — the divergence honesty beat

Plot On Pivot is a visual setting that matters more than it looks like it should. OFF (the default): the triangle prints at the right-shoulder confirmation bar — the bar the alert fired on. ON: the triangle back-shifts visually by the Pivot Len bars, landing on the original pivot bar.

Both visuals are truthful in different ways. OFF is truthful about the alert's bar of record. ON is truthful about where the geometry formed. The problem is that ON is easy to misread: a reader glancing at a historical chart with the triangle visually anchored to the original pivot bar can think "I could have seen this here." They could not have. The alert did not fire at that bar. The marker was not visible there in real time.

If you use Plot On Pivot ON — and there are good reasons to, because seeing where a pattern formed is useful for post-hoc analysis — keep both facts in your head every time:

• The triangle shows where the pattern formed.

• The alert fires at the right-shoulder confirmation bar, not the back-shifted bar.

That pairing is non-negotiable on this pack.

Three ambiguous states you will hit

These are the shapes the pane puts in front of you that require judgment. Each one is easy to misread and each one is worth reading deliberately.

Ambiguous state 1 — blend hovering near 50

The blend is close to the midline. The histogram is oscillating between faint green and faint red. The slots may or may not be disagreeing underneath.

What this usually means: convergence-divergence is near equilibrium. The blend has no clear directional story. The structure features may be doing something — a BBWP compression is common in this posture, a Donchian channel pinching is common — but the blend itself is not telling you direction.

What this is not: a reversal setup. Mean-reversion-style trades that treat 50 as a destination will fire often and perform poorly, because 50 is equilibrium on a normalized oscillator, not a magnetic value.

How to read it: look at the slot stack. If the slots are quiet and aligned, the blend is quietly waiting for evidence. If the slots are disagreeing, the blend is averaging disagreement — and an average of disagreement is not a conviction. Let the session develop.

Ambiguous state 2 — bearish divergence triangle into a continuation

A confirmed bearish divergence triangle prints. Price continues higher. The triangle looked clean and the move that followed argues with it.

What this usually means: divergence is a geometry statement, not a turn call. Price can continue making new highs while the blend makes lower highs at matched pivot offsets. The geometry describes that relationship accurately. It does not obligate price to stop.

What this is not: evidence that the tool is broken. The tool reported exactly what it promised — a confirmed divergence geometry. The mistake was in reading it as a call.

How to read it: treat the triangle as a trigger to check the slot stack. If all the slots are confidently above their signals and the stack reads bullish, the bearish divergence is a noteworthy geometry inside a continuation — interesting, not actionable. If slots are rolling over, the triangle is a geometry that agrees with the stack. Either way, the triangle is context, not a trigger.

Ambiguous state 3 — BBWP compressed on a violent price chart

Price is printing wide-range bars. BBWP is blue (below the threshold). A reader new to the pane sees that and thinks the column is lying.

What this usually means: the blended MACD line's own Bollinger band width is compressed by its own recent history. The blend can be in a compressed width regime while price is anything but — because the blend is a weighted mean of normalized slot reads, and the noise that makes price look violent can flatten the blended MACD as slots contradict each other.

What this is not: a price-volatility reading. BBWP is not price BBWP. The qualifier lives on every mention: from the blend, not from price.

How to read it: either BBWP is telling you something about the blend's regime (the blend is quieter than its own recent history, even while price is loud) or BBWP is simply describing the blend's width in a window when the blend's width doesn't correlate with what price is doing. Either way, BBWP is an input to judgment, not a volatility proxy.

The not-independent fact, up close

The four structure features are drawn on the blended fast line. Divergence reads confirmed price pivots against the blend at the same offset. Keltner envelopes the blend around a smoothed basis. BBWP percentile-ranks the blend's own Bollinger width. Donchian marks the blend's own highs and lows.

When a divergence triangle prints, the blend is probably also near a Donchian or Keltner extreme. The BBWP has probably just moved. On the pane, that looks like four features arriving at the same conclusion at the same moment. It reads like breadth.

It is not breadth. It is one line agreeing with itself.

Divergence in particular almost guarantees this effect by construction. A confirmed bullish divergence requires a blended-MACD value that is higher than a previous blended-MACD value at matched pivot offsets — which, in practice, often means the blend is near a local extreme of its own recent range. That is a Donchian touch. The blend being at an extreme of its own range is also often near a Keltner envelope extreme. That is Keltner agreement. Three features pointing in the same direction in that moment is mechanical, not confirmatory.

Reading two structure features as two agreeing witnesses is the misread that inflates confidence most cheaply. The pack does not use confirms to link features, and neither should you. The useful reads come from noticing how the features differ — a Donchian upper touch while BBWP is compressed is a different story than a Donchian upper touch while BBWP is expanding — not from stacking them as agreement. Think of the four features as four angles on one subject, not four witnesses to one event.

What the pane is and is not

It is a bounded, ATR-normalized read of the MACD triple, blended across active slots, with four interpretive layers drawn on the blend itself. It reports state. It reports geometry. It reports regime posture of the blend.

It is not a buy/sell flag service. It does not predict direction. It does not replace price action. It does not describe volatility on price. It is not RSI and it does not have overbought or oversold zones.

If you need ten-slot context breadth and cross-asset packs more than blended-line structure, use CTX. If you want three slots and a cleaner pane without the four structure features, use Base. STR is the structure version: five slots, stronger per-slot controls, and extra reads on the blended MACD output.

Next stop: Alerts if you are building the state descriptors into a workflow, MTF and Repainting if the per-slot On Bar Close? decision is on your mind, or Workflows if you want three named scenarios with reading routines.