Settings

Every control on this pane is a decision with a cost. This page walks them as decisions, not as a glossary. If you want the list of inputs, the input panel already has that. If you want to know which way to turn a knob and what you should expect to give up when you do, read on.

Read this first — the three default tripwires

Before you touch any individual slot input, know what these three controls do, because misreading them is where most first-session confusion comes from.

ATR Sensitivity (default 1.0)

This is the dial that scales each slot's normalized MACD, signal, and histogram before the sigmoid lands them in 0 to 100. 1.0 is the default baseline. Higher pushes readings toward 0 and 100 faster; lower flattens them toward the midline at 50.

The common misuse is cranking this higher when the pane looks "flat" in a quiet session. A quiet session is evidence — the convergence story is not pushing hard in either direction, and the pane is honestly reporting that. Pushing sensitivity to 3.0 to "see something" does not add information; it pushes more of the reading into the sigmoid's flat region at the boundaries, where movement of different sizes maps to roughly the same number. The pane does not become more sensitive. It becomes more saturated.

Move this dial when you have a reason you can say out loud ("on this instrument, 1.0 leaves the readings under 60 almost all session, so I want to pull in the useful range"). Do not move it because the pane looks calm.

Enable Master Smoothing (default off)

Off by default on purpose. When it is off, the blend uses the raw weighted blend from the active slots. When it is on, a single MA pass is applied to the blended K, D, and H before plotting, and blend-based alerts fire on that smoothed blend instead.

That is a noise-for-lag trade, and the lag is real. A smoothed blend can change the timing of blend alerts, usually by delaying turns and sometimes by muting short-lived state changes entirely. The chop that master smoothing hides was evidence about the slot evidence; hiding it does not repair it, and treating a smoothed-blend alert as "clean confirmation" overstates what the code is doing.

If you enable this, do it because you want a slower regime read, not because you are trying to fix a noisy alert experience. The fix for noisy slot evidence is a longer length pair on the slot or a different MA family choice, not a smoother wrapper.

On Bar Close? (global, default on)

This is the repaint switch, and in Base it is one switch for the whole pane. Every slot obeys it. On returns the confirmed previous-bar value from each slot's configured timeframe — stable but one slot-timeframe bar behind. Off returns the live intra-bar value — faster but capable of changing its mind as that slot's current bar develops.

Neither mode is "better." Each is honest about what it reports as long as you know which one you chose. The detail conversation lives in MTF and Repainting. For now, the working rule is: leave this on until you have read that page and decided, for a specific workflow, that the earlier-read tradeoff is worth the repaint cost.

The CTX and STR versions expose this switch per slot. Base does not. If you need some slots to repaint and others to wait, that is a version-fit question, not a Base setting hidden somewhere else.

Per-slot primary inputs (archetype: Slot 01)

Slots 02 and 03 expose the same inputs with the same semantics. Names in quotes are the exact labels in the input panel.

Enable MACD 0N (default true for all three slots)

Off turns the slot all the way off — no compute, no plot, no alerts, no contribution to the blend, no contribution to the alignment alerts. If what you want is "no influence on anything," this is the correct control, not a weight of zero.

Hide MACD 0N Plot (default false)

Removes the slot line from the pane without disabling the slot. The slot still computes, still contributes to the blend (if its weight is non-zero), and still fires its per-slot alerts.

This is useful when the slot's job is to feed the blend or provide alert coverage and its line would be clutter. The honest caution: hidden slots are easy to forget. A week later, when the blend moves in a way the two visible slots cannot explain, the answer is usually the hidden slot. If you adopt this pattern, keep a note somewhere — a chart annotation, a template name — that tells future-you a slot is hidden and why.

Source (default close)

The price series the slot reads. Closing price is the textbook MACD baseline, and it is what most traders have an intuitive feel for. hlc3, ohlc4, hl2, and their typed variants roll in more of the bar's range and react differently to long-wicked candles — a doji-heavy session can produce a meaningfully different slot line on hlc3 than on close, because the midpoint absorbs the wicks the close ignored.

The thing to resist is changing source across all three slots at once and then wondering why the pane behaves differently from what you remember. Change one slot's source, watch it for a full session, and only then decide whether you want the other slots to match. Mixing sources across slots is a legitimate configuration — a close-driven fast slot with an hlc3-driven slow slot is a defensible way to ask a different question of each cadence — but it is not a configuration you want to build accidentally.

TimeFrame (defaults 5 / 15 / 60)

The higher timeframe this slot runs on. Must be greater than or equal to the chart timeframe. If it is not, the script raises a runtime error that names the offending slot — a hard stop, not a silent downshift.

The default pattern is an escalating ladder: 5m, 15m, 60m on a 1-minute chart. That is the shape the pane was designed around. Going to a 1-hour chart with those defaults is the chart-timeframe collapse case in MTF and Repainting; the tool is behaving correctly, but the configuration has lost its multi-timeframe purpose.

Fast Length and Slow Length (defaults 12 and 26)

The two lengths inside this slot's MACD calculation. Fast Length must be strictly less than Slow Length; a violation raises a runtime error naming the slot.

12/26 is the textbook baseline. A longer pair (24/52, 48/104) gives you a calmer slot that reacts later to turns. A shorter pair gives you a twitchier slot that reacts sooner and flips more. The pane reflects the lag profile directly — faster pairs push the slot's color around more, which shows up in the four-state histogram as more accelerating/decelerating transitions.

A trap to name explicitly: keeping 12/26 on all three slots and only varying the timeframe is the most common way readers end up with three slots that correlate heavily. The pane works hardest for you when the three slots are three different slots, not three copies of the same slot on different cadences.

Signal Length (default 9)

The smoothing length that turns the slot's MACD into its signal line. The textbook 9 pairs with the textbook 12/26. If you move the length pair, think about whether your signal length should move with it. There is no mechanical rule; there is a coherence rule — the signal should smooth the MACD at a cadence that matches how fast the MACD is moving.

MACD MA Type and Signal MA Type (defaults EMA on both)

The moving-average family used for the fast and slow MAs inside the MACD, and for the signal line. Both are drawn from the Lite library in this trim. That matters, and it matters in two directions:

• The Lite family set is small by design: SMA, EMA, RMA, WMA, VWMA, and SWMA. HMA, DEMA, TEMA, ALMA, Jurik, KAMA, FRAMA, Laguerre, and VAMA are not exposed by this local Lite library.

• Because the Base trim does not expose Power User parameter blocks per slot, the family choice is your tuning surface for family character in this tier. One important exception: SWMA routes to TradingView's fixed symmetrically weighted moving average, so the length input does not tune SWMA the way it tunes SMA, EMA, RMA, WMA, or VWMA.

The honest framing: Lite is not a worse version of Pro. It is a smaller decision space. If you expected a family that is not listed here, the Base panel is not hiding it behind another setting; it is not in this library.

Line Width (default 2)

Visual only. No effect on computation or alerts.

Blended Weight (default 33.3 on each slot)

How much this slot steers the blend. Equal weights mean each slot contributes equally. Setting a slot's weight to 0 excludes it from the blend — but does not turn it off. The slot still computes, still plots (unless Hide Plot is also on), and still fires its per-slot alerts. It also still counts toward the alignment alerts in Alerts.

That last point is the one to hold onto. "Weight zero" is not "turned off." It is the observer pattern: a slot that participates in every part of the pane except the blend. If you want a slot to stay out of both the blend and the alignment count, disable the slot. If you want a slot to steer the blend less without leaving the alignment conversation, reduce its weight.

Oscillator group (applies to every slot)

ATR Length (default 14)

How far back each slot's ATR looks when normalizing. Longer gives a steadier volatility yardstick that is slower to react to recent volatility changes. Shorter reacts faster but can make the slot's behavior jumpier during volatility transitions. ATR is computed inside each slot's own request.security context, so a 60-minute slot's ATR is 60-minute ATR on the chart instrument — not a single shared ATR across the three slots.

ATR Sensitivity (default 1.0)

Covered in the tripwires section above. Re-anchor here: move it when you can name the reason. Never treat pinning as intensity; pinning is evidence that the pairing of sensitivity and current ATR has pushed the sigmoid into its flat region.

Overbought Level (default 70) and Oversold Level (default 30)

User-configurable reference lines. They drive the dashed-gray guides on the pane. They do not fire any alerts and they are not triggers. They are zones you can use for your own reference or for your own TradingView threshold alerts wired against the plotted series.

If 70 and 30 feel like "sell here" and "buy here" to you, that is RSI muscle memory talking. The pane is a normalized convergence/divergence read, not a mean-reversion gauge. Limitations and Trust Boundaries names this misread explicitly.

Display group

Plot Blended K/D (default true)

Draws the blended fast and blended slow lines and the tinted fill between them. Turning this off hides those lines but does not disable the blend calculation or the blend-based alerts. Alerts wired to the blend still fire when their condition is met, regardless of whether the blend is drawn.

Plot Blended Histogram (default true)

Draws the four-state histogram columns rooted at 50. Same independence — turning this off hides the columns but the histogram is still computed internally, and nothing downstream stops.

Think twice about hiding the histogram. The four-state color is the pane's speedometer; it is the cheapest way to read deceleration before it shows up on the slot lines. Hiding it is a deliberate information drop, not a cosmetic choice, and it costs you the best early surface the pane has.

Blended Line Width (default 3)

Visual only. Makes the blended fast and slow thicker relative to the slot lines, so the reading order (slots as evidence, blend as headline) is visually obvious.

Master Smoothing group

Enable Master Smoothing (default false)

Covered in the tripwires. Off keeps the blend faithful; on adds lag for visual calm.

Master MA Type (default EMA, from the Lite library)

Which family runs the single post-blend smoothing pass. EMA is the mild, symmetric choice. Other Lite families shift the lag profile. If you enable master smoothing, start with EMA, and only move to another family once you have a specific reason.

Master Length (default 3)

Length of the smoothing pass. Higher values smooth more and generally push turns later, but the exact alert timing depends on the shape of the blend, not on a fixed bar-count formula. A length of 3 to 5 gives you a visibly calmer blend that still reacts on a reasonable time budget. A length of 20 hides much of what the slots are doing; at that point you are no longer reading the slot evidence through the blend, you are reading a moving average of the blend, and the slot-level picture is where your eye has to live instead.

A concrete verification you can run: note the chart time when a blend-bull alert fires with smoothing off. Turn smoothing on with length 5, reload the chart at the same anchor, and compare the fire timing. Often it lands later; sometimes a brief state does not survive smoothing at all. Either way, you are seeing the timing cost of the smoother blend.

PU Settings group

On Bar Close? (default true)

Global repaint switch. Covered in the tripwires and in detail on MTF and Repainting.

The decision map — what to touch in what order

If you are configuring from defaults, this order will save you from configuring backwards.

1. Pick your chart timeframe first. The slot timeframes must live at or above it, not below it. Deciding the chart is a different decision from deciding the slots.

2. Decide what story you want each slot to tell. Fast intra-day, medium, higher-timeframe baseline — pick three genuinely different stories before you pick lengths. A trio that is three copies of the same slot on different timeframes is a trio the alignment alerts will fire on continuously during any trending session.

3. Pick a length pair per slot that matches the story. Short pair on the fastest slot if you want it twitchy, longer pair on the slowest if you want it behaved.

4. Pick an MA family per slot — MACD side and signal side. EMA is the default baseline. Reach for a different Lite family only if you already know what that family does and why you want it in this slot.

5. Pick a source per slot. close is the safe baseline. Other sources are legitimate but change the evidence in ways that are not always obvious.

6. Leave sensitivity at 1.0 and ATR length at 14 until you have evidence to move them. "Evidence" means a session or two of watching the pane behave, not a first-five-minute reaction to the defaults.

7. Leave master smoothing off until you have a specific reason to turn it on — and when you do, turn it on with a short length and watch whether it is helping or hiding.

8. Decide On Bar Close? deliberately. On for the stable, confirmed read. Off only after reading MTF and Repainting.

A short table: choosing a Lite-library MA family

Reach for this when you are tuning family per slot. Columns are honest about what the local Lite library exposes.

If a family you expect is not in this table, it is because Base is keyed to the Lite library. Do not hunt for HMA, DEMA, TEMA, ALMA, Jurik, KAMA, FRAMA, Laguerre, or VAMA in this panel; they are not exposed by the imported Lite library used by this script.

Where to go next

• What every element on the pane is telling you, and how to read them in the correct order — Visuals and Logic.

• The repaint switch, the slot-timeframe rule, and the chart-timeframe collapse case — MTF and Repainting.

• The ten alert conditions and the transitions, thresholds, and divergence events this tool deliberately does not alert on — Alerts.

• The anti-patterns that come from configuring three slots the wrong way — Workflows.

• Why pushing sensitivity to "fix" a flat pane is the opposite of the correct move — Limitations and Trust Boundaries.