Settings

This is the longest page in the pack for a reason. Every knob on STR has a cost alongside its benefit, and the settings page is where those costs get named. Read the decision map at the top if you are mid-configuration and need to touch the knob that actually matters. Read the rest when you have time.

If a section feels dense, it is dense on purpose. The shortest version of a knob's behavior is usually the version that gets a reader into trouble the first time the knob has to hold weight in a real session. Each group below names the benefit, the cost, and the condition under which the cost actually lands.

What this page is and is not

This page walks each input group as a deliberate decision. It does not prescribe best settings. It does not name a best MA family. It does not claim a specific sensitivity value is universal. When a knob has a cost in both directions, both directions are named. When a knob is only worth touching after you can state what you expect it to change, the page says that out loud — and saying it out loud is the whole point of the page.

For interpretation of what the pane is drawing, go to Visuals and Logic. For repaint posture and per-slot timing, go to MTF and Repainting. For the alerts, go to Alerts. For the construction of the four structure features at mental-model altitude, go to For the Geeks.

Decision map — what to touch in what order

1. Pick a source per active slot. Default is close. Change this only if you know why (e.g., hl2 for a less whippy read on a noisy instrument).

2. Pick a length pair you can defend on that slot's timeframe. The defaults are classical: 12/26/9. If you change them, state what you expect to change and check that it did.

3. Pick a MACD MA family and a signal MA family. EMA defaults are classical. If you choose a Pro family (ALMA, KAMA/FRAMA, Jurik, Laguerre, VAMA), do not open the Power User block unless you can describe what effect you expect from a specific parameter.

4. Pick the slot's timeframe. It must be at or above chart TF. If it equals chart TF the slot collapses onto the chart bar and the MTF benefit is gone for that slot.

5. Decide On Bar Close? per slot. Deliberately. This is not a global decision anymore. One side lags, the other can repaint. Pick with eyes open.

6. Leave ATR Sensitivity at 1.0 until a specific session gives you evidence to move it. Higher pins the pane at extremes faster. Lower flattens the read toward the midline. Neither is "more signal."

7. Decide which structure features belong on the pane for your reading style. Running all four at once is legal but can crowd a small chart. They share a source; loading more of them does not add reads.

8. Only touch master smoothing if you have a specific reason. It adds lag to blend-based alerts. Per-slot alerts are unaffected.

9. Leave Pivot Len at 20 unless you know what the tradeoff costs. Lower yields more triangles and more wrong ones. Higher yields fewer and later.

10. Leave Plot On Pivot OFF until you understand the honesty beat. Then if you turn it on, keep the honesty beat in your head.

Default tripwires to know about before the knob walk

A handful of defaults deserve explicit naming because readers who do not know they are there will misread the pane:

• Slot 01 through 03 ship Hide Plot ON. The slots are computing and voting in the blend. Their lines are simply not drawn. If you are confused about why the blend moved and no visible slot line moved, the answer is almost always a hidden slot pulling.

• Slot 04 and 05 ship with Enable OFF and Weight 0. If you enable a slot without setting its weight, it will plot and alert but contribute nothing to the blend. That is a legitimate observer configuration, not a bug.

• Keltner ships OFF. Donchian, BBWP, and divergence ship ON. The pane is deliberately busy enough without the envelope.

• BBWP Lookback defaults to 252. BBWP returns na until it has 252 prior valid width readings, after the Bollinger width itself has warmed up. The columns are not broken.

• Pivot Len is 20 on both left and right. Divergence confirmation arrives 20 bars after the actual pivot. That delay is not a bug; it is the cost of strict pivots.

• On Bar Close? defaults to ON on every slot. Enabled slots run lag-honest by default, one bar behind their own context. Turn any slot OFF deliberately, not casually.

Per-slot inputs (MACD 01 through MACD 05)

Five slots, identical shape. Defaults differ: slots 1–3 enabled at 5m, 15m, 60m respectively; slots 4 and 5 disabled and zero-weighted. Every setting below applies to all five.

Enable MACD 0N

Boolean. Defaults: slots 1–3 true, slots 4–5 false. Disabled slots do not compute, do not plot, do not alert, and do not vote in the blend. This is the full-off switch. Use it when you are not going to use that slot at all this session.

Hide MACD 0N Plot

Boolean. Defaults: slots 1–3 true, slots 4–5 false. Hide Plot removes the slot's line from the pane. It does not stop the slot from computing, voting in the blend, or firing its per-slot alert. Every time you hide a slot, remember that it is still pulling on the blend. If you intend to take a slot out of the blend, set its weight to zero instead (and know that zero-weight slots still alert).

Source

Price series. Default close. Fed into the MACD calculation on the slot's timeframe and, if the Optional Ticker is set, on that ticker. Changing source changes what the MACD is measuring, not how it is measured. hl2, hlc3, and ohlc4 are common alternatives when close feels too reactive to final-tick noise.

TimeFrame

HTF string. Defaults 5, 15, 60, empty, empty. Must be at or above chart TF. A slot TF below chart TF raises a runtime error — the slot label will name itself in the error message so you know which one. An empty string means "run on the chart TF." In that configuration, On Bar Close? ON gives you a one-bar-lagged read; OFF gives you a live intra-bar read that updates with ticks. Both are legitimate.

Fast Length / Slow Length

Int ≥ 1. Defaults 12 and 26. Fast must be strictly less than Slow — violating this raises a runtime error naming the slot. Wider pairs (12/52, 24/52) smooth the MACD and reduce whip at the cost of later crossings. Tighter pairs (8/17) tighten response at the cost of more crossings per session.

MACD MA Type / Signal MA Type

Enum from the Pro MA library. Defaults EMA. Choices include SMA, EMA, ALMA, KAMA, FRAMA, Jurik, Laguerre, VAMA, and the rest of the Pro library's roster. Non-EMA families expose Power User parameter blocks lower in the inputs dialog; those blocks only affect the read when the corresponding family is selected. Cross-reference the MA library manual for family behavior rather than the paragraph below, which is deliberately high-level.

Rough guide, not a recommendation:

• EMA is the classical MACD ingredient. Use it when you want the read to behave like the MACD you learned.

• SMA treats every bar in the window equally. Slower response, steadier line.

• ALMA is weighted by a gaussian; sigma controls the shape. Useful when you want the response curve tuned.

• KAMA / FRAMA adapt their effective window to recent volatility. Fast- and slow-ends of the adaptive range are what the Power User block exposes.

• Jurik is a low-lag filter with phase and power controls. The knobs change the filter's character; touch them only if you have a specific question.

• Laguerre has an alpha that trades lag for smoothness.

• VAMA adapts to a volatility window.

Pick a family you can describe. If you can't describe what you want differently from EMA, stay on EMA.

Signal Length

Int ≥ 1. Default 9. Used as the signal MA window applied to the MACD line inside the slot. Longer means slower crossings; shorter means earlier (and noisier) crossings.

Line Width

Int ≥ 1. Default 2. Pixel width of the slot's plotted fast line.

Blended Weight

Float ≥ 0. Defaults 33.3 for slots 1–3; 0.0 for slots 4–5. The weight a slot carries inside the weighted-mean blend. Zero means the slot plots (unless Hide Plot is on) and alerts (per-slot), but does not steer the blend.

Optional Ticker

Symbol. Default empty. When set, the slot runs its source and ATR on that symbol instead of the chart symbol. Cross-ticker slots are powerful and easy to misuse: pairing a large-cap benchmark with a tiny-cap name gives you a hybrid blend across two different liquidity regimes. Legitimate, sometimes — but be aware that the blend no longer reports a single symbol.

On Bar Close?

Boolean. Default true for every slot. Per slot, not global. STR differs from Base here, and the difference is more expressive and more error-prone than Base's single switch. Full repaint posture lives on MTF and Repainting. The two-line summary:

• ON returns the previous slot-bar value. Safer, one bar lagged.

• OFF returns the live slot-bar value. Faster, can repaint until the slot bar closes.

Flipping OFF on an HTF slot without thinking about how that interacts with your other slots will produce a blend whose repaint posture is mixed across its members. That is not automatically wrong, but you have to own it deliberately.

Power User blocks (K and D)

Each slot carries two Power User blocks — one for the MACD line's MA family (K) and one for the signal line's MA family (D). Fields include ALMA Floor, Offset, Sigma; KAMA/FRAMA Fast and Slow; Jurik Phase and Power; Laguerre Alpha; VAMA Vol Length. A field only affects the read when the corresponding MA family is selected on that line.

Touch Power User knobs only when you can state what you expect to change. The defaults were chosen to produce a reasonable read for each family. Moving a Jurik Phase you cannot describe is not "pro configuration." It is gardening blind — the kind of change that makes the pane behave differently, fails to teach you anything about why, and leaves you unable to say whether the new behavior is better or just unfamiliar.

The honest test before touching a Power User knob: can you finish the sentence "I expect this change to make the line [faster / smoother / less reactive to volatility spikes / more responsive in adaptive regimes / ...] because [...]"? If yes, change it and verify. If no, close the block. The surface is exposed because some readers earn their way into it. It is not a feature-density bragging surface, and the defaults are not beginner settings that serious traders are supposed to leave behind.

Cross-reference the Axiom MA library manual for what each family's parameters actually do. That manual, not this paragraph, is the authority on family internals.

Oscillator group

Global dials for the ATR-sigmoid normalization. They apply to every slot uniformly.

ATR Length

Int ≥ 1. Default 14. Window for the per-slot ATR used inside the normalization. Higher (20, 40) makes the normalization steadier and less reactive to individual-bar volatility pulses. Lower (7) makes it touchier — a volatility expansion on one or two bars will change the normalized read sharply.

ATR Sensitivity

Float. Default 1.0. Multiplier applied inside the sigmoid. Higher sensitivity (1.5, 2.0) pushes normalized values toward the 0/100 rails faster — the pane will saturate more often and the extremes become less discriminating. Lower sensitivity (0.5) flattens everything toward the 50 midline and the pane loses range.

Sensitivity is not a "find more signal" knob. Pinning is not intensity. If your slots sit pinned at 100 for long stretches, the fix is usually to reduce Sensitivity or increase ATR Length — not to act on the pinning, and not to raise Sensitivity hoping the pane becomes "more decisive." A saturated pane is the sigmoid doing its job under conditions where the raw scaled value has grown large relative to the sigmoid's linear region. Adjusting Sensitivity moves the line between "in the linear region" and "in the flat region." It does not put more information on the pane.

A diagnostic before touching Sensitivity: look at the slot lines during a range-bound hour. If they sit clustered within ten points of 50 and barely move, Sensitivity is probably too low for the current volatility regime. If they frequently pin at 100 or 0 during normal price activity, Sensitivity is probably too high. Adjust one step at a time and watch the same chart replay before committing.

Overbought Level / Oversold Level

Float 0–100, step 0.1. Defaults 70 and 30. Reference lines drawn on the pane. Nothing in the indicator responds to crossing them — they are visual bumpers, not OB/OS thresholds. The pane is not RSI; treating it as RSI is a misread the pack names directly on Limitations and Trust Boundaries.

Blend Core group

Plot Blend

Boolean. Default true. Toggles the blended fast line, the blended slow line, and the tinted fill between them.

Plot Histogram

Boolean. Default true. Toggles the four-state column histogram on the blend.

Line Width (Blend Core)

Int ≥ 1. Default 3. Thicker than a slot line by design — the blend is the headline.

Enable Smoothing

Boolean. Default false. When ON, the blended K, D, H triple is smoothed by a master MA after blending. Smoothing adds lag. Blend-based alerts fire later, not earlier. Per-slot alerts are unaffected. If you turned smoothing on because the blend looked choppy, remember that the chop was evidence — smoothing hid it, did not solve it.

Smooth Type / Smooth Len

Enum / Int ≥ 1. Defaults EMA / 3. The master-smoothing MA family and length.

Blend Smooth Power User block

Nine fields mirroring the per-slot Power User block, applied to the master-smoothing MA family. Same guidance: only touch what you can describe.

Blend Div (divergence) group

The divergence module is the most magnetic visual on the pane and the most dangerous to misread. Configure it deliberately.

Show Div

Boolean. Default true.

Pivot Len

Int ≥ 1. Default 20. Used on both left and right of ta.pivothigh and ta.pivotlow. This is a single control with two effects:

• It sets the strictness of the pivot (more bars of left-and-right means fewer, cleaner pivots).

• It sets the confirmation delay. Pivot Len at 20 means divergence confirmation arrives 20 bars after the actual pivot bar.

Lower Pivot Len (5, 10) gives you more divergences and more wrong ones. Higher (30, 40) gives you fewer and later. There is no best value. The pack does not name one.

Plot On Pivot?

Boolean. Default false. OFF plots the triangle at the right-shoulder confirmation bar — the bar the alert fired on. ON back-shifts the triangle visually by Pivot Len bars, so it lands on the original pivot bar. The alert still fires only on the confirmation bar.

ON is honest about where the geometry formed and misleading-looking about when it was visible. The triangle is not actually lying — it is a truthful picture of where the pattern finished forming. But a reader scrolling through a historical chart sees a triangle anchored to a bar on which the triangle was not visible in real time, and the reader's gut turns that into "I could have seen this here." The reader could not have. The right-shoulder pivot had not closed yet. The alert had not fired yet. The marker was not there.

Pair that understanding with every use. A good habit: say the two facts out loud the first few times you use ON — "the pattern formed here" and "the alert fired Pivot Len bars later." When the two facts live separately in your head, ON is a useful post-hoc visualization. When they collapse into one, ON is a lie about the past.

Blend KC (Keltner) group

Show Keltner

Boolean. Default false.

KC Length

Int ≥ 1. Default 20. Lookback for the range smoothing.

KC Mult

Float ≥ 0, step 0.1. Default 2.0. Multiplier applied to the smoothed range before adding and subtracting from the basis.

Basis Type / Basis Len

Enum / Int ≥ 1. Defaults EMA / 20. The MA family and length that smooths the blend to produce the envelope's midpoint.

Show Basis / Show Fill / Line Width

Visual-only controls. Show Basis draws the envelope midpoint. Show Fill paints the area between upper and lower with the source's fixed subtle transparency.

Blend KC Power User block

Nine fields applied to the Keltner basis MA family. Same guidance as per-slot Power User.

Keltner is a stretch envelope around the blend. A touch means the blend is stretched against its own recent range. It does not mean overbought or oversold. That reframe is the single qualifier every Keltner mention carries on this pack.

Blend BBWP group

Show BBWP

Boolean. Default true.

Length

Int ≥ 1. Default 20. Applied to both the Bollinger basis window and the stdev window. The match is deliberate — classical Bollinger construction.

Lookback

Int ≥ 1. Default 252. The percentile-rank window. BBWP returns na until the rank window is full with prior valid width readings, which takes at least 252 bars at default and can take longer because the Bollinger width has its own warm-up. If the columns vanish across the earliest visible history on a new symbol, that is usually why.

Threshold

Float 0–100, step 0.1. Default 50.0. Column color flips above and below this threshold.

Basis Type

Enum. Default SMA. Classical Bollinger uses SMA. Changing the family changes the basis, which changes the width, which changes the percentile. Stay on SMA unless you know what you want the change to do.

Blend BBWP Power User block

Same nine fields as elsewhere, applied to the BBWP basis family.

BBWP reports the blend's own band-width percentile, not price's. A low column on a visibly volatile chart means the blend's own width is compressed by its own history. The qualifier "from the blend, not from price" carries every mention.

Blend DC (Donchian) group

Show DC

Boolean. Default true.

DC Len

Int ≥ 1. Default 20. Lookback for ta.highest(blend, length) and ta.lowest(blend, length).

Basis Type / Basis Len

Enum / Int ≥ 1. Defaults SMA / 1. The midpoint smoother. Basis Len 1 is effectively no smoothing. Raising it smooths the visual midpoint without changing the steplines.

Show Basis / Show Fill / Line Width

Visual-only controls. Show Fill paints the channel with the source's fixed subtle transparency.

Blend DC Power User block

Same nine fields, applied to the Donchian basis family when Basis Len > 1.

A Donchian touch means the blend is pushing its own recent range edge. It does not mean price is breaking out.

Cross-reference

• Per-slot On Bar Close? deep dive → MTF and Repainting.

• The reading order and ambiguous-state walkthroughs → Visuals and Logic.

• What the sixteen alerts actually commit to → Alerts.

• The estimate boundary and the most common misreads → Limitations and Trust Boundaries.

• The ATR-sigmoid normalization and the four structure features at mental-model altitude → For the Geeks.

• Pro MA family internals → cross-reference the Axiom MA library manual.