Settings

Written By Axiom Admin

Last updated 22 days ago

Settings

This is the knob-by-knob reference. Every input is a choice with a cost in both directions, and this page names both directions before it names the default. Defaults are a restrained starting point for a reader who wants the pane to behave sensibly on first install; they are not an endorsement and they are not optimal. There is no optimal. There is what you chose, what it cost you, and what you decided to live with. Settings do not fix a bad read — they shape the evidence the read is working with. Come to this page when you want to understand a specific input, not when you want the tool to feel different without understanding why.

Decision map — what to touch first

If you are mid-configuration and do not want to read this page linearly, start with these decisions in this order. Everything else is secondary.

Pick a source on each slot. close is the textbook baseline for RSI. Swap to hlc3 or ohlc4 only if you have a reason — long wicks on the symbol you trade, a personal preference for body-over-wick weighting — not because the alternative looks more "precise."
Pick an RSI length you can defend on this instrument. 14 is textbook. 7 and 9 react faster and hop more. 21 and 28 calm the RSI and lag more. You have only one lookback per slot; there is no fast/slow pair to balance inside the RSI itself. Every downstream smoothing pass is a second-order effect on top of this choice.
Pick a smoothing pair. Short pairs (SMA 3 / SMA 3) stay close to textbook RSI behavior while letting the color encode the RSI-vs-signal relationship. Long pairs (SMA 7 / SMA 7 or similar) smooth more aggressively and lag more.
Pick an MA family on each pass. The Lite library exposes SMA, EMA, RMA, WMA, VWMA, and SWMA. In Base, the family choice is your effective tuning lever; there are no extra per-family parameters on this trim.
Pick a timeframe at or above the chart timeframe. Slot timeframes strictly greater than or equal to the chart's timeframe. A slot timeframe below the chart timeframe raises a runtime error that names the slot.
Leave master smoothing off for the first session. Turn it on later only if you have a reason; that reason should not be "the blend looks choppy."
Leave On Bar Close? on until you have read MTF and Repainting. The tradeoff between confirmed and live is real and worth understanding before you flip the switch.

Per-slot primary inputs

Three slots share the same input shape. Configuring slot 01 teaches slot 02 and slot 03 by pattern. The inputs below appear in the settings dialog in the order TradingView exposes them — that is the order the dialog position cues, and rearranging on pedagogical grounds would make the dialog harder to scan.

Enable RSI 0N

Type: checkbox. Default: on for all three slots.
On: the slot computes, plots (unless Hide Plot is on), contributes to the blend (unless weight is zero), and evaluates its per-slot alerts.
Off: the slot goes silent in every sense. No compute, no plot, no blend contribution, no alert.
Cost: disabling a slot removes it from the alignment tally as well as the blend. If you are trying to run a two-slot configuration for a session, disabling the third is cleaner than zeroing its weight (see the note on weight-zero observer slots below).

Hide RSI 0N Plot

Type: checkbox. Default: off.
On: the slot's line disappears from the pane; everything else about the slot keeps running. The slot still feeds the blend at its current weight. The slot's alerts still fire on its state. The alignment tally still counts it.
Off: the slot plots normally.
Cost: hidden slots are easy to forget. If a blend moves for reasons the two visible slots cannot explain, the hidden slot is usually the cause. Cross-check the Limitations and Trust Boundaries page for the symmetric misread.

Source:

Type: series. Default: close.
What it feeds: ta.rsi(source, length). The source is the price series the RSI reads. close is the conventional choice because RSI was defined on closes and published comparisons assume closing values.
Alternatives: hlc3, ohlc4, hl2, open, or any of TradingView's source choices. hlc3 dampens the difference between wicks and bodies; ohlc4 includes the open; hl2 ignores the close entirely.
Cost: changing the source away from close makes the slot incomparable to any textbook RSI reading you may have in your head from another tool. On doji-heavy sessions the divergence between close and hlc3 can be meaningful. This is not a trap, but it is a commitment — if you change the source, change it with a reason and remember the change the next time the pane confuses you.

TimeFrame:

Type: timeframe string. Defaults: "5", "15", "60" for slots 01, 02, 03.
Rule: the slot's timeframe must be greater than or equal to the chart's timeframe. A 5m slot on a 1m chart is normal. A 1m slot on a 5m chart raises a runtime error.
Error shape: the runtime error names the slot explicitly — for example, RSI 01 timeframe cannot be lower than the chart timeframe. If you see that string and the pane is blank, the slot timeframe is below the chart timeframe.
Blank: leaving the field blank inherits the chart's own timeframe. That is a legitimate configuration for a slot you want anchored to the chart's bar sequence.
Cost: a slot on the same timeframe as the chart loses the multi-timeframe contribution — it still runs the RSI-and-smoothing pipeline, but the answer is drawn from the chart's own bars. See MTF and Repainting for the same-timeframe collapse case.

RSI Length:

Type: integer. Default: 14. Minimum: 1.
What it feeds: the RSI lookback passed to ta.rsi. Shorter lengths react faster, overshoot more, and are more sensitive to single-bar noise. Longer lengths calm the RSI and introduce lag.
Pair with smoothing: the RSI line and the signal line both smooth the RSI that comes out of this lookback. A long RSI length plus long smoothing lengths produces a slot that is very slow by the time it reaches the pane — sometimes slow enough to lag the chart's story enough that the slot stops carrying useful same-bar evidence.
Cost: a 14 is the textbook baseline. Moving shorter speeds the slot and noises it up; moving longer slows it and pushes the slot's read behind the chart.

RSI Smoothing:

Type: integer. Default: 3. Minimum: 1.
What it feeds: the length of the first MA pass, applied to the raw RSI. The output is the slot's RSI line.
Short values (2 to 4): stay close to the raw RSI. The color still gives you the RSI-vs-signal read without pulling the line far from textbook territory.
Long values (7 to 12): visibly calm the line. The line spends more time near the midline and responds more slowly to direction changes.
Cost: increasing this length adds lag. The cost scales with the signal length as well, because the signal line is smoothing on top of the RSI line.

RSI Type:

Type: enum. Default: SMA.
What it feeds: the MA family used for the first smoothing pass.
Families: drawn from the Lite library: SMA, EMA, RMA, WMA, VWMA, and SWMA. Most use the length input; SWMA follows TradingView's fixed ta.swma() behavior.
Cost: in Base, the family choice is the effective tuning lever on the first smoothing pass. No inside-family knobs are exposed on this trim.

Signal Length:

Type: integer. Default: 3. Minimum: 1.
What it feeds: the length of the second MA pass, applied to the RSI line. The output is the slot's signal line.
Second-order effect: the signal line is smoothing on top of the RSI line. A given change in Signal Length is felt more strongly when the RSI smoothing is short than when the RSI smoothing is already long.
Cost: long signal lengths add direct lag to the color transitions on the slot. The slot's color flips when the RSI line crosses the signal line; smoother signal lines flip later.

Signal Type:

Type: enum. Default: SMA.
What it feeds: the MA family used for the signal smoothing. Can differ from the RSI Type on the same slot. A common two-family configuration is SMA for the RSI pass and an EMA or WMA for the signal — the EMA makes the signal line more reactive and the color transitions earlier.
Cost: same family-choice commitment as the RSI Type. Family choice is the primary tuning lever; inside-family knobs are not exposed in Base.

Line Width:

Type: integer. Default: 2. Minimum: 1.
What it feeds: the drawn thickness of the slot's RSI line.
Cost: purely visual. Higher values make the slot dominate the pane; lower values make it recede. Adjust when the pane is crowded.

Blended Weight:

Type: float. Default: 33.3. Minimum: 0.
What it feeds: how much the slot steers the blended pair. The blend is a weighted average — slots with higher weights pull the blended pair harder toward their own values.
Weight 0: excludes the slot from the blend entirely while leaving the slot computing, plotting (unless Hide Plot is on), and alerting. This is the observer-slot pattern and is documented in Workflows.
Consequence you must remember: a weight-zero slot still counts toward the alignment alerts. The alignment tally is driven by enable, not by weight. A reader who zeros a slot's weight to "turn it off" has not turned it off — alignment and per-slot alerts still fire.
Cost: equalizing weights creates the appearance of three independent opinions while the underlying measurement is autocorrelated by shared source and timeframe laddering. Differentiating weights does not remove that autocorrelation; it just re-apportions it. A slot at 80 is not a stronger claim than a slot at 40 — it is a slot you told the blend to listen to twice as hard. If the weight is ever doing more reasoning than your configuration-time intention, the blend is telling you a story about your weights, not about the market.

Oscillator group

Overbought Level

Type: float. Default: 70. Step: 0.1.
What it feeds: the upper dashed reference line.
Internal logic: none. No alert fires on this line. The 70 value is a reading convention drawn from textbook RSI, and the line is here so you can keep that convention visible on a pane that is not plain RSI. You can move the line — it is a visual guide, not a threshold the tool enforces.

Oversold Level

Type: float. Default: 30. Step: 0.1.
What it feeds: the lower dashed reference line.
Internal logic: none. Same design as the overbought line.
Cost of mis-reading: the 30/70 pair is a reference frame, not a trigger frame. Trading off the lines with the pane as the evidence is a habit transferred from textbook RSI that Base is not structured to support. Use the reference for orientation; do not expect the tool to confirm it.

Display group

Plot Blended RSI/Signal

Type: checkbox. Default: on.
On: the blended pair, the blended signal, and the tinted fill between them all draw on the pane.
Off: the three blend visuals are hidden. The blend still computes. Blend-based alerts still evaluate. If you want to hide the blend for visual reasons but keep alerts running, this is the toggle — and the separation between display and logic is deliberate. A reader who hides the blend and then wonders why a blend alert still fires is encountering the intended separation, not a bug.

Blended Line Width:

Type: integer. Default: 3. Minimum: 1.
What it feeds: the thickness of both the blended RSI line and the blended signal line.
Cost: visual only. Higher values make the blend dominate the pane over the slot evidence; lower values recede the blend to the background.

Master Smoothing group

Enable Master Smoothing

Type: checkbox. Default: off.
On: one additional MA pass runs on the blended pair before the pair is plotted. The per-slot lines and per-slot alerts are untouched — this is a blend-only layer.
Off: the blend is plotted as it came out of the weighted average.
What it trades: responsiveness for calm. A calmer-looking blend at a real lag cost. Long master lengths on top of long per-slot signal smoothing can flatten the blended pair right through a regime shift that is still legible on the slots underneath. If you enable master smoothing, keep the slot lines visible so the unsmoothed evidence is still on-screen; that is the configuration the pack expects. Master smoothing is not a remedy for a choppy blend — a choppy blend is evidence that the slots are disagreeing.

Master MA Type

Type: enum. Default: EMA.
What it feeds: the MA family used for the master pass. Note the default shift — per-slot smoothing defaults to SMA, but master smoothing defaults to EMA. The master pass is meant to be responsive relative to itself; the SMA defaults on the slots are meant to stay close to textbook behavior. If you change Master MA Type, you are selecting a different lag profile for the final smoothing pass.

Master Length

Type: integer. Default: 3. Minimum: 1.
What it feeds: the length of the master pass. Higher values calm the blend more and add more lag.
Cost: the cost compounds with the per-slot signal-smoothing lengths. You are smoothing the average of already-smoothed lines; the effective lag is the combination of both.

Global PU Settings

On Bar Close?

Type: checkbox. Default: on.
On: each slot's higher-timeframe call returns the values from the previous confirmed higher-timeframe bar. The slot stops moving within a live higher-timeframe bar. Stable read, one slot-timeframe bar of latency relative to the live value.
Off: each slot returns the live higher-timeframe bar's values. The slot moves while its higher-timeframe bar is still open and settles when that bar closes. Earlier read, revisable until the close. On historical bars, treat this mode as repaint-exposed hindsight, not proof of what you would have seen live.
Important property of this trim: there is one switch, and it applies to every enabled slot at once. Per-slot repaint control lives on CTX and STR.
This is a tradeoff, not a fix. Both modes report something truthful about the market; what you choose is what you want to be looking at. MTF and Repainting walks the full detail with a verification exercise.

Lite-library MA families — what this trim does and does not tune

The slot smoothing passes and the master pass all consume the Lite moving-average library. The table below is honest about which knobs appear on Base and which belong to the broader CTX and STR surfaces.

Family	Typical behavior on a smoothing pass	When a reader might prefer it	What Base does not let you tune inside the family
SMA	Uniform weight across the window. Calm but slow to turn.	You want the baseline that matches most textbook RSI references.	No per-family knobs on Base.
EMA	Recency-weighted. Turns faster than SMA of the same length.	You want the slot's signal line to flip earlier on direction changes.	No separate alpha control.
RMA	Wilder-style smoothing. Slow, steady, and familiar from RSI-adjacent tools.	You want a smoother response with a Wilder feel.	No per-family knobs on Base.
WMA	Linear weights with recency bias.	You want recency bias without EMA's exponential tail.	No per-family knobs on Base.
VWMA	Volume-weighted smoothing.	You want higher-volume bars to carry more weight in the smoothing pass.	No volume source override on Base.
SWMA	TradingView's symmetrically weighted moving average.	You want the fixed SWMA shape TradingView provides.	The length input does not tune SWMA.

Family choice in Base is the primary tuning lever. The inspected Lite library does not expose offsets, phases, efficiency bands, fractal windows, or other inner-family controls through this tool. Saying so plainly is a scope statement. If this palette covers the way you actually smooth RSI, the absence costs you nothing. If you need a deeper MA surface, Base is not the trim that provides it.

Tripwires before you change anything

Master smoothing. Off by default on purpose. Enabling it adds lag to blend-based alerts and does not change per-slot alerts. It is a visual calm setting, not a correctness setting.
On Bar Close?. On by default. Flipping off without reading MTF and Repainting means you will see a slot move intra-bar on a higher timeframe and may misread that motion as real-time sensitivity.
RSI length and smoothing pair. The second smoothing pass compounds with the first. A length 14 RSI with an SMA 3 RSI smoothing and an SMA 3 signal is not the same instrument as a length 14 RSI with an SMA 7 RSI smoothing and an SMA 7 signal — the second configuration is substantially slower and can sit behind the chart's story.
Weight zero vs disabled. Zero weight removes a slot from the blend. It does not remove the slot from alignment or from per-slot alerts. If you want full deactivation, disable the slot.
Hide Plot vs disabled. Hide Plot hides the line. The slot still computes, still feeds the blend at its weight, and still alerts. The visible pane is not the whole pipeline.

Where to go next

How the configured pane is supposed to read — Visuals and Logic.
The On Bar Close? switch and the slot-timeframe rule in full — MTF and Repainting.
What the ten alerts do and do not confirm — Alerts.
When a chosen configuration looks wrong — Troubleshooting.