Workflows

This page teaches validated use patterns for the oscillator and names the anti-patterns that commonly derail new users. Each workflow includes a concrete scenario with specific settings so you can adapt it to your own...

Written By Axiom Admin

Last updated About 1 month ago

Workflows

The oscillator measures how far price sits from its moving averages across multiple timeframes, normalized by volatility and bounded to a consistent scale. That measurement serves different analytical purposes depending on how you configure the slots, what you compare, and what you do with the information. The workflows below show five distinct ways to use the tool, followed by five patterns that produce misleading or useless results.

Validated workflows

Workflow 1: Intraday MA-distance layering

Purpose: See how stretched price is from its averages at three time scales simultaneously, in one pane, to gauge whether a move has multi-timeframe support or is only visible on one scale.

Setup:

Chart timeframe: 1m or 5m
Slot 01: 5m, EMA 20 (short-term stretch)
Slot 02: 15m, EMA 20 (intermediate stretch)
Slot 03: 60m, EMA 20 (session-level stretch)
Sensitivity: 1.0
Weights: 33.3 / 33.3 / 33.3 (equal)
On Bar Close: On

What to observe:

When all three Fast lines are positive and rising, the weight of evidence says price is above its averages at every scale and the distance is growing. This is not a signal to buy — it is context that tells you the move is broad-based. The distance from the MA is increasing at the short-term, intermediate, and session level simultaneously.

When the 5m slot starts compressing toward zero while the 60m slot holds at +60 or higher, short-term stretch is fading but the session-level picture has not changed. This often happens during pullbacks within a still-intact trend. The short-term MA is catching up to price while the longer-term MA is still far below.

When the 60m slot starts dropping from a high reading while the 5m has already turned negative, the longer-term structure is beginning to weaken. This does not predict a reversal, but it tells you that the distance from the session-level average is finally compressing after the short-term already compressed.

What the blended line adds here: With equal weights, the blend gives you a quick summary of the average stretch across all three timeframes. If the blend is strongly positive while one slot is negative, the disagreement is visible: the blend is being carried by the other two slots. Check the individual lines when the blend and any single slot diverge.

When this pattern stops being useful: In range-bound, choppy markets where price stays near all three MAs and the oscillator oscillates around zero without clear directional structure. The oscillator is measuring distance that does not tell you much when price is not trending on any timeframe. In those conditions, the regime color flickers constantly and the readings stay small. The oscillator is working correctly — it is just telling you there is nothing meaningful to measure.

Workflow 2: Cross-market MA-distance comparison

Purpose: Compare how far your primary instrument's price sits from its average against a broader market or reference instrument, on the same bounded scale.

Setup:

Chart: AAPL on 5m
Slot 01: 5m, EMA 20 on AAPL (chart symbol — no Optional Ticker)
Slot 02: 15m, EMA 20 on AAPL
Slot 03: 60m, EMA 20, Optional Ticker set to SPY
Sensitivity: 1.0
Weights: 33.3 / 33.3 / 33.3
On Bar Close: On

What to observe:

Slots 01 and 02 tell you how AAPL's price relates to its own 5m and 15m averages. Slot 03 tells you how SPY's price relates to SPY's 60m average. Because the oscillator normalizes each slot by the instrument's own ATR, the three readings sit on the same scale.

Compare the slot 03 (SPY) reading against slots 01 and 02 (AAPL). If AAPL's short-term slots are strongly positive (+70, +60) while SPY's session-level slot is negative or flat, AAPL is outperforming the broader market's recent direction. This is not a recommendation — it is an observation that the stock's MA-distance is moving against the market context.

If both AAPL's and SPY's readings are strongly positive, the move has market-level support. If AAPL turns negative while SPY stays positive, the stock is weakening against a still-healthy market.

When this pattern stops being useful: When the two instruments have no structural relationship. The ATR normalization makes the scores arithmetically comparable, but comparable numbers from unrelated instruments do not create meaningful composite readings. Blending AAPL with crude oil futures produces a number on the right scale that does not represent anything coherent. Use this workflow with instruments that share some logical connection — a stock and its sector ETF, a stock and the broad index, a forex pair and a correlated commodity. See Multi-Ticker Mixing for a deeper treatment of when cross-ticker comparison is useful and where it breaks down.

Workflow 3: Blend-only MA-distance pane

Purpose: Reduce the pane to a single composite MA-distance oscillator for a minimal chart layout.

Setup:

Chart: your preferred instrument and timeframe
Slot 01, 02, 03: configured to your preferred timeframes
Hide Plot: On for all three slots
Plot Blended Fast/Slow: On
Weights: set according to which timeframe you consider most important

What to observe:

You see one thick line (Blended Fast), one gray line (Blended Slow), and a fill between them. The composite reading tells you the weighted-average MA-distance across your configured timeframes.

This is the cleanest visual layout — no overlapping slot lines, no color complexity. Good for small screens, multi-chart layouts, or situations where you want a single "how stretched is price?" reading without the detail.

What you lose: All visibility into slot disagreement. The blend can show a calm near-zero reading while the individual slots are in active conflict. You cannot see whether the blend is being driven by one heavy slot or by genuine agreement. If you use this layout and the blend does something unexpected, un-hide the individual slots before trying to diagnose it.

Who this works for: Traders who have already calibrated their settings, trust their weight distribution, and understand the blend well enough that they know what it hides. This is a configuration for someone who has already done the learning. If you jump straight to this layout because it looks cleaner, you will not know when the blend is lying to you — and it will lie to you, quietly, every time the individual slots disagree.

Workflow 4: Confirmed-bar oscillator for strategy comparison

Purpose: Use the oscillator pane alongside a strategy script on the same chart, with confidence that the oscillator's historical values match what was available in real time.

Setup:

On Bar Close: On (critical)
All other settings: configured to match the timeframes and MA types your strategy process cares about
A strategy script or set of entry/exit markers on the same chart

What to observe:

Scroll through historical entries and exits. For each entry, look at the oscillator pane and note the Blended Fast and individual slot readings. Because On Bar Close is on, the values you see on those historical bars are the values that were available when each bar closed. The oscillator is not retroactively showing you a cleaner picture.

This lets you build genuine intuition: "When the oscillator looked like this, the trade did this." The correlation may or may not be useful, but at least it is honest. If you were doing this comparison with On Bar Close off, the oscillator values on historical bars would include HTF data that was not yet confirmed — making the correlation unreliable.

When to use this: Any time you want to evaluate whether the oscillator adds information to an existing strategy, or when you are building a discretionary process that incorporates the oscillator's readings and want to study how those readings appeared in the past.

When this gets dangerous: If you turn On Bar Close off and forget. The historical comparison becomes contaminated without any visible warning. There is no indicator on the chart that says "these values are from the future." The chart looks the same — it just lies about the past. See MTF & Repainting for the full risk analysis.

Workflow 5: Regime-alert monitoring with weight-zero isolation

Purpose: Monitor one timeframe's regime state through alerts without letting it influence the blended oscillator.

Setup:

Slot 01: 5m, weight 50
Slot 02: 15m, weight 50
Slot 03: 60m, weight 0
Slot 03 alerts: set up "MA Osc 03 Regime Flip" in TradingView

What to observe:

The blended oscillator reflects only slots 01 and 02. Slot 03's Fast line is still plotted in the pane (visible as a blue line), but it does not pull the blend. The slot 03 alert fires independently when the 60m regime flips.

This gives you a clean separation: the blend shows your short-term and intermediate MA-distance, while the 60m slot acts as an independent watchdog. You get alerted when the 60m regime changes without that change distorting the blend you are watching.

One catch: weight 0 does not remove slot 03 from the alignment alerts. If slot 03 stays enabled, it still counts in All Slots Bullish/Bearish.

When this is useful: When you want the blended oscillator focused on a narrow timeframe range (for example, 5m and 15m for intraday scalping) but you also want awareness of a larger structural shift. The weight-zero slot gives you the structural alert without the structural data muddying your intraday blend.

Anti-patterns

Anti-pattern 1: Same-timeframe stacking

What it looks like: All three slots set to the same timeframe (e.g., 15m / 15m / 15m) with different MA lengths or types.

Why people do it: They want to see multiple MA distances at different speeds — EMA 10, EMA 20, EMA 50 — on the same timeframe.

Why it fails the tool: The oscillator's architecture is built for multi-timeframe comparison. Each slot requests data at its configured timeframe through request.security(), and the blending assumes each slot brings a structurally different perspective. Three slots on the same timeframe with different MA lengths produce three measurements of the same data at different smoothing speeds. You get no timeframe diversification. The blend averages three variations of the same picture, which is not meaningfully different from a single slot with a mid-range length.

If you want to see multiple MA distances on the same timeframe, you can — but you should understand that you are using the tool for a purpose it was not designed for, and you are paying the cost of three request.security() calls for something that could be approximated more simply.

Anti-pattern 2: Backtesting with On Bar Close off

What it looks like: The oscillator is on the chart with On Bar Close disabled. The user scrolls back through history, notes how the oscillator "predicted" moves, and builds confidence in patterns that include future-leaked data.

Why people do it: On Bar Close off feels faster and more responsive. They forgot to turn it back on, or they do not understand the repainting consequence.

Why it fails the user: Every historical oscillator value includes information from higher-timeframe bars that had not yet closed at the time of the chart bar. The oscillator appears prescient — it seems to show extreme readings right before reversals, clean regime flips right before moves — because it knew the HTF outcome before it was finalized. Any pattern the user identifies in this mode is built on data that was not available in real time.

This is not a small error. It is the specific mechanism that builds false confidence in indicator-based setups, and it has a predictable lifecycle: the trader scrolls back, sees the oscillator "calling" turns with uncanny precision, develops a strategy around those patterns, goes live, and discovers the patterns do not hold — because they never existed in real time. The oscillator was not prescient. It was looking at the answer sheet. The trader built conviction on hindsight dressed as foresight, and that conviction cost real money.

What to do instead: Keep On Bar Close on for all historical review. Accept the one-bar lag as the cost of honest data. If the lag bothers you, that discomfort is telling you something important about how much you are relying on the most recent bar's value — which is the one bar that has not been confirmed yet regardless of this setting.

Anti-pattern 3: OB/OS as reversal triggers

What it looks like: The user sets an alert for "Blended MA Osc Overbought" and treats each alert as a sell signal (or at least a strong caution that a reversal is imminent).

Why people do it: Decades of RSI-based education have trained traders to associate overbought readings with impending reversals. The association is partially valid for RSI in range-bound conditions but does not transfer to this oscillator.

Why it fails here: The tanh-bounded oscillator's overbought zone has different properties than RSI's. The oscillator can sit above +70 for extended periods during trending markets because the distance from the MA in ATR units remains large, and the tanh function holds it near the bound. A reading above +70 that persists for fifty bars is not "failing to reverse" — it is accurately measuring that price remains stretched. The oscillator has no opinion about when the stretch will end.

Using OB/OS crossings as reversal triggers produces a specific failure mode: premature countertrend entries during sustained trends. The user sells when the oscillator crosses above +70, price continues higher, the oscillator stays above +70 (or goes to +90), and the trade loses.

What to do instead: Treat the OB/OS levels as attention thresholds, not action triggers. When the oscillator enters your defined zone, it means the distance from the average is large enough that you decided you want to notice. What you do with that notice — whether you tighten stops, reduce position size, look for structural reversal evidence elsewhere, or simply observe — is your decision, not the oscillator's.

Anti-pattern 4: Ignoring hidden slots in the blend

What it looks like: The user hides a slot's plot to clean up the pane, then interprets the blended line as if it reflects only the visible slots.

Why people do it: Hide Plot seems like it should mean "this slot is off." Visually, it does disappear. The mental model "if I can't see it, it is not there" is natural but wrong.

Why it fails the user: The hidden slot continues to compute, contribute to the blend, and fire its own alerts. If the hidden slot is in strong bearish territory while the two visible slots are bullish, the blended line will be pulled toward center (or below), and the user will not understand why. They see two bullish lines and a blend that contradicts them.

What to do instead: If you want a slot to stop affecting the blend, either disable it or set its weight to zero. Hide Plot is a cosmetic toggle — it affects what you see, not what the oscillator computes.

Anti-pattern 5: Maximum sensitivity without understanding saturation

What it looks like: The user sets sensitivity to 2.0 or 3.0 because they want the oscillator to be "more responsive" or to "reach the OB/OS zones more often."

Why people do it: The name "sensitivity" implies a straightforward more/less dial. Higher must be better if you want the oscillator to react.

Why it fails the understanding: High sensitivity does not make the oscillator more informative. It pushes the oscillator toward the bounds faster, which means it spends more time in the saturated, low-resolution zone near ±100 and less time in the high-resolution center. The oscillator becomes more decisive but less discriminating. At sensitivity 3.0, a moderate move that would produce a reading of +40 at sensitivity 1.0 already reads +90. The remaining range from +90 to +100 absorbs everything larger — from "moderately stretched" to "extremely stretched" — into ten points of oscillator space.

The user who sets high sensitivity and sees the oscillator frequently hitting ±100 may feel validated ("it is working hard"), but they have lost the ability to distinguish between moderate distance and extreme distance. The oscillator has become a binary above/below indicator with a fancy wrapper.

What to do instead: Start at 1.0. Look at a few weeks of data and check whether the oscillator uses a reasonable portion of the range. If it rarely reaches ±50, consider increasing sensitivity. If it is constantly near ±100, consider decreasing it. See Settings — ATR Sensitivity and For the Geeks for a deeper explanation and a hands-on experiment.

Adapting workflows to your own process

The workflows above use specific instruments and timeframes as examples. The principles transfer to any setup:

Choose slot timeframes that represent structurally different scales in your trading. A 5m/15m/60m configuration makes sense for intraday traders. A 15m/60m/Daily configuration makes sense for swing traders. The goal is that each slot tells you something the other two cannot.

Set weights based on which timeframe carries the most analytical weight in your process. If you make decisions primarily on the 15-minute structure and use the 5m for entry timing and the 60m for directional context, weight the 15m slot highest.

Start with On Bar Close on. Turn it off only after you understand the repainting consequence and have a specific reason that outweighs the loss of historical integrity.

Run the sensitivity experiment from For the Geeks on your own instrument before finalizing your configuration. What works on ES may be wrong for a low-volatility forex pair. Ten minutes with the sensitivity experiment saves weeks of reading an oscillator that is miscalibrated for your market.

Use the blend as a summary, not a substitute. Always have a way to check the individual slots when the blend does something unexpected. The blend is a convenience for reading the pane quickly — it is not a higher-quality reading than the individual slots. In fact, the individual slots always contain more information than the blend, because the blend averages away the very disagreement that matters most.