Workflows

The indicator is a workbench. It does not come with one "correct" way to use it, because the correct way depends on what question you are asking. What it does come with is a set of named patterns that have held up against real readers, and — more importantly — the anti-patterns they most often get confused with.

Every workflow below is framed the same way: what you are trying to learn, how to configure it, the reading it produces, and the anti-pattern you should not mistake it for. If you cannot state the question the workflow is answering out loud, the workflow is not the right one for that moment. Add slots only when you can name what each new slot is teaching you that the slots you already have are not.

Workflow 1 — The curated three-slot blend

What you are learning. Whether the short, medium, and longer intraday MACD reads on your chart instrument agree or disagree, and how the blend of the three is currently oriented.

Setup. Leave defaults. Three slots, 5m / 15m / 60m, equal weights, MACD 12/26/9, EMA on both the MACD line and the signal. Master smoothing off. OB/OS at 70/30 for reference only.

Reading. Three slot lines, each colored by its own fast-vs-slow posture. A blended fast/slow pair with a translucent fill in the regime color. A blended histogram columned in the four-state color code. When the three slot lines run roughly in the same direction and the blend's fill is solid in one color, the multi-timeframe convergence story is aligned. When the slot lines diverge — say 5m is above 50 and rising while 60m is below 50 and falling — the blend will sit near the midline and its histogram will be small. That is not a failure of the blend; that is the blend being honest about disagreement.

Anti-pattern to avoid. Enabling slots 4–10 "to see more." With default settings those slots would mostly replicate the same convergence story at more cadences, and you would end up reading ten lines that agree because they share ingredients, not because they have independently confirmed anything. The alignment alerts would look louder without the underlying pane getting any wiser. If you cannot state what slots 4–10 would be teaching you that 1–3 do not, they do not belong on the pane yet.

Workflow 2 — The weight-zero observer slot

What you are learning. Whether a slot you are curious about would have been contributing, without actually letting it contribute.

Setup. Enable the observer slot. Configure it deliberately — pick a timeframe, MA family, source, and (optionally) ticker that you have a reason to watch. Set its weight to 0. Leave the three default slots steering the blend.

Reading. The observer's line appears on the pane and is colored by its own fast-vs-slow posture. Its per-slot bullish/bearish alert is active — so MACD 04 Is Bullish fires on confirmed chart bars while slot 04's fast is above its slow — but the blend does not move because of slot 04. You can see what slot 04 is doing without letting it pull the blend around.

What to remember. The observer does count toward the alignment alerts (All MACD Slots Bullish / Bearish). Weight does not filter alignment; only disabled or na slots are excluded from that count. If you add an observer and then notice the alignment alert going quiet more often, that is the observer doing exactly what you added it to do — voting its own mind without steering the blend.

Anti-pattern to avoid. Adding three or four observer slots at weight 0 on ingredients similar to the active ones and then using alignment as "confirmation." You have not added confirmation; you have added voices that mostly echo the chart-instrument story and made the alignment alert easier to fire. If you want observers to be useful for alignment, differentiate them — a different MA family, a different source, a different ticker.

Workflow 3 — The cross-asset context slot

What you are learning. How another instrument's MACD posture sits at the moment, as context for what you are trading. Examples that come up: SPX while trading a single name, DXY while trading a dollar-sensitive pair, BTC while trading an altcoin, VIX proxies during equity hours.

Setup. Pick a slot you are willing to spend. Set its Optional Ticker to the context symbol (e.g., "SP:SPX" or "FX:DXY"). Pick a timeframe on that symbol that matches the question — often 60m or higher for a regime read, 15m for a faster context pulse. Choose weight deliberately: 0 if you want the cross-asset slot to be pure context, a non-zero value only if you genuinely believe that instrument should influence your blend.

Reading. The slot line reflects the cross-asset symbol's normalized MACD on its own timeframe, in its own ATR. It plots in your chart's pane only because that is where it has to go; the evidence is entirely from the other instrument. Its color, its posture, and its alerts are about that symbol.

Anti-pattern to avoid. Treating the cross-asset slot as confirmation of the chart symbol. It is not. A slot set to SPY on a chart of a single stock is telling you what SPY's MACD looks like. That is useful context when the single name is correlated; it is actively misleading when the single name is diverging from the index for its own reasons. And letting a cross-asset slot carry heavy weight into the blend means the blend is no longer a statement about the chart instrument — it is a mix of two instruments wearing one number. If the chart is AAPL and half of your blend weight is sitting on SPY, Blended MACD Is Bullish is a statement about SPY's contribution as much as AAPL's, and you should not pretend otherwise.

Workflow 4 — The two-MA-family comparison

What you are learning. How much the choice of MA family changes the normalized read on the same underlying data.

Setup. Enable two slots on the same timeframe (say, both at 15m), the same source, the same fast/slow/signal lengths, and differ only in the MA family. Slot A: EMA on both the MACD line and the signal. Slot B: ALMA or Jurik. Give them small weights (or 0) so they do not dominate the blend. Leave the default three slots running.

Reading. Slot A and slot B will agree most of the time — they are the same MACD, computed with different smoothers — and disagree at the family-specific moments. Jurik-on-signal tends to read less jittery around whipsaw; ALMA with certain offsets reads more responsive. You are not looking for a "winner"; you are calibrating what the choice is actually buying and costing on your instrument and your timeframe. That calibration is the workflow's whole value.

Anti-pattern to avoid. Keeping A and B running at meaningful weights in a production pane "for comparison." Once you have learned the difference, pick the family you believe in for the slot that needs to do work, and take the comparison slot's weight to 0 or disable it. Otherwise you are paying blend influence for a learning exercise that has already finished.

Workflow 5 — The master-smoothing regime read

What you are learning. A calmer, slower read of the blend, intended for regime orientation rather than real-time reaction.

Setup. Enable master smoothing. Pick a short MA (EMA length ~10 is enough to feel the effect; longer if you want a heavier filter). Keep slots 1–3 as they are. Observe the blend for a session.

Reading. The blended fast and slow will trail the slot lines. The fill between them changes regime later. The blended histogram reports a calmer version of the convergence story. That is what you asked for when you turned the switch on — less twitch, more posture. The slot lines underneath remain unchanged and unsmoothed. If you want to see what the slot lines are doing right now while the blend gives you where the blend's regime has settled, that is the stacked read master smoothing is for.

What to remember. Master smoothing applies only to the blended fast, slow, and histogram. Slot lines and per-slot alerts are unaffected. The blend alerts (Blended MACD Is Bullish / Bearish) are computed on the smoothed values — meaning they fire later than they would on the raw blend. Treat them accordingly.

Anti-pattern to avoid. Using a smoothed blend alert as an early entry trigger. It is not early; it is late. And treating master smoothing as a correctness improvement — "the blend looks clearer now, so it must be more right" — is mistaking visual calm for information. The underlying read is the same data with a lag.

A shared meta-anti-pattern

Across all five workflows, the single most common mistake is adding surface to compensate for uncertainty. When a reader is unsure what the chart is telling them, the temptation is to enable more slots, increase ATR Sensitivity, turn on master smoothing, and wire up more alerts. The pane gets louder; the read does not get sharper. The slot math does not multiply confidence when the ingredients overlap, and visual smoothing does not create information that was not in the raw data.

The opposite move — turning slots off, asking what question each remaining slot is answering, and tolerating a pane that reports disagreement honestly — is almost always the higher-value one. The workbench is designed to let you add as much surface as you can defend with a reason. Defending the surface is the work.

Where to go next

If you are building an alerting plan around these workflows, Alerts walks the 24 conditions and their gating. If you want the misread patterns named in full, Limitations & Trust Boundaries is the page that carries them. For the input surface behind the workflows, Settings walks every control and its tradeoff.