Workflows

A page about how this indicator shows up in actual trading, with two end-to-end workflows that a serious reader can adapt to their own process. Each workflow carries its own anti-pattern list — the ways the same tool gets turned into a footgun by a reader who skipped the reading work. The point of the anti-patterns is not to scold. It is to name the traps in enough detail that you recognize them the moment they start happening in front of you.

Neither workflow here is a trading system. This indicator does not make trading decisions, and nothing on this page is trying to dress up the pane as an entry or exit generator. The workflows are reading patterns: structured ways to use the pane as an input to a process you already have.

If you do not have a process the pane can feed into, this page will read as abstract. That is honest information. The pane sharpens an existing decision frame; it does not produce one. A reader who is still searching for a methodology will get more from time spent on the methodology than from time spent tuning a tool inside a process that does not exist yet.

Before either workflow makes sense

Two things need to be true before the workflows below are worth studying.

• You have a working concept of what CVD is and is not. CVD is a participation read, not a price read. If you are still catching yourself reading the pane as "where price is going next," slow down and re-read Visuals and Logic before going further.

• You have at least one instrument you know well. Both workflows depend on being able to tell when the pane is describing the market and when the pane is describing a quiet window. On an instrument you do not know, you cannot make that call reliably. Run the workflows on a familiar chart first and port them outward as your calibration grows.

Workflow 1: the context stack

What this workflow is for. Using three slot timeframes deliberately as a short/middle/long stack, and letting agreement or disagreement across the stack shape how much weight you give a setup.

What this workflow is not for. Producing entry signals. Producing exit signals. Making a single trading decision off a single slot's color. Turning the blend into a verdict.

The reading model

Pick three slot timeframes that line up with how you already think about the instrument:

• Execution frame. The fastest slot. Typically close to the timeframe you trade on, or one below it. This is the slot that moves the most and carries most of the short-term character.

• Confirmation frame. The middle slot. Usually three to four times the execution frame. This is the slot you check to ask "is this a momentary push or something developing."

• Regime frame. The slowest slot. Usually three to four times the confirmation frame, or at the natural regime cadence for the instrument. This is the slot that tells you whether the short-term push is running with or against the instrument's current character.

A reasonable concrete configuration for a 5-minute trader on a liquid intraday instrument is the default stack: 5m / 15m / 60m. A 15-minute swing trader on the same instrument might prefer 15m / 60m / 240m. The point is the ratio, not the specific numbers. Keep the ratios roughly consistent and let the cadence of your own process pick the numbers.

Leave weights equal on the first pass. Equal weighting is not the "correct" setting; it is the cleanest starting point. Move weights later, not first.

What you are watching for

• All three slots aligned. When the execution, confirmation, and regime frames all lean the same way, the participation context is unusually consolidated. This is the state that makes the All CVD Slots Bullish or All CVD Slots Bearish alignment alert meaningful. It is not a trade signal — there is no "correct" trade to take against aligned pressure — and the script does not prove a probability edge. It only tells you the three enabled reads agree right now. Whether that agreement has value depends on the instrument and the rest of your process.

• Regime and execution aligned, confirmation disagreeing. The confirmation frame is the middle frame for a reason — it often rotates between agreeing with the execution frame and agreeing with the regime frame. Disagreement on the middle frame, with the other two aligned, usually means the middle frame is rotating. Not information by itself; combined with price structure, sometimes useful.

• Regime disagreeing with execution. The slowest frame is leaning one way and the fastest is leaning the other. This is often the state of a counter-trend push into the regime. Whether that push continues or fails is not a question the pane can answer; what it can tell you is that the push is not (yet) being supported by the regime's participation context.

• All three spread. Each slot is in a different part of the pane. The pane is telling you the timeframes are arguing. The blend will sit near 50 and look quiet. Do not read that quiet as silence. See Visuals and Logic for the full ambiguity walkthrough.

A concrete reading example

Assume a 5-minute chart on a liquid equity-index future during regular trading hours, with the default 5m / 15m / 60m stack, weights 33.3 each, Session mode with D window, On Bar Close ON.

• 09:35: chart opens, slots are warming up. Ignore the first three or four chart bars.

• 09:55: CVD 01 at 74 (lime), CVD 02 at 68 (aqua lime), CVD 03 at 62 (blue lime). Blend at 68, well above its signal. All three slots lime. The participation context says buyers have been present on every frame since the open. If price makes a pullback here, the stack's posture is "the pullback is into a session that has been bought so far," which is different from "the pullback is into mixed context." That difference is the useful information.

• 11:10: CVD 01 at 42 (faded teal), CVD 02 still at 62, CVD 03 at 58. Slot 01 flipped faded. The execution frame is reporting short-term selling participation; the confirmation and regime frames are still leaning bullish. The stack is telling you the short-term push has reversed inside a still-bullish day. Whether that matters for the decision in front of you depends on your strategy; the stack has given you information, not an opinion.

• 13:15: CVD 01 at 28 (faded), CVD 02 at 34 (faded aqua), CVD 03 still at 58. The middle frame has now joined the short frame. Only the regime frame is still lime. The stack is telling you that the bullish context is wearing thin across the faster frames and hanging on only at the regime cadence. Again: not a trade signal. The reading is "the consolidated bullish context from this morning is eroding."

• 14:40: all three faded. The All CVD Slots Bearish alert fires on this bar's close. The stack has fully rotated.

None of this is a trading system. The day is full of decisions no single alignment state can make for you. What the stack is doing, bar after bar, is keeping your read of participation honest across three frames at once, so that you do not accidentally take a short-frame signal as if it were operating in a vacuum.

Anti-patterns in the context stack

• Taking an alignment alert as a trade trigger. Alignment is a state. The alert fires on every confirmed bar that the state holds. Wiring it to an automated entry will produce an entry on the first bar of alignment, and then another entry on the next bar, and the next. That is not what the alert is for. Alerts has the full description.

• Running a stack where the three timeframes compress together. 5m / 10m / 15m is three slots that see almost the same thing. The middle slot stops doing the rotating work it is supposed to do, and the regime slot stops describing a regime. If your three slots have ratios less than roughly 2x apart, you are not running a context stack — you are running three copies of the same read with slight smoothing differences.

• Running a stack where the three timeframes are so spread that they never interact. 5m / 240m / 1W might be legitimate on a swing-trading chart, but on an intraday chart the 1W slot will just sit there and never contribute usable movement. Keep the top slot inside a cadence that actually changes during your session.

• Changing the weights before understanding the instrument. Equal weights is the honest starting point. A reader who starts with a 50/30/20 weighting because "my execution frame matters more" is pre-committing a reading choice to an instrument they have not yet observed. Watch equal weights first. Move weights when you can describe, in a sentence, what you are correcting.

• Forgetting that Hide Plot does not remove a slot from the blend. Three slots you can see plus a fourth slot of your own attention on a slot you hid — and then wondering why the blend is "off." It is not off. The math is including the hidden slot. If you want it out of the blend, set its weight to 0.

• Mixing Session and Rolling on three slots for no specific reason. Sometimes the right call (a rolling intraday slot against a session regime slot). Often just a configuration the reader set up once and forgot. If you are mixing modes, be able to name why.

Workflow 2: alert triage

What this workflow is for. Using the state-descriptor alerts as a feed into a review process, not an execution process. The pane tells you that pressure has consolidated; you decide, looking at price and the rest of your context, whether that is a thing you act on.

What this workflow is not for. Automated entries. Automated exits. Any flow where an alert triggers a trade without a human read in between.

Why triage, not execution

The alerts on this trim are repeat-fire. The alignment alerts fire on every confirmed bar that the alignment holds. The per-slot alerts fire on every confirmed bar the slot state holds. That is the correct behavior for describing a state; it is terrible behavior for anything resembling an entry trigger.

Triage is the pattern that respects this. You wire alerts that tell you something about the state of the pane, you receive those alerts into a channel you can see (webhook, Discord, pager), and when one fires you look at the chart and decide. The alert is the prompt; the decision is yours.

The minimal triage configuration

• Wire All CVD Slots Bullish and All CVD Slots Bearish as the "stack consolidated" pair. These are the alerts you want fast — alignment across all three frames is relatively rare and worth seeing quickly.

• Wire Blended CVD Is Bullish and Blended CVD Is Bearish as the "summary changed" pair. Noisier than the alignment pair, but useful as a secondary prompt.

• Do not wire the six per-slot alerts unless you have a specific single-slot question. They will mostly reproduce what the blend alerts already tell you, with more noise.

Configure your alert sink to handle repeats sensibly. "Fire every bar close while the state holds" is what TradingView will do; "ping me once and then stay quiet for fifteen minutes" is a receiving-side policy you have to implement. Common receiving-side rules: ignore repeats within N minutes of a prior alert on the same state; count alignment alerts as a single "consolidated" event and only re-notify on a change of direction; batch per-bar alerts into a rolling summary.

The triage process

When an alert fires:

1. Name what the alert is telling you, in one sentence. "The three slots have aligned bullish on the 5-minute chart." Writing it or saying it out loud forces the correct framing and prevents the drift into "this alert says go long."

2. Pull up the chart. Do not skip this step even if the decision feels obvious. The purpose of triage is to make the chart read the center of the decision, not the alert.

3. Check the alignment is not a post-reset artefact. If a slot window just rolled over in the last couple of bars, alignment is often spurious. The Visuals and Logic page covers the post-reset calibration behavior. A pause of several bars after any window reset is the right posture before trusting alignment.

4. Check the stack against price. Does the price structure support the participation read? If the stack is aligned bullish during a meaningful pullback into a known support zone, the read has a plausible home. If the stack is aligned bullish in the middle of a range with no structure nearby, the read is context, not opportunity.

5. Decide. The decision is yours, it is not the indicator's. Common outcomes: act on an existing plan whose preconditions now include the alignment read; hold off because the price structure does not support the read; log the alignment for review and do nothing. All three are legitimate triage outcomes.

6. Stop looking at the alert. Repeat-fire alerts will keep arriving while the state holds. Once the triage decision is made for this alignment, further repeats of the same state are not new information. Your receiving-side policy should have silenced them already; if it did not, silence them manually.

What the alert is doing to your attention

Worth naming, because it bites people who are otherwise careful. An alert that fires while you are doing something else captures attention with a force that is not proportional to its content. A bullish alert on the alignment alert at 2pm will pull your eye to the chart even if it is the seventh repeat of an alignment that started at 1pm. The first one was useful. The other six are noise that the receiving-side configuration should already be silencing — and even when it is, the act of glancing at the chart on each fire pulls you out of whatever else you were doing.

The defensible posture is that alerts on this pane belong in a channel you check on your own cadence, not a channel that pages you. The alignment alerts are meaningful enough to be worth a glance when they first appear. They are not meaningful enough to interrupt anything. Setting up the alert sink so the second through twentieth repeats do not interrupt is not a TradingView problem; it is a discipline you build into the channel that receives them.

Anti-patterns in alert triage

• Treating the alert as a trigger. The state-descriptor design is categorical on this pack. An alert is not a signal. The receiver has to impose any execution semantics on top.

• Wiring the per-slot bullish/bearish pair with no receiving-side deduplication. The result is a full-time buzz. On most active instruments, at least one per-slot alert is in the "state holds" condition at any moment. Your channel will be constantly alerting. This makes the alerts useless — when everything is alerting, nothing is.

• Acting on an alignment alert fired during a session reset. The first bar or two after a reset can produce spurious alignment because the slots have just been renormalized. Build the post-reset pause into the process and the noise drops out.

• Treating the absence of alignment as a sell or short signal. Alignment is a relatively rare state. Most of the time the three slots are not all on the same side. The normal state of the pane is mixed, not aligned. "The alignment alert has not fired in an hour" is a legitimate description of the pane; it is not a bearish signal.

• Letting the triage process degrade into "fire and forget." The point of triage is the human read. If you find yourself responding to alerts without looking at the chart, either tighten the alert set until the alerts are rare enough that you look at each one, or rethink whether alerts are the right tool for the current process.

Optional: session versus rolling as a workflow choice

This is a shorter, optional third workflow about the Window Mode selection, because the right choice depends on how you think about the instrument more than on anything in the indicator.

Session mode is the right default when the instrument respects a specific clock — regular trading hours on an equity, the open of a futures contract, the daily cut on an FX pair, anything where "today" and "yesterday" are different things to a participant. Session mode gives you cumulative deltas anchored to that clock, and the dashed reset markers on the pane give your eye a landmark it can use.

Rolling mode is the right default when the instrument does not respect any specific clock — 24-hour crypto, some FX, anything where the notion of "a session" is mostly an arbitrary slicing of continuous activity. Rolling gives you a consistent lookback that does not privilege a specific boundary. The absence of reset markers is deliberate; there is no reset event to mark.

A mixed configuration — say, CVD 01 in Rolling and CVD 02/03 in Session — is occasionally the right answer. The case where it shines is when you want a rolling intraday view on your execution frame while still honoring the session anchor on the regime frame. The case where it confuses you is when you set it up once, forget you set it up, and later try to read the stack without remembering which slot is on which mode.

If you are mixing, leave yourself a note. Seriously. The pane does not label which slot is in which mode; your memory has to, and memories of indicator configurations are short.

A harder workflow pitfall — emotional fit

A pressure indicator does something psychological that a price indicator usually does not. It gives a name to the read underneath a decision that did not go well. A losing trade with the pane disagreeing in hindsight is not evidence that the pane is right and you were wrong; often it is hindsight finding a signal that was not actually visible at the time. But the pane is a tidy-looking surface, and the eye wants to find a reason the loss was avoidable. That search, repeated enough times, turns the pane into a therapy tool instead of a reading tool.

If you catch yourself doing this, take the pane off the chart for a session. Trade without it, write down the decisions and their reasoning, and put the pane back on the next session. A tool that is honestly useful to a process will be missed when it is not there. A tool that was mostly post-hoc explanation will feel like nothing changed. Running the experiment is cheaper than not knowing which one you had.

A note about the defaults

The default stack, the default weights, the default lengths, the default Pressure Sensitivity, the default Wick Weight — these are all reasonable starting points on liquid intraday instruments. They are not universal. A reader who ports the defaults straight onto a session-traded equity in overnight hours, a thinly-traded altcoin, a weekly chart, or a product the defaults were never calibrated against will get a pane that is doing its job honestly and producing reads that are hard to use. That is not a bug; that is the cost of defaults being defaults.

The workflows above are the part of the pack that turns a default configuration into a deliberate one. Sitting with the defaults for a session, writing down what you want to change, and then changing one thing at a time is the faster path than trying to tune three things simultaneously and losing track of which change caused which behavior.

Related pages

• Alerts for the mechanics of each alert and the state-descriptor design.

• Settings for the specifics of slot identity, weighting, window modes, and smoothing choices.

• Visuals and Logic for the ambiguous-middle reading that both workflows depend on.

• Limitations and Trust Boundaries for what these workflows cannot tell you and what the pane will not reveal.