Axiom CVD Osc

Axiom CVD Osc is a three-slot cumulative-volume-delta oscillator. Each slot runs on its own configured timeframe at or above the chart timeframe, estimates signed volume from lower-timeframe OHLCV using a wick-aware participation model, and maps its cumulative delta into a bounded 0-to-100 pane. The three slots fold into a single weighted blend so you can watch several timeframes argue and see where the weight of the argument is leaning.

The whole pack is built around one idea: a CVD line is only useful if you know what it is reading and what it is not reading. Most CVD tools hand you a picture and stop there. This one hands you the picture, names the estimator underneath, and keeps inviting you to check it. If that posture is unfamiliar, the rest of this page is the orientation that makes it worth your time.

Where this version sits

Axiom indicator families use a Base -> CTX -> STR progression when the full set exists. Base is the free, focused version: three context slots, chart-symbol only, limited filtering, and one global bar-close posture. CTX expands that same core CVD logic into broader context: up to ten slots, per-slot symbol and timing control, and deeper Power User surface for shaping the read. STR is the structure expansion at the end of the series: CTX-style per-slot controls with five slots instead of ten because the extra processing budget goes into structure on the blended CVD itself.

This page covers Base. Use it when three CVD slots are enough and you want the smallest surface that still gives a legitimate working read. Reach for CTX when ten-slot context breadth is the job. Reach for STR when you want the more advanced structure read around the blended CVD and are willing to trade away five CTX slots to get it.

What the pane actually shows

When you add the indicator, three lines appear inside a dedicated pane beneath price, all bounded between 0 and 100:

• CVD 01 on the 5-minute timeframe (teal)

• CVD 02 on the 15-minute timeframe (aqua)

• CVD 03 on the 60-minute timeframe (blue)

Each slot draws one line — its smoothed, normalized CVD — and flips between a bright "up" tone and a faded "down" tone depending on whether that slot's CVD value is above or below its own signal line. The blend draws two more lines on top of the stack: a lime-or-red blended CVD, a gray blended signal, and a translucent fill between them whose tint tracks the current blend state. Horizontal guides sit at 0, the oversold level (default 20), 50, the overbought level (default 80), and 100. In Session window mode, a dashed vertical line in the slot's color marks each window rollover. That line is an accounting event, not a trade event, and the rest of the pack keeps teaching that distinction.

The thing to internalize on first read is that everything you see on the pane is a read about participation, not a read about direction. Price can rally against selling pressure. Price can sell off against buying pressure. When that happens, the slots are not "wrong" — they are doing the job the pane was built for. The rest of this pack is about keeping that distinction alive when a chart is live and the stakes are not hypothetical.

Why this tool exists

Traders reach for CVD to answer one honest question: is there real participation behind the move in front of me, or is price just drifting? That question deserves better than most of the answers the market has for it.

• Raw cumulative lines drift forever. A CVD line that never resets keeps climbing because the number keeps getting bigger. Visual slope on that kind of line is not an instrument; it is a story the eye tells itself. Every slot in this pack runs inside a window you choose, and every line is renormalized into 0-to-100 using the window's observed high and low, so the slope you see is the slope against a reference frame you picked.

• Naive body-direction models call every green bar buying and every red bar selling. That is correct on decisive bars, wrong on indecisive bars, and actively misleading on bars where a long wick is the real event. The estimator underneath each slot reads body, close location, and wick asymmetry as separate pieces of evidence and classifies the bar into strong, weak, or indecisive buckets. The behavior degrades gracefully on bars that do not have a lot to say; it does not invent conviction where there is none.

• Most CVD tools never name the data source. The delta a TradingView indicator can compute is not tick-resolution bid/ask delta. It is an estimate built on top of OHLCV at a finer timeframe. If a tool does not say that out loud, a reader ends up trusting the line more than the line has earned. This script tells you in its own preamble that it is an estimate, and the pack keeps that word in front of you every place it matters.

• Window choice is usually hidden. A CVD read from a daily session reset and the same data inside a rolling lookback are not the same line. Each slot here exposes the choice, and a dashed vertical marker shows you where the reset landed when you are in Session mode.

Where a lot of CVD tools hand you a picture and hope you stop asking where the numbers came from, this one hands you the estimator and keeps inviting you to ask.

The three-slot frame, quickly

Three slots is enough to make a multi-timeframe argument legible. It is not enough to feel like a workbench, and that is intentional. A stack of three gives you:

• A working frame near your execution cadence — the lowest timeframe on the stack.

• A confirmation frame one step up — a slower version of the same read, moving in arcs rather than jitter.

• A regime frame on top — a gentler drift that tells you what the day or the week looks like relative to itself.

The default 5m / 15m / 60m stack is a reasonable starting point for intraday reading on a liquid symbol. It is not a universal truth. A session-traded equity, an overnight futures product, a thinly-traded pair, a weekly chart — each wants something else, and the Settings page walks you through what changes when you change a slot's timeframe. The pack will keep returning to this: defaults are a working chart, not a prescription.

The blend is a summary of the three-slot argument. It is not a verdict. When the three slots agree, the blend says so cleanly. When they disagree, the blend averages the disagreement, and a reader who treats the blended line as the verdict will end up on the wrong side of contested pressure more often than they expect. The Visuals page walks through a specific version of this trap — the blend sitting near 50 while the underlying slots are spread — because it is the single most common way this pane lies to a reader who stopped looking at the stack.

What this is not

• Not a signal. A slot flipping color, an alignment alert, the blend crossing its signal — these are state descriptors. They describe the state of the pressure read. They are not entries, not exits, not triggers. The Alerts page is explicit about this, and the "All CVD Slots Bullish" alert is framed the same way: it tells you something about how three estimators happen to agree right now, not something about where price is going next. Wiring it to an automated entry will fire that entry on the first bar of agreement, then again on the next bar, and the next, until the agreement breaks. That is the design behaving correctly. It is also the wrong tool to put in front of a router that buys on every ping.

• Not footprint or bid/ask delta. The delta under each slot is estimated from lower-timeframe OHLCV. That estimate is defensible, and For the Geeks walks you through the mental model, but no knob in this indicator turns it into tick-resolution data. If your process requires footprint data, this is not the right tool, and no setting on this trim will change that.

• Not repaint-proof. It is repaint-honest. On Bar Close is a global switch; ON waits for the slot-timeframe bar to finish before publishing, which costs one slot bar of lag and gives you a settled read; OFF uses the live slot-timeframe bar, which is faster but can update until that slot bar closes. The MTF & Repainting page walks through both modes and the tradeoff, and the default ships as ON because a reader new to the indicator should see settled values first.

• Not a replacement for price action. The pane describes participation. Your chart describes price. They are different readings and they answer different questions. The pane is useful when it is read alongside your chart, not in place of it. A reader who tries to make the pane carry both jobs will end up annoyed at the pane for not doing the job it was never built to do.

• Not the "beginner version" of the family. The base trim is compact. It is not lesser. A serious trader who chooses to run three slots instead of ten is making a reading choice, not settling. Three slots is enough to make a multi-timeframe argument legible, and not so many that the eye loses track of who is saying what.

Who this is likely to serve

• A trader who is newer to Axiom's take on CVD and wants a focused instrument to learn on before reaching for the wider workbench.

• A trader who already reads multiple timeframes and wants three pressure reads they can compare at a glance.

• A trader who has been stung by a CVD tool that looked sharp on a backtest and fell apart on live data, and who wants a pack that names the estimator's boundaries every place a reader might skim.

• A trader who respects their own time enough to learn one configuration well rather than skim three.

Who will be better served elsewhere

• A trader who needs cross-symbol pressure reads (a spot feed versus a perp, an index against a constituent). The optional-ticker slot lives in the CTX trim.

• A trader who wants fine-grained control of the smoothing families with Power User parameters like ALMA Offset, KAMA/FRAMA Fast and Slow, Jurik Phase or Power, Laguerre Alpha, or VAMA Volatility Length. Those controls are in the CTX trim. This pack points at the Axiom MA library manual when you want family-level context, but it does not expose those knobs itself.

• A trader who wants a buy-sell arrow. The pack's alerts are state descriptors and nothing in it will ever become an entry generator.

• A trader who has not yet sat with the distinction between price (direction) and CVD (pressure). The pane will still read as abstract until that distinction is internalized. A primer on that, not a product manual, is the better first stop.

How to read this pack

Three reading modes are supported on purpose:

• Fast orientation. README, then Quick Start. Twenty minutes of reading and one chart session and you have a working pane and a short list of traps. Stop there if that is all you came for.

• Active configuration. Settings, then Visuals and Logic, then Workflows. This is where the pack expects most of its time to be spent. The order matters — pick the slots and windows first, learn what the pane is showing once those choices are made, and only then sit with the workflows that turn the pane into a part of a process.

• Trust calibration. Limitations, MTF and Repainting, and For the Geeks. These are the pages a reader returns to when something feels off, when an alert behaves strangely, or when a trade is about to be sized larger than usual. None of them are decoration. The pack invests heavily in them because the moments they help with are the moments that actually matter.

If you find yourself reading every page in order on the first pass, slow down. The pack rewards being open at the right time more than it rewards being read end to end.

Where to go next

• First time opening the indicator. Read Quick Start. One validated setup, one sanity check, what a healthy pane looks like on first load, and the traps that will make you think something is broken when nothing is.

• Configuring slots for your own process. Settings walks the inputs in the order a real configuration session uses.

• A live read feels ambiguous. Visuals and Logic teaches the stack-versus-blend read, the ambiguous middle, and the states most likely to be over-read.

• Deciding whether to wire alerts. Alerts is explicit about what each alert commits to, what it refuses to commit to, and what the script does not expose.

• Serious use. Workflows walks two end-to-end patterns — the context stack and alert triage — each with its own anti-pattern list.

• Calibrating how much weight to place on the tool. Limitations and Trust Boundaries is the honest inventory of what this pane can and cannot tell you.

• Understanding the slot-timeframe plumbing and the repaint tradeoff. MTF and Repainting.

• Wanting the participation model explained. For the Geeks. Mental-model altitude — enough to decide whether to lean on the estimator, not enough to rebuild it.

• Something on the chart does not look right. Troubleshooting is organized by what you are seeing, not by what setting caused it.