Settings

Settings

This page covers every configurable setting in Axiom CVD Osc Pro. The settings are organized by the decisions that matter most, not by their order in the panel. If you are configuring from scratch, work through the first six sections in order — they cover the choices that shape what the oscillator shows you and how much you can trust it. Everything after that is refinement.

The Pro variant has ten independently configurable slots. Each slot carries the same settings structure. This page teaches the pattern once and then documents what varies between slots.

The slot pattern

Every slot has the same configurable surface:

• Enable / Hide Plot — whether the slot runs, and whether it draws. These are two separate switches and the distinction matters. A disabled slot does nothing — it does not compute, does not contribute to the blend, does not fire alerts. A hidden slot (Enable ON, Hide Plot ON) computes fully, contributes to the blend at its full weight, and fires its alerts — you just cannot see its line on the chart. This is the most common source of "why doesn't the blended line match what I see?" confusion. If you are debugging the blend, always check for hidden slots before anything else.

• Timeframe — what resolution the slot computes at

• Lower TF Precision — how granularly the participation model reads each bar

• Window Mode and Window — how the slot accumulates and normalizes

• CVD Length / Type and Signal Length / Type — how the raw delta is smoothed

• Blended Weight — how much this slot matters in the composite

• Pressure Sensitivity and Wick Weight — how the participation model classifies bars

• Optional Ticker — whether the slot reads a different symbol

• On Bar Close — whether the slot uses confirmed or live data

• Power User MA params — advanced knobs that activate only for specific MA types

Slots 01-03 are enabled by default with a 5m/15m/1h timeframe spread. Slots 04-10 are disabled. The defaults give you a working three-slot setup from first load. The remaining seven slots are there when your questions get more specific — not as a suggestion to enable them all.

On Bar Close

Setting: On Bar Close? Default: ON (all slots) Location: PU Settings (per-slot)

This is the most important setting in the indicator. It controls whether each slot uses confirmed historical data or live-updating data from the building higher-timeframe bar.

When ON (default): The slot displays the CVD and Signal values from the last completed higher-timeframe bar. The current bar's values do not appear until that bar closes. Historical values are stable — they will not change after the fact.

When OFF: The slot displays the current HTF bar's values as they form in real time. The oscillator responds faster, but historical values reflect each bar's final state, not what was visible while it was forming.

The tradeoff is not about speed preference — it is about what kind of trust you can place in the chart's history. With On Bar Close on, you can scroll back and study the oscillator knowing it matches what was visible live. With it off, the chart looks better in hindsight than it did in real time. If you use the oscillator for historical study, comparison, or any kind of pattern review, leave this on.

This is a per-slot setting in the Pro variant. You can run some slots confirmed and others live. There are valid reasons to do this — see MTF and Repainting for a full treatment of the mechanics and how to verify behavior yourself.

Timeframe

Setting: TimeFrame Defaults: Slot 01 = 5m, Slot 02 = 15m, Slot 03 = 60m, Slots 04-10 = empty

The timeframe at which the slot computes and accumulates CVD. If left empty, the slot uses the chart's timeframe.

Constraint: The slot timeframe must be equal to or greater than the chart timeframe. Setting it lower produces a runtime error and the indicator will not load. This is a safety rail — requesting data at a resolution finer than the chart cannot provide meaningful results.

When to change: Set each slot to a timeframe you actually analyze. The standard approach is progressively higher timeframes: something like 5m/15m/1h for intraday, or 1h/4h/D for swing. The exact values depend on your chart timeframe and the timeframes that matter in your workflow.

What to avoid: Setting multiple slots to the same timeframe produces duplicate readings. The blend treats them as independent data when they are not, which creates the appearance of multi-timeframe agreement where none exists. Use meaningfully separated timeframes.

Setting adjacent timeframes too close together (5m/6m/10m) produces highly correlated readings. The slots will move nearly in lockstep, and their agreement tells you less than it appears to.

Lower TF Precision

Setting: Lower TF Precision Defaults: Slot 01 = 1m, Slot 02 = 1m, Slot 03 = 5m, varies for 04-10

This controls the sub-timeframe resolution used to fetch intrabar OHLCV data for the participation model. For each bar at the slot's timeframe, the indicator pulls candles at this lower resolution, classifies each one through the participation model, and sums the directional contributions to estimate that bar's net delta.

Constraint: Must be strictly lower than the slot's timeframe. Setting it equal or higher produces a runtime error.

The tradeoff: Lower values (more granular) give the participation model more sub-bars to classify. This can produce a more detailed estimate but increases data requests and may introduce noise from very small candles on some instruments. Higher values (coarser) produce fewer sub-bars per slot bar — a rougher estimate, but potentially more stable.

Fallback behavior: When intrabar data is unavailable (sparse history, symbol limitations), the indicator silently falls back to estimating delta from the slot bar's own OHLC and volume. The oscillator still works, but with less granularity. The indicator does not warn you when this happens.

Window Mode

Setting: Window Mode Default: Session (all slots) Options: Session, Rolling

This changes what the oscillator remembers and how it normalizes — and it changes it fundamentally enough that two users looking at the same slot on the same chart can see very different pictures if one uses Session and the other uses Rolling.

Session mode anchors the CVD accumulation to a reset boundary defined by the Window setting. At each boundary (daily by default), accumulated delta, period high, period low, and the normalization range all restart from zero. The oscillator answers: "Where is net estimated pressure within this session?"

Rolling mode uses a continuously sliding window. Old bars age out as new bars enter. There are no resets. The accumulation and normalization range drift as the window contents change.

The difference is not cosmetic.

Session mode gives a clean per-session read but loses all prior context at each reset. The first bars after a reset produce thin-data readings — the normalization range is narrow, so even modest volume can push the oscillator toward extremes. A +80 five bars into a new session is not the same evidence as +80 after 200 bars.

Rolling mode maintains continuity across session boundaries but introduces a different artifact: the oscillator can change direction not because new volume is directional, but because old directional volume is leaving the window. A reading that drops from +60 to +30 without any new bearish bars may mean that the strongly bullish bars from earlier have aged out of the window.

When to change: Session mode is natural for traders who think in daily or weekly cycles and want a fresh read each day. Rolling mode suits instruments that trade near-continuously (crypto, futures) or traders who want a pressure view across session boundaries. The choice depends on how you think about time and what question you are asking the oscillator to answer.

Window

Setting: Window Default: D (daily, all slots)

In Session mode, this sets the anchor period for the reset boundary. D resets at the daily session boundary. W resets weekly. In Rolling mode, this sets the lookback duration of the sliding window.

Constraint: Must be equal to or greater than the slot's timeframe. Setting it below produces a runtime error.

What "too short" looks like: A narrow window means the normalization range rebuilds frequently (Session) or covers very few bars (Rolling). Individual bars have an outsized effect on the normalized reading. You may see the oscillator swing from +60 to -40 on two consecutive bars — not because pressure reversed, but because the range is so narrow that any directional bar dominates. The oscillator becomes reactive rather than contextual.

What "too long" looks like: A very wide window accumulates so much data that the normalization range becomes anchored to old extremes. Recent pressure shifts barely move the needle because the range is wide and the current delta is a small fraction of it. A fresh directional push that would read as +50 in a daily window might register as +10 in a weekly window because the weekly window includes a much larger historical range. The oscillator becomes sluggish and loses its ability to surface what is happening now. If you find yourself staring at a flat-looking oscillator during a session that clearly has directional pressure, the window may be too long.

Practical guidance: For intraday Session-mode slots, a daily window is the natural starting point — it gives a fresh read each day with enough bars to build a meaningful range within the first hour. For Rolling-mode slots, the window should span the analysis horizon you care about. A 4H or 8H rolling window shows developing context within a session. A 2D window carries overnight flow. A weekly window provides broad context at the cost of responsiveness.

Pressure Sensitivity

Setting: Pressure Sensitivity Default: 1.50 Range: 0.25 to 4.0, step 0.05 Location: Per-slot

This controls how aggressively the participation model classifies sub-bars into directional buckets. It is the most impactful tuning parameter for how the oscillator reads volume — and in the Pro variant, it is configurable independently per slot.

At higher values, the model is more decisive. More bars land in the strong bullish or strong bearish buckets, and ambiguous bars carry more of the prior classified direction. The oscillator swings further and transitions more sharply. On instruments with clean candle structure and decisive trending, this can produce a more vivid read. On choppy or ranging instruments, it over-classifies noise as directional commitment.

At lower values, the model is more conservative. It requires stronger candle structure before committing to a directional bucket. Ambiguous bars carry less prior bias. The oscillator is calmer and less prone to whipsawing, but it may underweight genuine moves.

Do not treat higher sensitivity as "more accurate." It is more decisive, which is a different thing. Higher sensitivity means the model is more willing to commit to a direction — that confidence is earned during trends and overextended during chop.

What "too high" looks like on-chart: The oscillator whipsaws between bullish and bearish regime on nearly every bar during ranging or low-volume conditions. The slot line swings aggressively between +50 and -50 even though the candle structure is mixed and indecisive. The blend looks hyperactive. If you find yourself wondering whether the market is really this directional — and the answer is no — sensitivity is too high.

Why per-slot tuning matters: A 5m slot on a volatile crypto pair may want lower sensitivity (the candle noise is real and should not be over-classified). A 1h slot on the same pair may tolerate higher sensitivity because the hourly bars have already absorbed some of that noise. Per-slot tuning lets you match the model's aggressiveness to each timeframe's characteristics.

For the mechanics of how Pressure Sensitivity shifts the classification model, see For the Geeks.

Wick Weight

Setting: Wick Weight Default: 0.20 Range: 0.0 to 0.50, step 0.05 Location: Per-slot

Controls how much wick structure influences the participation model's classification of each sub-bar. Higher values give wick rejection more power to shift a bar's directional bucket. At zero, the model ignores wicks entirely and classifies based only on body direction and close position.

When wicks are informative — instruments with genuine rejection wicks, like crypto pairs that spike and reverse, or commodities during thin-volume periods — increasing wick weight may improve the estimate by letting the model account for rejection dynamics that body direction alone misses.

When wicks are noise — low-liquidity instruments where wicks reflect erratic fills rather than genuine price rejection, or very short timeframes where a single tick can create a disproportionate wick — reducing wick weight keeps the model focused on body structure.

The interaction with Pressure Sensitivity: These two knobs work together. Pressure Sensitivity determines how easily a bar reaches a strong directional bucket. Wick Weight determines how much wick rejection can redirect that classification. A bar with a strong bullish body but a long upper wick (rejection) will be classified differently at Wick Weight 0.0 (body wins) than at Wick Weight 0.40 (wick rejection pulls it back toward neutral or even bearish). Adjusting one without considering the other can produce unexpected behavior. Change them one at a time and observe the difference.

CVD Length, CVD Type, Signal Length, Signal Type

Settings: CVD Length / CVD Type / Signal Length / Signal Type Defaults: Length = 3, Type = SMA (all four, all slots)

These control the smoothing applied to the raw normalized CVD value and the Signal line derived from it.

CVD Length and Type determine the first smoothing pass — from raw normalized delta to the displayed CVD line. Longer lengths or heavier MA types (EMA, ALMA, Jurik) reduce noise but add lag.

Signal Length and Type determine the second pass — from the CVD line to the Signal line. The Signal is what determines regime classification: bullish when CVD is above Signal, bearish when below.

The gap between CVD and Signal is where regime lives, and how you set these lengths shapes what "regime change" means. If both lengths are short (3 and 3, for example), the Signal tracks the CVD line closely. The regime flips frequently and responds to small pressure shifts — useful if you want sensitivity, but prone to whipsawing during noise. If both lengths are long, the Signal lags the CVD substantially. Regime transitions are heavily delayed — you see the CVD line decline for many bars before it finally crosses below the Signal and flips the regime label. The move is already well underway by the time the label changes.

A common approach is to keep the CVD length short (for a responsive pressure line) and the Signal length slightly longer (for a more stable regime boundary). This creates a setup where you can watch the CVD line's direction for early signs of pressure change while the Signal provides a slower, more deliberate threshold for calling a regime flip. But "slightly longer" depends on the slot's timeframe and the instrument's noise level. On a volatile 5m slot, a CVD length of 3 and Signal length of 7 may be reasonable. On a smooth 1h slot, the same spread might add unnecessary lag. Start with the defaults (3/3), observe how often the regime flips on your instrument, and adjust from there. Change one length at a time so you can isolate the effect.

The Pro variant offers the full Axiom moving average library, including SMA, EMA, WMA, DEMA, TEMA, ALMA, KAMA, FRAMA, Jurik, Laguerre, VAMA, and others. Each MA type has different lag and smoothing characteristics. If you are not sure which to use, SMA and EMA are reliable starting points. The more exotic types are available for traders who have specific smoothing requirements — they are not "better," they are different. Picking an unfamiliar MA type without understanding its response characteristics can produce regime behavior you did not intend. If you want to experiment, change the CVD type first and observe before changing the Signal type.

Blended Weight

Setting: Blended Weight Default: 33.3 for Slots 01-03, 0.0 for Slots 04-10

The relative influence of each slot in the blended CVD and Signal calculation. Weights auto-normalize: they are divided by the sum of all active slots' weights, so only the ratios matter. Three slots at 33.3 each produce the same result as three slots at 100 each.

When to change: When you want one timeframe to carry more influence in the composite. Weighting the highest-timeframe slot more heavily biases the blend toward the larger-structure read. Weighting the lowest more heavily makes the blend more responsive to recent pressure.

What equal weights actually imply: With three slots at 33.3 each, each slot has the same influence in the blend — regardless of how often each one updates. A 5m slot on a 5m chart produces a new confirmed value every bar. A 1h slot on the same chart produces a new confirmed value every 12 bars. Under equal weights, both have the same pull on the blended line, even though the 5m slot is updating twelve times as often. This means the blended line is more reactive to the 5m slot's bar-to-bar changes while the 1h slot's step-changes produce visible jumps in the blend whenever the hourly bar closes. That behavior is not wrong, but it is worth understanding. If you want the blend to reflect the higher-timeframe direction more steadily, weight the higher-TF slot more heavily to dampen the short-TF churn.

Zero weight: A slot with weight 0 still computes, still draws its individual line, and still fires its alerts. Only its contribution to the blend is removed. This is the "observer slot" pattern — useful for watching a reference timeframe or ticker without contaminating the composite. But it catches people off guard: if you forget about a weight-zero slot, you may wonder why the blend does not reflect a reading that is visible on the chart. If every enabled slot is weight 0, or no positive-weight slot currently has a valid reading, the blended CVD/Signal stay na until a contributing slot comes online.

If you want a slot to have no influence and no visibility, disable it entirely.

Optional Ticker

Setting: Optional Ticker Default: Empty (uses chart symbol) Location: Per-slot

When set, the slot fetches OHLCV data from the specified symbol instead of the chart's symbol. This enables cross-market CVD stacking — watching BTC's estimated volume pressure while charting ETH, for example, or comparing ES and NQ delta side by side.

What to understand: Normalization makes readings from different symbols look directly comparable — both produce -100 to +100 values. But a +60 on BTCUSDT and a +60 on a mid-cap altcoin represent very different volumes. The numbers are range-equivalent within each symbol's own window, not magnitude-equivalent across symbols. The blend mixes these normalized scores, not raw volume, which means the blended line from a cross-ticker stack is a composite of relative extension, not a composite of actual capital flow.

If you use cross-ticker slots, you need to understand what the blend is averaging and what it is not. See Limitations and Trust Boundaries for the full treatment.

Power User MA parameters

Each slot carries two full sets of advanced MA parameters — one for the CVD moving average and one for the Signal moving average. These activate only when the corresponding MA type needs them.

The key point: If you are using SMA, EMA, or any other standard MA type for both CVD and Signal, none of these parameters do anything. They are inert until their corresponding MA type is selected. Seeing them in the settings panel does not mean you need to configure them — it means the option is available if you ever want it.

Each slot has its own independent set of these parameters. This means you can run ALMA with one set of shape parameters on Slot 01's CVD smoothing and Jurik with different parameters on Slot 03's Signal smoothing. The flexibility is there for traders who need it. If you do not know what these parameters do, you do not need them yet.

Global settings

Overbought and Oversold Levels

Settings: Overbought Level / Oversold Level Defaults: +70 / -70

Horizontal reference lines on the pane. They also serve as thresholds for the blended OB/OS crossing alerts.

These are reference points, not reversal signals. A reading above +70 means the blended oscillator is extended relative to its normalization range. In a trend, it can stay above +70 for hours. The defaults are reasonable starting points — adjust them based on how your instrument and session length typically distribute readings.

Plot Blended CVD/Signal

Default: ON

Shows or hides the blended CVD line, blended Signal line, and the fill between them. Turning this off removes the visual but does not affect blended alerts — those still fire on the computed values.

Blended Line Width

Default: 3. Cosmetic only.

Master Smoothing

Setting: Enable Master Smoothing Default: OFF

When enabled, applies an additional MA pass to the blended CVD and Signal values after the blend is calculated. This is a second smoothing layer on top of the per-slot smoothing.

When to use: When the blended output is still too noisy, typically on lower chart timeframes or when using short per-slot MA lengths. The cost: Additional lag. If per-slot MAs and master smoothing are both long, the blended output will respond very slowly to directional changes. The responsiveness you gain from short per-slot MAs can be negated by heavy master smoothing.

Master MA Type / Length: Default is EMA / 3. Same MA library as per-slot smoothing, with its own set of power-user parameters.

Settings decision sequence

If the settings panel feels overwhelming, here is a reasonable order for configuring from scratch:

1. On Bar Close — decide per slot whether you need stable history (leave on) or faster updates (turn off with awareness of the tradeoff).

2. Slot timeframes — set each enabled slot to a timeframe you actually analyze. Make sure each is at or above your chart timeframe, and that each slot targets a meaningfully different resolution.

3. Window Mode — choose Session if you think in daily/weekly sessions, Rolling if you want continuous accumulation. This changes the oscillator's memory model. You can mix modes across slots.

4. Pressure Sensitivity — start with the default (1.50) on each slot. If the oscillator feels too jumpy on your instrument, lower it. If it feels too flat during genuine moves, raise it. Move in small increments and observe.

5. Wick Weight — start with the default (0.20). Adjust based on whether wick structure is informative on your instrument.

6. Blend weights — start equal. Adjust when you have a reason to emphasize one timeframe over another.

7. Everything else — MA types, lengths, display settings, power-user params. These are refinements. Get the first six right before optimizing the rest.