Limitations and Trust Boundaries

This page exists so the indicator stays useful without being asked to do a job it never claimed. Axiom CVD Osc Lite can organize multi-timeframe participation context very well. It can also look more certain than it really is if you forget what the stack is built from. The point of this page is not to scare you away from the tool. It is to keep the tool in the right role. That is especially important once the pane starts looking tidy. Clean visuals are helpful here, but they can also hide how many earlier choices are still carrying the final read.

What you can trust

You can trust these statements because they are observable in the tool:

• the indicator can run up to 3 configurable slots
• each enabled slot must respect the chart-timeframe, lower-timeframe, and window rules
• each slot is normalized into a bounded -100 to +100 oscillator
• the blended pair only summarizes enabled slots whose Blended Weight: is not 0
• the alert conditions only go true on confirmed chart bars
• the whole stack shares one higher-timeframe confirmation choice through On Bar Close?

Those are stable structural truths. They are not the same thing as market truth.

What you still need to verify

These are the questions the indicator cannot answer for you automatically:

• Is this slot stack actually fit for this market and timeframe?
• Does confirmed mode or live-forming mode match the way I intend to use the output?
• Are my chosen weights helping the summary, or just making it look more decisive?
• Does the alternate-ticker slot add useful context, or just more noise?
• Am I reading normalized extremes as context, or as orders?

The better you get at answering those questions, the more honest the tool becomes in your process. If you notice yourself answering them with guesswork, move back into verification before you add more complexity.

The first trust boundary: this is an estimate

This script estimates participation from OHLCV structure. It uses:

• lower-timeframe sampling when available
• bar body and close placement
• wick behavior
• a carry mechanism when bars are neutral

What it does not have is exchange-side trade classification. So the output can be useful without becoming true footprint or bid/ask delta. That distinction matters because the pane can look very convincing at moments of strong agreement. If you forget the estimate boundary, the cleanest-looking parts of the chart can start borrowing authority they did not earn.

The second trust boundary: higher-timeframe behavior depends on one shared mode

On Bar Close? is the stack's global trust switch.

When it is on:

• each slot uses the last closed higher-timeframe reading
• history and live behavior are easier to compare honestly
• you give up some early movement

When it is off:

• each slot can update from the still-forming higher-timeframe bar
• the stack can react earlier
• the live read can move before that higher-timeframe bar closes

There is nothing morally wrong with live-forming mode. The mistake is acting like that trust tradeoff does not exist anymore once the pane looks normal again.

The third trust boundary: normalization helps comparison, but it hides scale

Turning active-window CVD into a bounded oscillator is one of the reasons this tool is easier to read than several raw cumulative lines. That gain is real. It comes with a cost.

Normalization helps you compare shape and position across slots. It also means:

• raw cumulative magnitude is no longer the thing you are looking at
• threshold lines are chosen references inside a normalized space
• two slots can look comparable even though they came from very different contexts underneath

So use the bounded scale for comparison, not for mythology.

The fourth trust boundary: the blend compresses earlier decisions

The blended CVD and blended Signal pair are useful because they reduce chart clutter. They are risky when you forget they inherit:

• your chosen slot timeframes
• your chosen window model
• your chosen smoothing
• your chosen participation tuning
• your chosen weights
• your chosen symbol mix

The blend is not independent evidence. It is a weighted summary of configured slot outputs. That is why one of the healthiest habits in this indicator is being able to name which slots are shaping the blend before you trust the blend at all.

Limits that belong near the front of your process

Runtime geometry limits

An enabled slot will fail if:

• its timeframe is below the chart timeframe
• its Lower TF Precision: is not lower than the slot timeframe
• its Window: is smaller than the slot timeframe

Those are hard limits. The tool is not being fussy. It is protecting the geometry it actually needs.

Fallback visibility limit

The script can fall back from lower-timeframe array sampling to a slot-bar estimate when lower-timeframe arrays are unavailable. That handoff is not marked with a dedicated on-chart warning. So if you are working near the edge of available intrabar detail, the manual matters more, not less.

Hidden-slot logic limit

A hidden slot can still matter. Visual absence is not logical absence here.

Zero-weight logic limit

A zero-weight slot leaves the blend, but it can still keep its own line, its own slot alerts, and its role in all-slot alignment.

Blend disappearance limit

If every effective blend contributor has zero weight, the blended lines disappear because the summary has nothing to summarize. That is not a mystery state. It is a direct result of your weighting choices.

What not to assume

• this is true traded-side delta
• stronger agreement means objective truth
• smoother is automatically safer
• overbought and oversold are complete trade instructions
• hiding a slot removed its influence
• turning the blend off visually removed blended alerts
• confirmed chart-bar alerts erase live-forming higher-timeframe risk

If you keep those assumptions out, the tool stays much easier to use honestly.

A practical verification routine

Run these checks whenever the stack starts feeling more certain than you can explain:

1. Confirm every enabled slot is legal on the current chart.
2. Toggle On Bar Close? and watch one unfinished higher-timeframe bar.
3. Set one slot's Blended Weight: to 0 and confirm what changed and what did not.
4. Hide one enabled slot and confirm the logic still carries it.
5. Compare one Session slot with one Rolling slot so the window difference stays real in your head.
6. If using an alternate ticker, compare that slot against the source market on a separate chart.

That routine is usually more valuable than adding one more setting tweak.

What a healthy use of this indicator sounds like

"This stack gives me a structured estimate of participation across the contexts I chose. I still have to verify the mode, the weighting, and the fit." That sentence is not modesty theater. It is the posture that keeps the tool useful instead of turning it into rented conviction.

Visual placeholder: Trust-boundary diagram showing four layers of caution: estimated participation, shared higher-timeframe confirmation mode, bounded normalization, and weighted blend compression.