Limitations & Trust Boundaries

This is not a disclaimer page. It is a working frame for what to trust in Axiom DC CTX, what to verify before you rely on it, and what never to assume. Everything below is something the tool does (or deliberately does not do), stated honestly enough that you can decide how to build your own process around it.

The habit worth forming: every time the tool's output is about to influence a decision, ask which trust boundary the decision sits on. Some boundaries are rock-solid. Some require verification. Some are where the tool hands the judgment back to you and says "this is where your reading starts."

What this tool is reliably good at

Holding several Donchian ranges across timeframes and, optionally, across symbols, in one frame, under a configurable blend, with a per-slot repaint posture you chose on purpose. That sentence is long because every clause in it earns its place.

• Range reads are exact. When On Bar Close? = ON, the upper and lower lines are the exact highest-high and lowest-low of the slot's Length: bars at the slot's timeframe. Nothing is approximated. Nothing is estimated. The numbers are what the market did.

• The blend is a declared, auditable composite. The blended upper, lower, and basis are exact weighted averages of the contributing slots. You can compute them on paper from the visible slot values and check the chart against your arithmetic.

• Repaint posture is a switch you own. ON is confirmed-history; OFF is live/current and repaint-prone; the distinction is stable and verifiable (see MTF & Repainting). Do not give OFF history the trust contract that belongs to ON.

• Alerts fire at chart-bar close on computed conditions. Not intrabar. Not on a prediction. On a specific condition you can read off the source of truth if you want to.

These are the boundaries inside which the tool is straightforwardly reliable.

What this tool cannot tell you

What is going to happen next. This is not a hedge — it is the most important thing to name plainly. The indicator reports relationships (where the basis is relative to itself some number of bars ago; where the chart close is relative to the basis; how enabled slots are oriented) at specific points in time. None of those relationships claim anything about the future, and no combination of alerts on them is a signal that the next bar will go a specific way.

If you want predictive behavior, you build a strategy on top of these reads. The tool is the instrument. Your methodology is whatever you layer on top.

Blend-honesty boundary

The blended channel is useful and the blended channel is also easy to over-trust. A short list of what the blend does not do:

• The blend does not remove disagreement. It averages it. When two slots say different things, the blended line sits between them, weighted. A middling blended value can reflect genuine consensus in the middle or disagreement averaged down — those are different situations that look identical on the blended line. Open the individual slots when the blended picture matters.

• Hidden-plot slots vote in full. A slot with Hide DC NN Plot ON still contributes to the blend's weighted average and to the blended basis-trend majority vote. "Out of sight" is not "out of the math." If you see the blended channel move while no visible slot appears to have moved, check for hidden slots with non-zero weights.

• Zero-weight slots do not steer the blend. A slot with Blended Weight: 0 keeps its lines on the chart but does not contribute to the blended sum. It is visible, it is enabled, but it is a silent member of the composite.

• The blended basis-trend alert uses a weight-majority vote with ties resolving to uptrend. This is a deterministic rule, not a regime call. A 2-slot stack with one up and one down at equal weights fires Blended DC Basis Is Uptrend purely by tie-break convention. Know the rule before acting on the alert.

• Shipped weights 40 / 35 / 25 are a load posture, not a ranking. They are not "optimized." No study produced them. They bias the blend toward faster timeframes because that is a coherent composite for a first-load reader. Your weights should reflect your trust distribution across the timeframes you are reading, not the shipped defaults.

Cross-ticker-honesty boundary

The Optional Ticker: feature is a visibility aid. It lets you draw a related market's range envelope on your chart in the chart's own price space by rescaling the alternate symbol's channel with a single close-ratio at the slot's timeframe. Several things are worth knowing:

• The scaling is not predictive. The alternate symbol is not being treated as a leading indicator. If a QQQ-scaled channel on a SPY chart sits above the SPY range, it does not mean SPY is about to "catch up." It means QQQ's range at that timeframe, expressed in SPY prices by the close ratio, currently sits there.

• Session mismatches warp the scaling. If the alternate symbol and the chart symbol have non-overlapping sessions (for example, an overnight session one has and the other does not), the scaling ratio computed during the non-overlapping window can be stale or asynchronous. The scaled channel during those windows should not be read as authoritative.

• Volatility-regime mismatches are not corrected. The ratio is a single-point scaling, not a volatility-aware transform. Two symbols with very different realized volatility regimes will produce scaled channels whose widths look sensible in price space but are not comparable as volatility envelopes.

• The ratio inherits the slot's On Bar Close? posture. A cross-ticker slot under OFF scales against live HTF closes on both symbols and can redraw; its old/reloaded bars are not a faithful live record. A cross-ticker slot under ON scales against confirmed closes. For cross-ticker usage, ON is almost always the posture you want — see MTF & Repainting.

The honest reading of a cross-ticker slot: this is what the other symbol's range looks like, if you squeeze it onto the chart symbol's price axis by this much. That is a piece of context. It is not an arrow.

Alert-honesty boundary

Alerts are narrow by design. See Alerts for the full inventory. A few points belong on this page because they are trust boundaries rather than feature descriptions:

• Alerts fire at chart-bar close, not HTF-bar close. A chart bar can close while the slot's HTF bar is still forming. Under On Bar Close? = OFF, the alert may have evaluated against a live HTF read. The alert message by itself does not tell you whether the HTF was confirmed — your slot's posture does.

• Alignment alerts require every enabled slot. A stack where three of four slots agree does not fire All DC Slot Bases Uptrend. This is not a bug — the alert is deliberately strict. Read silence as information.

• No alert confirms intent or outcome. A DC 02 Basis Is Uptrend alert confirms that the slot's basis is currently at or above where it was Basis Trend Length: bars ago, on the chart bar that just closed. It does not confirm that you should trade. It does not confirm that the basis will continue rising. It does not confirm that price will follow the basis.

Configuration-honesty boundary

The shipped configuration — three slots on 5/15/60 with weights 40/35/25, every slot ON, length 20, basis MA length 1, basis trend length 3 — is a posture, not a recommendation. Several implications follow:

• If you rely on the shipped defaults as if someone somewhere had proven them to be correct for your instrument, your timeframe, and your method, you are reading a configuration you did not actually choose. The manual is explicit on this because it matters: defaults are there so the tool loads cleanly with something sensible on the screen, not to save you from choosing.

• Adaptability is the point. Ten slots, per-slot timeframe, per-slot symbol, per-slot MA family on the basis, per-slot repaint posture, per-slot weight — this tool is adaptable and customizable. The amount of configurability is not there to overwhelm you. It is there because different readers need different layers, and the tool refuses to pretend one layering is right for everyone.

• The smoothing asymmetry is non-negotiable. Basis MA Length: and Type: smooth the midpoint only. The outer bounds are raw. You cannot configure your way out of this; it is how the tool is designed. The right response is to pick Length: deliberately for the outer envelope, and to pick basis smoothing deliberately for the midline — two different knobs for two different jobs.

Behaviors that are documented, not limits

Worth separating, because it is easy to file "things that surprised me" under "product limits" when some of them are doing exactly what they are supposed to do.

• The chart goes blank with a red error banner when a slot's timeframe is below the chart's. That is the timeframe guard working. Raise the slot. (See Troubleshooting.)

• A hidden slot steers the blend. That is by design. Hide is cosmetic.

• A zero-weight slot draws lines but does not steer the blend. That is by design. It is how you keep a slot visible as context without letting it vote.

• A slot under OFF redraws between HTF closes. That is OFF. Its historical/reloaded bars are not proof of what was available live. Read MTF & Repainting before accepting or rejecting that behavior.

• Alignment alerts silent while most slots agree. Alignment requires every enabled slot. Partial agreement is not alignment.

• Basis Trend Length: 1 produces flickering alerts. That is what "current basis at/above or below one bar ago" produces. Lengthen the trend window if the flicker is not informative.

Behaviors that are genuine limits

Three things the tool cannot do, stated plainly:

• It cannot read a slot below the chart timeframe. The runtime error is a hard stop.

• It cannot correct for session mismatches on cross-ticker slots. If the two symbols' sessions do not align, the scaled channel will not know.

• It cannot tell you that your weights are wrong. The weights are yours. The tool executes your chosen composite honestly and makes no claim about whether the composite you chose is appropriate for the question you are asking. That judgment is outside the tool.

A short list of things never to assume

• Never assume the outer channel is smoothed by the basis MA settings. It is not.

• Never assume a hidden slot is out of the blend. It is not.

• Never assume alignment alerts exclude hidden slots. They do not.

• Never assume a cross-ticker channel predicts the chart symbol. It does not.

• Never assume ON and OFF are "slow" and "fast" versions of the same posture. They are different contracts.

• Never assume shipped defaults are ranked or optimized. They are a load posture.

These six assumptions are the ones that cost people time. Keep them on the side of your mental model and you will stop making them without having to think.

Using the trust-boundary frame in practice

A habit worth forming when you are looking at the chart under any kind of pressure: before you act on something the tool is showing you, silently locate the trust boundary that decision is crossing. Three categories are enough to sort most situations.

• Reliable layer. Exact readings, exact arithmetic, exact alert conditions. If a decision only relies on this layer, it is on firm ground. Example: "the 15-minute high is 101.50 and price is at 101.00" — that is literal; the tool reports it with precision.

• Verify-first layer. Behaviors that are correct given the configuration, but where the configuration might not be what you remember. Hidden slots steering the blend, weights pulling the composite somewhere unexpected, a slot under OFF that you forgot. Open the inputs dialog before acting.

• Your judgment, not the tool's. Whether the composite you built is appropriate for the question you are asking, whether the 240-minute context is relevant to a 5-minute decision, whether a cross-ticker context slot is telling you something about the chart symbol or only about itself. No amount of inspecting the tool answers these; the tool hands them back to you on purpose.

When the distinction between those three categories is clear in your head, the tool stops feeling mysterious and starts feeling like an instrument with known accuracies in known dimensions — which is what it is.