IQ Wizard: Manual Signal Builder
The Manual Signal Builder lets you define a signal by chaining together indicator/labeling functions yourself, step by step, instead of training an AI model against a target. The exact same builder, opened in a different mode, is also how you define a Custom Indicator — a reusable indicator column you compute yourself and make available as an input feature elsewhere in the wizard.
Opening the builder
From an Entries/Exits row, choosing Manual as the signal type opens the builder in Manual Signal mode.
From an AI or AI Regression signal's indicator picker, Add Custom Indicator opens the same builder in Custom Indicator mode, scoped to a specific Base/Helper timeframe.
From the Templates page, View & Edit Configuration on a Manual or Custom Indicator template reopens the builder pre-filled with that template's steps.
The two modes share the exact same step-building mechanics described below; they differ only in what happens once you're done:
A Manual Signal ends with a Final Signal Trigger step that picks which output label actually fires the signal.
A Custom Indicator has no trigger — its last step's raw output (numeric or categorical) is the indicator, and you just give it a name.
Step 1 · Base Indicators
Every pipeline starts with one or more Base Indicators — numeric indicators only, browsed by category (or via the search box). You can add the same indicator more than once (e.g. an EMA with a window length of 50 and a separate EMA with a window length of 100) so they can later be compared against each other. Each base indicator you add becomes its own step, with its own parameter fields (window length, source column, etc.) editable directly on its step card.
Adding modifiers
Once you have at least one Base Indicator, use + Add Modifier to layer a labeling or derived-numeric function on top of your existing steps. The picker only shows functions whose required input type (numeric or categorical) matches at least one of your current steps' output — so, for example, a function that needs a categorical input won't be offered until you already have a categorical-output step to feed it.
Popular
The most commonly reached-for modifiers are pinned to a Popular section at the top of the picker, ahead of the full category-grouped list, so you don't have to search for them every time:
Apply Expression, Cross, Sort — general-purpose staples.
Oscillator Group — RSI/Stochastic-style overbought/oversold zone reads.
Confirmed Cross — a cross that must hold for a configurable number of bars before counting, filtering out one-bar false crosses.
Level Interaction — support/resistance-style touch/break/approach detection against a level.
Confluence Score — a weighted score combining several existing categorical signals into a single bull/bear score and label.
The rest of the available modifiers are grouped by category below the Popular section (QuantiX Labeling Functions, QuantiX Numeric Functions, and so on) — see QuantiX Labeling Functions and QuantiX Numeric Functions for what each one does.
Wiring inputs between steps
Every modifier has one Input picker per parameter that needs to reference another step's output (e.g. Cross needs a "First Feature" and a "Second Feature", each pointing at an earlier step). The picker only offers steps whose output type is compatible. If the step you're referencing produced more than one output column at once (some functions do — see below), the picker lets you choose exactly which one.
Multi-output steps
Most functions produce a single output column, but a few produce several at once — for example, Moving Average Group produces the moving average itself plus its distance from price, cross state, and several other related columns in one step; Divergence Group and several of the new Advanced Quant wrapper functions (see below) work the same way. Adding one of these as a step makes every one of its outputs individually selectable in later steps' Input pickers, labeled with the step name plus the specific output (e.g. "Step 2 · Oscillator Group · zone").
Removing a step
Removing a step also clears it from any later step's Input picker that was referencing it — you'll need to pick a new input for anything that pointed at the removed step.
Final Signal Trigger (Manual Signal mode only)
Once your last step produces a categorical output, the Final Signal Trigger card shows the predicted output labels for that step and lets you pick which one fires the signal. Predicted labels are worked out automatically wherever possible — from a step's own configured label list (e.g. the names you typed into a Streak Label or Distance Label step), from a fixed, verified vocabulary for functions whose output values are hardcoded (e.g. Cross's "cross over"/"cross under"/"not cross", or the new Oscillator Group's "overbought"/"neutral"/"oversold"), or by tracing back through the chain for functions whose output is derived from an earlier step's own labels (History, Memory, Majority, Labeling Shift).
If a step's real output vocabulary can't be predicted this way (this happens for a small number of multi-output functions whose first output happens to be numeric even though the step is flagged categorical overall — Moving Average Group, Band Context, and Confluence Score are examples), a Custom label (advanced) text field lets you type the exact value to match instead.
Custom Indicators
In Custom Indicator mode, there is no trigger step — instead, once you have at least one step, you give the pipeline a name, and the last step's raw output (numeric or categorical) becomes a new indicator column under that name, available for the specific Base/Helper timeframe you opened the builder from. Custom Indicators can then be picked alongside the predefined indicators anywhere the wizard lets you choose input features for an AI/AI Regression signal.
Saving as a Template
Any pipeline — Manual Signal or Custom Indicator — can be saved as a reusable Template from within the builder, so you don't have to rebuild the same chain of steps from scratch next time. See IQ Wizard Templates for how templates are managed, edited, and reused.