667 catalogue entries, including 175 roadmap entries

The census contains 667 family-and-name entries across 28 families. It marks 175 as roadmap; the other 492 carry the existing non-roadmap classification. This census is not a fresh execution audit.

667
nodes in the shipped catalogue
Measured
492
nodes validated by an executable check
Measured
175
nodes marked for later work
Measured
379
nodes with user-settable parameters
Measured
667
validated plus roadmap, equal to the total
Measured
0
roadmap nodes with no published limitation
Measured

A node is in one of two states: validated by an executable check, or marked for later work with its limitation written. The two counts add up to the total, and an automated check fails the build if a node falls between them.

The family and the name make the key

A catalogue entry carries a family, a name and a one-line note. The counting key is the pair family/name, so the same name under two families counts as two entries, and a rename produces a new identity.

Of the 667 entries, 379 expose at least one parameter. Parameter blocks define settings and search ranges. Entries without parameters take their behaviour from their implementation and graph inputs.

The recorded strategy in the current visual builder: connected inputs, trading rules and validation nodes.Open full size
The recorded strategy in the current visual builder: connected inputs, trading rules and validation nodes.

6 sections, 28 families, and the custom one

The catalogue groups its families into sections, and the app draws its palette from that grouping. Each line gives the validated count over the family total, on 2026-09-21.

Inputs

Data & Splits10 validated of 19
Macro & Alt-Data4 validated of 4
Timeframe3 validated of 5

Signals

Core Signals24 validated of 36
Breakout9 validated of 9
Mean Reversion10 validated of 12
Trend & Momentum14 validated of 17
Price Action & Patterns11 validated of 12
Volume & Order Flow10 validated of 10

Logic

Entry Filters42 validated of 43
Cost & Slippage17 validated of 26
Position Sizing4 validated of 6
Stops & Exits0 validated of 13
Prop-Firm Rules13 validated of 17

Models

Regime & Labeling7 validated of 18
Machine Learning24 validated of 38

Search & Eval

Optimisation & Portfolio33 validated of 44
Monte Carlo22 validated of 23
Synthetic Paths & Null Models16 validated of 55
Validation & Metrics41 validated of 47
Statistical Tests11 validated of 14
Time-Series & Volatility13 validated of 17

Output

Run & Execution6 validated of 6
Live Display6 validated of 6
Analysis & Plots132 validated of 159
Export & Codegen7 validated of 8

The custom family carries no section and holds 2 entries, described further down.

InputsSignalsLogicModelsSearch & EvalOutputData & Splits10 / 19Macro & Alt-Data4 / 4Timeframe3 / 5Core Signals24 / 36Breakout9 / 9Mean Reversion10 / 12Trend & Momentum14 / 17Price Action & Patterns11 / 12Volume & Order Flow10 / 10Entry Filters42 / 43Cost & Slippage17 / 26Position Sizing4 / 6Stops & Exits0 / 13Prop-Firm Rules13 / 17Regime & Labeling7 / 18Machine Learning24 / 38Optimisation & Portfolio33 / 44Monte Carlo22 / 23Synthetic Paths & Null Models16 / 55Validation & Metrics41 / 47Statistical Tests11 / 14Time-Series & Volatility13 / 17Run & Execution6 / 6Live Display6 / 6Analysis & Plots132 / 159Export & Codegen7 / 8
Figure 1. The shipped catalogue, family by familyFilled bar: nodes validated by an executable check. Outline: the family total. Common scale across all 28 families, so the widths are comparable. Measured 2026-09-21.

The gap between the two bars is where the work is, and it is concentrated: Synthetic Paths & Null Models (16/55), Analysis & Plots (132/159), Machine Learning (24/38), Stops & Exits (0/13) carry 53 % of it, on 4 of the 28 families. The rest of the catalogue sits at or near its own total.

Search the node catalogue by name, here for the Ridge regression model, before adding it to the strategy.Open full size
Search the node catalogue by name, here for the Ridge regression model, before adding it to the strategy.

Validation requires an executable result

A node reaches the validated state when an executable check wires it and obtains two results that differ. Display labels such as live, ported or implemented do not establish that result.

The split is recalculated from the shipped catalogue, so no published count is stored separately. The catalogue distinguishes six cases:

validated        an executable check produces two different results
roadmap          the catalogue states what is still missing
executable       a matching kernel or function resolves; this is only a ceiling
output           the node displays, renders or exports
declared inactive its family is announced in-app as not running
label only       declared implemented, but no matching function resolves

A badge carries what the node still lacks

Each roadmap entry should name what it still lacks. Of the 175 entries marked for later work, 0 carry no reason; that count is recalculated from the shipped catalogue.

A badge with no reason is a promise with no content behind it. The count goes on the page beside the others, at the same rank as the counts that flatter the product.

Custom nodes run Python you wrote yourself

The custom family runs a signal(df) function you author. The call goes through exec with the app's privileges. Withholding import names is a convenience, not a security boundary.

A graph carrying custom code is confirmed in the interface before an autonomous run. The causality gate applies to it as well: run() is re-evaluated on truncated bars, so a signal reading the full series or a negative shift fails the gate.

The shipped catalogue and the enriched one differ

The counts above come from the catalogue inside the desktop build. A second catalogue enriches those entries at runtime and therefore lands on a higher figure.

Both are correct on their own object, and the numbers published here describe the build. What a validated node establishes is bounded by the executable check behind it, and the validation protocol is described under Research.