PTC Confidence Framework

PTC grades reliability differently for different kinds of published content. This page describes each system and how they relate to one another. It does not restate the model-specific mechanics documented on the Active Trader, Funded Trader and Monthly Payout model pages, or the shared rules in the Estimate Methodology.

What confidence means at PTC

Confidence is a grade PTC attaches to a modelled estimate, answering one question: how strongly does the available evidence support this particular figure? It is not a single grade applied to everything PTC publishes.

PTC keeps four questions separate rather than folding them into one score: what kind of figure is this (provenance), how strong is the underlying evidence (evidence strength), how confident is PTC in a modelled figure specifically (estimate confidence), and should this be shown publicly (publication policy). A directly recorded fact, a ranking position, a research finding and a modelled estimate are each graded for reliability in their own way, described in the sections below.

Provenance is not confidence

Every PTC figure carries a provenance label stating what kind of figure it is. Provenance says nothing about how confident PTC is in it.

Observed: Directly measured or recorded data.
PTC Estimate: Modelled values based on a disclosed PTC methodology.
Derived: Calculated from PTC datasets, rankings or published snapshots.
Research Finding: Evidence-backed findings from PTC research.

An Observed figure and a PTC Estimate figure can each be strong or weak evidence in their own domain — provenance classifies the figure, it does not grade it. Provenance and confidence are recorded and displayed separately, and a provenance label should never be read as a confidence grade.

Estimate confidence

Each of PTC's three estimate models assigns every firm-level estimate a confidence score, reported publicly as High, Medium or Low.

The score reflects factors such as how much of the estimate rests on the firm's own disclosed evidence versus a calibrated market relationship, the quality and freshness of that evidence, and — for a model built on top of another PTC estimate — the confidence already carried by that upstream figure.
PTC publishes the resulting band, not the internal weights or numeric thresholds behind the score.
Every published estimate also states its basis in plain language — an evidence-led estimate, a modelled estimate, or, for Funded Traders, a scenario estimate — so a reader can see whether a figure rests mainly on firm disclosure or on a calibrated relationship, independent of its confidence band.

Evidence strength

Confidence in a modelled estimate is a different system from the reliability of a single piece of research evidence.

PTC records sources against a quality framework — Official, High quality, Secondary and Community — described in the Research Methodology. A finding drawn from firm documentation or a regulator carries more weight than one drawn from a single community report, and PTC treats a claim as corroborated only where more than one independent source supports it. This source-quality framework feeds research findings and firm dossiers; it is a separate system from the High/Medium/Low confidence band attached to a modelled estimate, even though both describe evidence strength within their own domain.

Confidence differs by model

Confidence is not calculated the same way twice. Each model defines its own confidence factors and its own conditions for reaching each band, because the evidence available differs by metric.

Active Trader Model™

Disclosure-associated in practice

Confidence weighs signal coverage, disclosure evidence and validation against other observed data. The firms that reach High confidence today are all anchored on a firm's own disclosed trader count; a purely calibrated estimate has not, in current production data, reached High on its own.

Funded Trader Model™

High requires current-stock evidence

Confidence carries the upstream Active Trader confidence forward, alongside the strength of the firm's own funded evidence. High confidence structurally requires a firm's own disclosed current funded-person or funded-account count; a calibrated scenario estimate alone cannot reach High, however strong the upstream Active Trader estimate is.

Monthly Payout Model™

Structural caps on modelled values

High confidence structurally requires current firm-specific payout evidence; a purely modelled payout figure cannot reach High regardless of score, and modelled Futures estimates are held to a stricter ceiling than modelled Forex estimates because the Futures calibration rests on thinner evidence.

Because the conditions differ, the same band name means "the evidence clears this model's own bar for that band," not "the evidence looks identical across models." A Medium in one model and a Medium in another are not built from the same ingredients.

Confidence and uncertainty ranges

Every estimate is published with a low–central–high range alongside its confidence band.

The range expresses model uncertainty under the stated methodology — how wide PTC considers the plausible values to be, given the evidence and calibration behind that estimate. It is not a statistical confidence interval, and PTC does not publish a sampling-based margin of error for any estimate.
Confidence and range width are related but not interchangeable: a Low-confidence estimate typically carries a wide range, and firm-specific evidence typically narrows it — but a narrow range does not by itself imply High confidence, and a wide range does not by itself imply Low confidence.
Aggregate totals are published as a central sum across modelled firms. PTC does not publish an aggregate-level range, because summing firm-level ranges would assume the firms' errors are independent, which the models do not claim.

Publication is a separate decision

Confidence informs, but does not by itself decide, whether a figure is shown, shown range-first, or withheld at firm level.

Publication gate: Each model applies its own gate on top of confidence — for example whether the estimate clears a minimum materiality threshold, and how wide its range is relative to its central value.
Low ≠ withheld: A Low-confidence estimate is not automatically withheld. Several PTC estimates publish Low-confidence figures range-first, with the central value marked as approximate rather than removed.
Withheld ≠ zero: A firm can be modelled with reasonable confidence and still be withheld at the firm level if it falls short of a model's publication threshold. A withheld figure is not a claim that the underlying quantity is small, and it still contributes to the published aggregate for its market.

Rankings and research

Not everything PTC publishes carries a confidence band.

Rankings & indexes

Coverage and eligibility, not a grade

PTC rankings and indexes are comparative scores, not estimates, and are not graded High/Medium/Low. Reliability is expressed instead through signal coverage and eligibility rules that exclude firms without enough signal coverage from an index entirely, rather than scoring them on thin evidence — see the Ranking Methodology.

Research & dossiers

Source quality and editorial review

Research findings and firm dossier content pass through source-quality assessment and editorial review — accepted, edited, held or rejected — rather than a numeric confidence score. Where evidence is thin or conflicting, PTC publishes at a lower precision, discloses the conflict, or withholds the claim — see the Research Methodology.

Limitations and interpretation

Confidence describes the strength of the evidence behind a modelled estimate; it is not a probability, a percentage or a guarantee of accuracy.
A High band does not mean a figure is certain, and a Low band does not mean a figure is unreliable in an absolute sense — only that it rests more heavily on calibrated assumptions than on firm-specific disclosure.
Confidence bands are not directly comparable across models, because each model's conditions for each band differ.
Not all published PTC content carries a confidence band: provenance, source quality and editorial review play that role for observed data, rankings and research findings respectively.
Confidence reflects the evidence available at the model run behind a published baseline, not a live, continuously updated assessment.

Governance and versioning

Confidence scores and bands are produced by a model run and stored with the estimate they describe, tied to that run's methodology version.

A later model run can change a firm's confidence band as new evidence becomes available or as the model version changes. The prior value is retained as history rather than silently overwritten, in keeping with the baseline and versioning rules described in the Estimate Methodology.