About the Prediction Basis Standard

What Is the Prediction Basis Standard?

The Prediction Basis Standard (PBAS) is the third Parent Standard within PGA™ — Prediction Governance Architecture .

It governs the informational foundation upon which a prediction is constructed.

Every prediction depends on something: data, observations, historical information, operational signals, expert information, model outputs, market information, simulation outputs, real-time measurements, or combinations of multiple sources.

PBAS establishes how those informational elements become an identifiable and traceable Prediction Basis .

Its purpose is simple but fundamental:

Before a prediction can be governed, it must be possible to establish what that prediction was actually based on.

Why the Standard Exists

Modern predictive systems can process enormous quantities of information while making the informational foundation of a prediction increasingly difficult to understand.

A prediction may appear precise while important questions remain unanswered:

What information actually entered the prediction?

Where did it come from?

Were multiple inputs genuinely independent or derived from the same underlying source?

Was relevant information missing?

Was information deliberately excluded?

Was the information current when the prediction was made?

Did the predictor possess information unavailable to others?

Was the prediction based partly on simulation, AI-generated information, market signals, expert knowledge, or direct Operational Reality?

Without governance of these relationships, the final prediction may be visible while the informational foundation beneath it remains opaque.

PBAS establishes that foundation as a governed object.

What PBAS Governs

PBAS governs the relationship between a prediction and the information upon which it is based.

This includes governance of:

Prediction Inputs; information sources; source relationships; source dependencies; signal relevance; information inclusion and exclusion; unavailable information; derived information;

information redundancy; conflicting information; informational sufficiency; Predictor–Reality Proximity; information advantage; relevant outcome-knowledge relationships; Prediction Basis change; Prediction Basis Decay; historical Prediction Basis continuity; Prediction Basis traceability.

The result is a Governed Prediction Basis that can remain associated with the prediction throughout its lifecycle.

Prediction Basis Is More Than Data

Prediction Basis should not be understood merely as a dataset.

A prediction may depend upon combinations of:

Data + Observations + Signals + Expert Information + Model Outputs + Market Information + Simulation-Derived Information + Contextual Information

Different predictive systems may therefore possess fundamentally different Prediction Bases even when predicting the same outcome.

PBAS makes those differences governable.

Prediction Basis ≠ Evidence Governance

This is one of the most important boundaries of PBAS.

Prediction Basis ≠ Evidence Governance

PBAS determines whether and how information forms the informational basis of a prediction.

It does not independently determine whether that information constitutes valid, admissible, authenticated, or sufficient evidence.

Those questions remain within the appropriate evidence and validation governance architectures, including ADIT® and VALIDOS™ where applicable.

Information may therefore be relevant to a prediction without PGA™ claiming that the information has been independently validated as evidence.

Likewise:

Valid Evidence ≠ Relevant Prediction Basis by Default

The two governance questions are different.

Source Relationships Matter

Knowing which information was used is not always enough.

It may also be necessary to understand where that information originated.

Ten different reports may appear to provide ten independent signals while all ten ultimately derive from the same original source.

Similarly, several AI systems may appear to provide independent predictions while relying upon materially overlapping information.

PBAS therefore preserves material source relationships.

Multiple Inputs ≠ Independent Sources

This prevents apparent information volume from being mistaken for informational independence.

Relevance Matters More Than Volume

Predictive systems increasingly have access to enormous quantities of information.

More information does not automatically produce a better governed Prediction Basis.

PBAS therefore distinguishes information availability from predictive relevance.

Available Information ≠ Governed Prediction Input

More Information ≠ Better Prediction Basis

Signal Presence ≠ Predictive Relevance

A governed Prediction Basis should contain information because it has a legitimate relationship to the prediction — not merely because the information is available.

Predictor–Reality Proximity

PBAS also addresses an important characteristic of predictive sources: their proximity to the reality being predicted.

Predictor–Reality Proximity describes the governed relationship between a predictor or predictive source and the information, actors, processes, authorities, systems, decisions, or operational conditions from which the predicted outcome may emerge or by which that outcome may be materially determined or influenced.

A predictor may operate using only public information.

Another may possess privileged operational information.

Another may work directly with decision-makers.

Another may already know information strongly determining the outcome.

Another may possess some ability to influence the outcome itself.

These conditions are not equivalent.

A useful governance continuum is:

Public-Information Predictor → Information-Advantaged Predictor → Decision-Proximate Predictor → Outcome- Informed Predictor → Outcome-Influencing Predictor → Outcome-Controlling Predictor

PBAS does not assume wrongdoing merely because a predictor is close to Operational Reality.

High Predictor–Reality Proximity ≠ Misconduct

Instead, it makes the informational relationship visible.

Information Advantage ≠ Predictive Capability

Predictive performance can be misunderstood when information access is ignored.

A predictor may repeatedly outperform other predictors because it has superior access to information rather than because it possesses a superior prediction methodology.

PGA™ therefore preserves an essential distinction:

Information Advantage ≠ Predictive Capability

And where the outcome is already materially known:

Outcome Knowledge ≠ Predictive Capability

In simple terms:

Knowing an outcome before others is not the same as predicting it better than others.

PBAS preserves this informational context so later predictive-performance governance can interpret results more meaningfully.

The Prediction Basis Is Time-Bound

Prediction Basis must be associated with the information available when the prediction was actually formed.

Information discovered afterward must not silently become part of the historical Prediction Basis.

Later Information ≠ Original Prediction Basis

This becomes especially important when evaluating whether a prediction was impressive, poorly founded, or methodologically justified.

Historical evaluation must consider what was actually known at the time.

Prediction Basis Can Change

The informational foundation of a prediction may change after the prediction has been formed.

New information may appear.

Existing information may become obsolete.

Signals may weaken.

Operational conditions may change.

Previously unavailable information may become available.

Contradictory information may emerge.

PBAS therefore recognizes Prediction Basis Decay .

Valid Initial Prediction Basis ≠ Valid Continuing Prediction Basis

PBAS makes the change governable.

Later PGA™ standards determine whether that change requires review, revision, supersession, or withdrawal of the prediction.

Position Within PGA™

PBAS occupies the third governance space within the Prediction Lifecycle.

The complete PGA™ Governance Flow is:

Prediction Object & Purpose → Prediction Methodology → Prediction Basis → Prediction Assumptions → Prediction Formation → Probability & Uncertainty → Prediction Feasibility → Collective Prediction where applicable → Communication & Reliance → Monitoring & Revision → Outcome & Performance → Prediction Record & Continuity

PBAS operates at:

Prediction Object & Purpose → Prediction Methodology → → Prediction Basis → Prediction Assumptions

It receives a sufficiently governed predictive frame and methodology and establishes the informational foundation required before predictive assumptions are governed.

How PBAS Is Operationalized

PBAS is operationalized through five Core Modules:

PIIM — Prediction Input Identification Module Identifies the information that materially enters Prediction Basis.

PSMM — Predictive Source Mapping Module Maps source origin, source relationships, dependencies, and materially relevant Predictor–Reality Proximity.

SRVM — Signal Relevance Module Determines whether identified information and signals are relevant to the governed prediction.

ISUM — Input Sufficiency Module Determines whether the informational basis is sufficiently complete for the methodology and predictive frame.

PBTRM — Prediction Basis Traceability Module Preserves source lineage, historical basis state, change, and continuity.

The internal progression is:

Prediction Input Identification → Predictive Source Mapping → Signal Relevance → Input Sufficiency → Prediction Basis Traceability

What PBAS Does Not Govern

PBAS does not determine:

whether a prediction is correct; whether evidence is legally admissible; whether evidence has been independently validated; Operational Reality itself; predictive assumptions; prediction formation; probability or uncertainty representation; prediction feasibility; collective prediction legitimacy; permitted reliance; accountability; predictive performance; final validation status.

These remain governed by subsequent PGA™ Parent Standards or other applicable OOF® architectures.

This boundary prevents Prediction Basis governance from expanding into general evidence, validation, reality, accountability, or prediction-performance governance.

Why PBAS Matters

Organizations increasingly evaluate predictions by looking at the output:

72% probability.

High confidence.

Failure expected within 30 days.

Market likely to decline.

Candidate likely to win.

But the output alone reveals very little about the informational foundation beneath it.

Two predictions producing the same probability may be based on radically different information.

One may use broad public information.

Another may use proprietary operational data.

Another may rely heavily on one expert.

Another may aggregate dependent sources.

Another may incorporate information from someone directly involved in determining the outcome.

PBAS makes these differences visible.

It allows organizations to move from:

“What does the prediction say?”

toward:

“What information legitimately supports the prediction?”

Use Case 1 — Ten Sources That Are Really One Source

A financial forecasting system receives information from ten different research reports.

The system appears to have a diverse Prediction Basis because ten independent documents are being processed.

Closer examination reveals that eight reports ultimately rely upon the same original industry dataset and several reproduce the same underlying market estimate.

The prediction therefore appears to possess significantly more independent informational support than actually exists.

The critical governance question is:

Can a Prediction Basis legitimately be represented as diversified when multiple apparently independent inputs originate from the same underlying informational source?

PBAS requires material source relationships and dependencies to remain identifiable.

The result is not an automatic rejection of the prediction.

It is a more accurate representation of its informational foundation.

Multiple Inputs ≠ Independent Sources

Use Case 2 — The Predictor Who Already Knows More Than the Market

A prediction market asks participants whether a major commercial transaction will be completed.

Most participants rely upon public information.

One participant has close operational relationships with individuals involved in the transaction and receives information that materially increases their understanding of the likely outcome.

The participant repeatedly makes highly accurate predictions and appears to possess exceptional forecasting ability.

The critical governance question is:

Does the observed predictive performance demonstrate superior predictive capability, or does it materially reflect superior proximity to the information and decision processes from which the outcome emerges?

PBAS does not automatically classify the participant's behavior as improper.

Instead, it requires the informational relationship to become governable through Predictor–Reality Proximity.

Information Advantage ≠ Predictive Capability

Outcome Knowledge ≠ Predictive Capability

This distinction later allows PGA™ to evaluate predictive performance without confusing informational advantage with forecasting skill.

Canonical Boundary

PBAS governs what information forms Prediction Basis, where that information comes from, how it relates to the prediction, whether material source relationships and informational conditions remain identifiable, whether the basis is sufficiently established, and how that basis remains traceable through time.

It does not determine whether the information constitutes valid evidence, whether Operational Reality has been established, or whether the resulting prediction is correct.

PBAS governs the informational foundation of the prediction — not the truth of the predicted outcome.

Canonical Closing Statement

A prediction cannot be understood only by examining what it predicts. Its informational foundation must also remain visible.

PBAS makes it possible to know what the prediction knew when the prediction was made.

About Prediction Basis Standard (PBAS)
Prediction Information Identification Module (PIIM)
Prediction Signal Mapping Module (PSMM)
Source Relevance & Validity Module (SRVM)
Information Sufficiency Module (ISUM)
Prediction Basis Traceability Module (PBTRM)

←