A forecast is a probabilistic assessment of an uncertain future event—not a statement of fact or a promise of outcome.
Aegis Decision Systems is designed to measure uncertainty, not to conceal it.
These disclosures apply to Aegis forecasts, probability estimates, scenarios, target ranges, forecasting records, and related performance presentations.
Forecast Date and Information Cutoff
Every formal Aegis forecast should identify:
the forecast date;
the event or variable being forecast;
the forecast horizon or resolution date;
the information cutoff;
the applicable resolution or scoring rule; and
the official forecast, and — where methodologically justified — a numerical probability or confidence measure.
A forecast reflects the evidence and assumptions available at the stated cutoff. Information becoming available afterward was not part of the original forecast unless expressly identified as having been available before the cutoff.
Meaning of Probability
An assigned probability expresses Aegis’s assessed likelihood of an outcome under the stated definitions, evidence, assumptions, and time horizon.
A probability of 70% does not mean that the event is certain. It means that the event is assessed as more likely than not while retaining a meaningful possibility of failure.
Even properly calibrated forecasts with high probabilities will sometimes be wrong. Likewise, low-probability events will sometimes occur.
Probabilities may be rounded and should not be interpreted as greater precision than the evidence supports.
Scenarios Are Not Necessarily Predictions
A scenario describes a possible path or outcome. Its inclusion does not necessarily mean that Aegis considers it the most likely outcome.
Where probabilities are assigned to scenarios, they should be interpreted as mutually exclusive or otherwise defined according to the applicable report.
Where no probability is assigned, the scenario is exploratory and should not be treated as a formal prediction.
UNKNOWN and OTHER Outcomes
Aegis may assign probability to UNKNOWN, OTHER, UNRESOLVED, or a comparable category when:
evidence is insufficient;
the outcome universe cannot be reliably enumerated;
an unidentified actor or event may determine the result;
competing hypotheses cannot yet be distinguished; or
false precision would be analytically misleading.
The use of an unknown or residual category is an expression of uncertainty, not a defect to be hidden.
Frozen Forecasts
When a forecast is designated FROZEN, its official substance is preserved after publication.
Aegis does not retroactively change the original probability, direction, target, candidate ranking, scoring rule, or forecast window to improve the result after subsequent events become known.
Formatting corrections or corrections of obvious transcription errors may be made only if clearly documented and without changing the original analytical judgment.
Updated Forecasts
Material new information may justify a new forecast.
An updated forecast must be identified as a separate dated assessment with its own information cutoff. It does not replace the original forecast in the historical record.
Changes between forecast versions may be analyzed to determine:
what new evidence became available;
whether the probability update was directionally justified;
whether the original assumptions failed; and
whether the methodology responded appropriately.
No Hindsight Adjustment
Forecasts are evaluated against the information reasonably available at the time they were made.
Aegis does not judge an earlier forecast as though later information had already been known. Conversely, later events do not excuse analytical errors that were identifiable from evidence available before the cutoff.
The original official version—not an unpublished draft or later reconstruction—is the version used for evaluation.
Resolution and Scoring
Where practicable, Aegis defines the resolution source, date, event, price, index level, measurement rule, and treatment of ambiguity before the outcome occurs.
A forecast may be classified as:
correct;
incorrect;
partially correct;

unresolved;
invalidated by a predefined condition; or
not objectively scorable.
Quantitative forecasts may be evaluated using appropriate measures such as absolute error, directional accuracy, ranking accuracy, calibration, Brier score, or another stated metric.
Different forecast types require different scoring methods. A price forecast should not be scored in the same manner as an election probability, geopolitical scenario, or binary event.
Forecasting Record
The Aegis Forecasting Record may include formal forecasts preserved before their resolution.
The record should distinguish, where applicable:
live out-of-sample forecasts;
historical case studies;
backtests;
simulations;
illustrative examples;
revised forecasts; and
forecasts that could not be objectively resolved.
Backtested or simulated performance is hypothetical and may benefit from methodology selection, data availability, or assumptions that would not have existed in real time.
A limited forecasting record may not establish long-term skill. Results can also vary by time period, asset class, event type, market regime, and forecast horizon.
Selection and Presentation Risk
A selected example may not represent the full set of Aegis forecasts.
Where performance statistics are presented, Aegis seeks to identify the population of forecasts included, the relevant period, the scoring method, and any exclusions.
Users should not infer overall accuracy from a small number of successful examples.
Signal-Before-News Analysis
Aegis may search for lawful, observable precursors that could become visible before a conventional news narrative develops.
Such signals may include anomalies, institutional behavior, filings, logistics, market structure, operational traces, or clusters of indirect evidence.
A precursor is not inside information and is not proof of the predicted event. An anomaly may have multiple explanations, including mechanical, accidental, seasonal, or data-quality explanations.
Signal-before-news analysis remains subject to falsification, competing hypotheses, missing evidence, and model risk.
Sources, Models, and Judgment
Forecasts may combine public sources, licensed data, quantitative analysis, artificial intelligence, structured methodology, and human judgment.
Errors may result from:
inaccurate or delayed data;
source bias or deception;
missing variables;
incorrect causal assumptions;
regime change;
unexpected intervention;
flawed model selection;
misinterpretation; or
ordinary uncertainty.
Source verification and analytical discipline can reduce these risks but cannot eliminate them.
Financial Forecasts
A forecast concerning a security, index, commodity, currency, digital asset, interest rate, or economic variable does not guarantee an investment result.
Actual user performance may differ because of execution price, timing, fees, taxes, spreads, liquidity, portfolio construction, position size, leverage, and risk controls.
Forecasts should be read together with the Financial & Investment Disclosures.
No Guarantee
Aegis does not guarantee that any forecast will be correct, profitable, timely, complete, or suitable for a particular decision.
Forecasting quality must be evaluated across a sufficiently large and appropriately defined body of forecasts—not by whether every individual forecast proves correct.
Contact
Questions about a forecast, its information cutoff, resolution rule, or historical classification may be submitted through:
https://aegisdecision.com/contact