Coverage
Whether a distribution is supported by earnings and cash flow, and how confident we are in the underlying data.
METHODOLOGY / EXPLAINABLE RISK
Research framework in development · Version 0.1
A transparent framework designed for testing.
Our target design assesses dividend-reduction risk for eligible S&P 500 dividend-paying common stocks. The figure currently shown in the product is an earlier and broader estimate. It is not a validated model, a probability that a dividend will be cut, a price target or a recommendation.
Development status
No weights, thresholds or display bands are published as validated. They will be published only after point-in-time historical testing is complete, documented and approved. An in-development estimate appears in the product today: it covers a wider set of securities than the target design below, and it must not be read as a validated assessment. Illustrative examples elsewhere on this site are sample data, not live research.
What the current estimate uses
The estimate in the product today is built from the five inputs below. We do not publish their weights or thresholds, because those have not yet been tested. These inputs are not the candidate factor families listed further down — those describe the target design.
Distribution instability
Whether the distribution amount has moved in ways that suggest deterioration.
Yield anomaly
Whether the yield is unusual relative to the issuer’s own history and to comparable issuers.
Cut history
Whether a reduction, omission or suspension has been observed.
Peer stress
Whether comparable issuers show similar yield pressure at the same time.
Bands
Bands describe how much risk evidence was observed. They are labels for evidence, not forecasts, and a band is never a prediction that a dividend will be cut.
Stable
Little observed risk evidence.
Moderate
Some evidence of deteriorating distribution safety.
Elevated
Corroborating evidence across several inputs.
High risk
Sustained or multiple indicators of distribution pressure.
Not applicable
Insufficient or irreconcilable data. This is an absence of evidence, not evidence of safety.
Candidate factor families
These are the target design: each family is a hypothesis to test, not a finished input. Higher eventual scores would indicate more observed risk evidence — not a probability, prediction or recommendation.
Distribution coverage
Earnings, operating-cash-flow and free-cash-flow coverage, including persistence and trend.
Cash-flow quality
Conversion, working-capital dependence, recurring cash generation and volatility.
Balance-sheet pressure
Leverage, liquidity, interest coverage, debt structure and maturity concentration.
Operating deterioration
Revenue, margin and operating-income direction using point-in-time evidence.
Dividend & market evidence
Dividend behavior, governance signals, yield stress and relative market pressure.
How a published score will be produced
Define
Specify the eligible universe, dividend-cut event and every input before testing.
Collect
Use filing acceptance times, issuer documents and dated constituent snapshots.
Normalize
Resolve periods, units, currencies, signs, amendments and restatements.
Validate
Reconcile sources, test formulas and block publication when critical checks fail.
Backtest
Separate training, validation and out-of-time tests while preventing hindsight leakage.
Explain
Publish score drivers, confidence, data dates, limitations and model version.
Required publication record
Data-through dateModel versionSource manifestNormalized inputsValidation statusDrivers & counterevidence
A numeric score will not be published when material data is missing or cannot be reconciled. Missing data is uncertainty — not evidence of safety.
Scope discipline
Different structures need different models.
Our target design covers common stocks. REITs, BDCs, preferred securities, covered-call funds and structured products require purpose-built definitions rather than being forced into the common-stock framework. Until those definitions exist, the current in-development estimate includes those structures, and its figures should be treated as provisional for them.
Limitations
No model eliminates uncertainty. Assessments can be affected by delayed disclosures, data errors, regime changes and issuer-specific events. The estimate described on this page has not yet been validated against historical outcomes.
Dividend Portal does not provide individualized advice, price targets or guarantees. Methodology and key compliance language will receive limited securities-counsel review before launch.