How the Veridion Score is built

What the Score is

The Veridion Score is a 0-100 conviction index on actively traded US stocks. It is published daily after the US close. The public methodology lives at /methodology.

The Score is a calibrated composite of six independent factor categories. Each factor is computed from observable public data and normalized to the shared 0-100 scale before aggregation.

• Valuation
• Hype
• Momentum
• Earnings
• Analyst rating
• Sentiment

The categories are published because an institutional reader should know what drives the number before relying on it as research context. The exact weights, normalization, and calibration stay server-side.

What each factor measures

Sentiment. News flow on the ticker over a recent rolling window, classified by direction and net-weighted. Duplicate-article suppression keeps a syndicated headline from inflating the read. Sentiment is a confirmation factor and can lag fast-moving names.

Hype. Attention velocity, abnormal volume, and breakout pressure. The factor uses only real market bars and ticker-level article counts, with financial safeguards for deeply unprofitable companies.

Analyst rating. The current covering-analyst rating mix. The model uses the rating tape rather than price targets because targets can update slowly.

Momentum. Recent return and moving-average structure, normalized against the ticker's sector context. This keeps broad market beta separate from ticker-specific relative strength.

Valuation. Sector-relative valuation discipline using available multiples and free-cash-flow yield. Missing financial data is excluded rather than filled.

Earnings. Trailing earnings track record using actual results versus consensus where both fields exist. Earnings revisions and surprises are durable inputs in the research literature without making one factor dominate the composite.

How the composite works

For each ticker, every factor with usable data contributes through the server-side calibration. Factors without usable data are dropped from the composite. They are not filled with a neutral value.

The score is shown with a confidence badge. The badge describes how broad the data coverage was for that snapshot without exposing internal gate logic.

The band assignment is also coverage-aware. Sparse coverage can still produce a numeric read, but the confidence badge tells the reader that the read rests on a smaller subset.

What the Score is not

The Score is not a forecast. It observes how the six public factors aligned as of the snapshot date.

The Score is not financial advice. Veridion is not a registered investment adviser. The platform does not direct a transaction.

The Score is not complete. It is a fundamentals and flow composite. It does not measure options structure, macro regime, or short interest. The methodology page documents deliberate omissions.

Evidence layer

Forward returns by band are captured every trading day and aggregated against the original Score over 30, 60, and 90-day windows. The public evidence page at /backtest shows mean return, sample size, and standard error per band per window once the sample-size gate clears.

Rows without complete forward-return windows are dropped. They are not zero-filled.

Closing note

The Score is a structured read on public data. The math is documented at /methodology. The evidence is at /backtest. The data is observation, not transaction advice.

Not financial advice. Just receipts.