Ravenpine

Our approach

How Ravenpine decides what to highlight — and where it's honest about what it doesn't know yet.

Six time horizons

Every highlight is tagged to a horizon, because what makes a good short-term mover is rarely what makes a good long-term hold.

1 Week

Weighted heavily toward recent price momentum and news coverage.

1 Month

Still momentum-led, with more weight on analyst sentiment.

6 Months

A more even blend of momentum and fundamentals.

1 Year

Leans toward valuation and quality over short-term moves.

5 Years

Driven mostly by valuation and quality, barely by momentum.

Long-Term

A qualitative read on quality and value, not a price bet.

Five signals

Each ticker is scored on five factors, compared against the rest of our curated watchlist on the same day, then combined using weights tuned per horizon.

Momentum

Trailing price return, computed from our own daily snapshots rather than a data vendor's history. It's neutral until we've collected enough of our own data to trust it — a genuine cold start, not a shortcut.

Valuation

How a ticker's price-to-earnings and price-to-book compare to the rest of the watchlist that day. Cheaper, relative to peers, scores higher.

Quality

Profitability signals — return on equity and gross margin — compared across the same watchlist.

Analyst sentiment

A weighted tally of buy/hold/sell ratings from covering analysts.

News coverage

A relevance-weighted read of same-day news sentiment across financial media, so a ticker with genuinely notable coverage stands out from routine mentions.

Personalized to you

The scoring above runs once a day and is shared by everyone — your dashboard doesn't trigger new analysis. What's personal is how it's filtered and ranked for you: your sector interests act as a hard filter, and your risk tolerance re-orders results, favoring lower- or higher-risk tickers depending on your preference.

A model that grades its own homework

Every highlight is logged with the price at the time. Once its horizon elapses, we record what actually happened. Periodically, if one factor's picks have meaningfully out- or under-performed for a given horizon, we nudge that factor's weight — a bounded, explainable adjustment, not machine learning.

We're upfront that this takes real time to mean anything: a 1-week highlight gets its first verdict in a week, but a 5-year highlight won't for five years. The scoring works from day one; the self-improving part earns itself gradually.

Not financial advice. Highlights are generated by an automated, experimental model for informational purposes only — always do your own research before making investment decisions.