Where the data comes from, and how firm track records are computed

Consensus ratings, rating distributions, price targets, and per-firm rating actions (upgrades, downgrades, initiations, reiterations) are sourced from Yahoo Finance and are delayed. Every rating is the published opinion of the named firm at the stated date. We display them verbatim; we do not adjust, weight, or editorialize them.

Rating labels differ across firms ("Overweight", "Outperform", "Accumulate"…). For statistics only, we map each label to a buy / hold / sell equivalence class using a fixed public table. Labels we can't confidently map are excluded from statistics, never guessed.

For every buy-equivalent rating action in our coverage:

1. Take the stock's closing price on (or the first trading day after) the rating date, and again 3 months (91 days) and 12 months (365 days) later.
2. Compute the stock's return over each window, and the excess return vs the S&P 500 (SPY) over the exact same window.
3. A rating "beat the market" if its excess return is positive. A firm's success rate is the share of its evaluated buy-equivalent ratings that beat the market; we also report its average excess return.

Ratings too recent to evaluate (younger than the window) and ratings without sufficient price history are excluded from the denominators. Every statistic is shown with its sample size (n=), and firms enter the leaderboard only with at least 5 evaluated 12-month ratings. Track records are computed at the firm level (analyst-person attribution is not in our source data).

Track records are descriptive history, recomputed nightly and versioned (current: v2). They are not predictions. Past performance does not guarantee future results.