How Openbook scores stocks and portfolios.

The Reward Score

The Reward Score is a weighted average of four factor scores, each itself on a 0–100 scale. Higher means better expected upside. The factor weights are fixed across the universe (a 40/25/20/15 split) and were chosen to roughly match the empirical weighting used in academic factor research while giving Growth and Momentum more influence for our retail audience, who skew toward growth-tilted UK and US equities.

The Risk Score (equity)

The Risk Score is also a weighted average of four factors on a 0–100 scale, with the convention that lower is safer. Volatility carries the largest weight because, for retail investors, day-to-day price behaviour is the risk experience that actually matters. Financial solvency, operational quality, and size factor in second-order risks that mostly express through volatility but warrant their own line in the score.

Portfolio scores

Portfolio-level Reward and Risk are not just averages of the stock scores. The Reward Score aggregates each holding's growth, momentum, profitability, and valuation factors using current position weights. The Risk Score is built differently. Averaging stock-level risks misses the failure modes that actually blow up retail portfolios (too much in one ticker, too much in one sector, six oil names that all move together), so the portfolio Risk Score is built from four structural factors instead.

Portfolio Reward Score

For each of the four reward factors (Growth, Momentum, Profitability, Valuation), we take the position-weighted average of each holding's score on that factor. We then blend those four with the same 40/25/20/15 weights used at the stock level. The result is interpretable as "the average reward profile of the portfolio, weighted by how much money you have in each position."

Portfolio Risk Score

The portfolio Risk Score uses a different factor model than the equity Risk Score because portfolio risk is structural, not aggregated.

The Concentration and Sector Concentration factors use the Herfindahl-Hirschman index (HHI), a standard measure of concentration. We normalise the HHI between its theoretical minimum (perfectly equal weights across N holdings, 1/N) and its maximum (a single position, 1.0), then map the result to a 0–100 risk score.

Refresh cadence and data sources

• Stock-level scores are recomputed on a weekly cycle from the underlying fundamentals and price data, with intra-week refreshes for major corporate events (results, takeover bids, profit warnings).
• Portfolio scores are computed live in the browser whenever you load the Analytics tab. Holdings, weights, and the latest stock-level scores drive the calculation. There is no batch step.
• Price data is end-of-day, sourced from the same feed used by FTSE-listed brokers (LSE for UK-listed names, exchange of primary listing for US names).
• Fundamentals are based on the most recent reported annual and quarterly statements, normalised across reporting calendars. Companies with non-December fiscal year-ends are handled correctly.
• Analyst estimates and price targets are sourced from the consensus reported by listed brokers, refreshed weekly.

What we cover today

• LSE-listed equities. Full Reward and Risk score coverage across the FTSE 100, FTSE 250, FTSE Small Cap, and AIM-listed names with at least three years of reported financials. Approximately 1,650 named businesses.
• US-listed equities. Full fundamental and price data coverage across major exchanges (NYSE, NASDAQ, AMEX). Proprietary Reward and Risk scores are in beta and may not appear for every ticker. Expanding coverage is the highest-priority data engineering project for the platform.
• UK and US listed ETFs. Holdings, sector exposure, and total expense ratios for the major UK and US listed ETFs. Per-ETF Openbook Reward and Risk scores are limited to a curated set of ~200 large UK ETFs; uncurated ETFs render without scores.

Known limitations

We do not believe in pretending a model is more complete than it is. The current scope of the methodology has the following limitations, in approximate order of how much they could matter to you:

• US factor scoring is incomplete. Some US tickers do not yet carry a Reward or Risk score. We display blanks rather than estimates in those cases.
• Sector classifications are imperfect for conglomerates and special situations. A holding company sitting across three sectors will be assigned the parent's primary sector, which may understate sector concentration in the portfolio score.
• Correlation risk uses a sector proxy, not returns-based correlation. True pairwise correlation from daily returns is more accurate but data-heavier. We use the simpler proxy (same-sector pair density) until the returns-based version is ready.
• The scores are model output, not analyst opinion. A 75/100 Reward Score does not mean the analyst team thinks the stock will outperform. It means the underlying data places the business in the strong band of the universe across the four reward factors. Our analyst notes in Insights are where you find the opinion.
• Score history before 2024 is reconstructed. Where we display "+12 over 2 years" or similar score-trend annotations, the historical data points are model output run against prior fundamentals, not contemporaneous published scores.

Versioning

Methodology changes are versioned. When a factor weight, an input, or an aggregation rule changes, we bump the version on this page and date the change. We do not retrospectively rewrite historical scores; if you want to see how a stock would have scored under a prior version, contact us and we will run the prior model on request.

Questions