The data

Company filings plus Börsdata (fundamentals and prices). We adjust reported accounts toward economic reality — for example capitalising research & development the company expensed, and stripping one-off items — to get a truer picture of returns and profit.

Scoring each company

Every company gets a quality score (returns on capital, margins, balance-sheet strength, and the return on its newest capital) and a value score (price versus the value of its economic earnings, and a reverse-DCF check on the growth the price assumes). A composite blends quality 40% / value 40% / price-trend 20%.

Banks and insurers can't be measured the same way, so they're scored on return-on-equity and price-to-book instead.

Building the portfolio

We take the top ~30 by composite that also pass a price-trend gate (trading above a rising long-term moving average), size them by inverse volatility, and cap any one name at 8% and any one sector at 25%. We rebalance roughly monthly, holding each name until it clearly drops out of the top 50 (hysteresis) to keep trading low.

Two layers of depth

Most companies are scored automatically. The ones we hand-read in detail are badged Streamlined or Manual and are protected from being overwritten.

Honesty & limits

Where a number is missing we mark it missing — we don't fill gaps with guesses. Back-test figures already deduct modelled trading costs (~40 bps per side) but not tax, and the past is not a forecast.

Execution timing. The back-test fills at each weekly signal's own closing price. That is slightly optimistic — in reality you'd act on the close and trade the next session. We tested a deliberately harsh version of that (a full one-week execution lag, more conservative than a next-day open): it is roughly neutral for the Nordics (CAGR 22.5% → 22.5%) and trims about 1.5 points a year off North America (25.8% → 24.3%). The edge survives a pessimistic fill assumption — it isn't an artefact of trading at a price you couldn't get. We don't yet model intraday opens (the price history is weekly closes).

Data as of 2026-06-21.