Methodology

Selecting companies

We check over 1,000 companies listed in the US and consider only the large ones. We leave out companies at risk of bankruptcy and extremely expensive ones. Every month the author picks the candidates from them, and the system scores each of them in three categories (profitability, fundamentals and valuation) and in total, all in percent.

The scores come from 10 indicators: 5 historical ones are computed from the companies' financial statements, and 5 forward-looking ones are rated by the author. Profitability weighs the most, half of the total score, because over the long term share prices tend to follow a company's earnings. Up to 22 companies with the highest total score make the report. With equal scores, the profitability score decides, then valuation.

Where the data comes from

Historical indicators are computed from the financial statements that companies file with the U.S. Securities and Exchange Commission (SEC, the EDGAR database): quarterly 10-Q and annual 10-K reports. Foreign companies file annual 20-F reports instead, Canadian companies 40-F reports (some also interim 6-K reports): for them we compute the indicators from the latest available period. Amounts in a currency other than the dollar (also in some 10-K reports) stay in the currency of the filing, and for the Ohlson model we convert assets to US dollars at the FRED exchange rate. We compute the market cap, P/E and P/S of such companies in US dollars: the filing's amounts are converted at the latest FRED exchange rate, and the price of a depositary share (ADR) is divided by the number of ordinary shares it represents. We read that ratio from the cover or the text of the 20-F on EDGAR (e.g. one TSMC ADR is five shares). For every company we state which filing the data comes from and when it was filed.

Forward-looking indicators, that is revenue, earnings and cash flow estimates and analysts' price targets, are rated by the author from publicly available information, including sites such as Yahoo Finance, Investing.com, MarketWatch, Nasdaq.com, TipRanks, Zacks, MarketBeat, Seeking Alpha and Stock Analysis. The author uses AI tools to search the web and these sites and to help assess the data found. The report gives the author's rating, not the values from these sites: these ratings are the author's opinion, not a data provider's data.

Prices of the shares and of the SPY and QQQ ETFs are entered by the author from public quotes (the US market close). The date of the prices is given in the report.

Indicators, scales and weights

The weight tells how much an indicator counts in its category score and in the total score; the weights of all indicators add up to 130. Indicators from filings are scored on a scale from the worst to the best value, and forward-looking indicators are rated by the author.

For the bankruptcy probability the scale runs from 4% to 0%: the lower the value, the better.

How the scores are computed

An indicator from filings is scored in proportion to where its value lies on the scale: score = (value − worst) / (best − worst) × 100%. A value worse than the worst gets 0%, a value better than the best gets 100%. For example, a value of 75 on a 50 → 100 scale scores 50%.

When the 5-year change cannot be computed because the value 5 years earlier was negative or zero, the indicator scores 0%.

A category score is the weighted average of its indicator scores: the sum of weight times score divided by the sum of the category's weights (profitability 65, fundamentals 25, valuation 40). The total score is the weighted average of all 10 indicators, that is the same sum over all indicators divided by 130.

Indicator scores are rounded to two decimals, and the averages are computed from the rounded scores. This way the category and total scores can be recomputed from the scores shown in the report.

Example

A sample company with made-up data, scored step by step.

Profitability

(5 × 58.33 + 15 × 100 + 5 × 50 + 5 × 50 + 15 × 100 + 5 × 75 + 15 × 50) / 65 = 4,916.65 / 65 = 75.64%

Fundamentals

25 × 75 / 25 = 1,875 / 25 = 75%

Valuation

(15 × 50 + 25 × 25) / 40 = 1,375 / 40 = 34.38%

Total score

(4,916.65 + 1,875 + 1,375) / 130 = 8,166.65 / 130 = 62.82%

Results are rounded to two decimals, halves up.

Definitions

TTM (trailing twelve months)

The result of the last four quarters. When the latest filing is an annual report, it is the annual result. Otherwise: the result of the last fiscal year + the result since the start of the current year − the result for the same period of the previous year.

5-year change

(TTM now / TTM 5 years earlier − 1) × 100%. EPS is adjusted for stock splits so that both values are comparable.

EPS Basic

Net income attributable to common shareholders divided by the weighted average number of shares, as reported by the company.

EBITDA margin

(EBIT + depreciation and amortization) / revenue, for the trailing twelve months. EBIT is operating income; when a company does not report it, we use pretax income plus interest expense.

Bankruptcy probability (Ohlson)

James Ohlson's O-Score model (1980) estimates the probability of bankruptcy within a year from nine ratios: company size, leverage, working capital, liquidity, profitability, operating cash flow and the change in net income. Size is the logarithm of total assets in USD adjusted for the price level (1968 = 100, using the GDP deflator from FRED). Probability = eᴼ / (1 + eᴼ), where O is the model's score.

When the filings lack data

We assess only companies whose SEC filings contain all the data the historical indicators need. Companies with incomplete data do not enter the ranking. This applies, for example, to banks and insurers, which report no current assets (the EBITDA margin and the Ohlson model do not fit them).

Valuation

For each company we show the market cap (the price times the shares outstanding from the latest filing), the P/E ratio (the price divided by diluted EPS for the trailing twelve months; not shown with negative earnings) and the P/S ratio (the market cap divided by revenue for the trailing twelve months).

The author also gives a valuation verdict in one of three states: undervalued, fairly valued or overvalued. The reasoning is part of the qualitative analysis. The verdict is not part of the total score.

Qualitative analysis

21 questions in six areas. Each answer earns the company 0, 0.5 or 1 point. The report shows every question, answer and points, a summary, the full analysis and the good signals, bad signals and warning lights. The qualitative analysis complements the numerical score and is not part of the total score. The author prepares it from publicly available information, using AI tools to search for information and to help assess the answers. The author reviews the AI's output, but errors may still occur.

• Industry and growth potential3
• Product and business model5
• Competitive advantage (moat)6
• Customers and pricing power3
• Resilience and risks3
• Management1

Portfolio changes

Every report is compared with the previous one: new companies, changes in rank and removed companies together with the reason for removal.

Portfolio simulation

The simulation answers the question of how much an investor following the reports since the first one would have earned. It starts with 100,000 USD. At every report the portfolio is valued and rebalanced to equal weights: removed companies are sold, new ones bought, and the others topped up or partly sold.

It uses fractional shares and ignores fees, spreads, taxes and dividends. The author enters prices from public quotes: as of each report and a few times a month between reports. Each time the portfolio is revalued and compared with the SPY (S&P 500) and QQQ (Nasdaq-100) ETFs since the first report.

Simulation results are hypothetical. Past results do not guarantee future results.

Publication and disclosures

Reports are published on the 15th of every month. Each report states the author, the date and time it was prepared and first published, the data sources, the author's holdings and each company's rank history for the last 12 months, as the Market Abuse Regulation (MAR) requires.