About Důchodce investuje

How AI assesses companies and share prices

ChatGPT produces both scores, including the assumptions and calculations behind them. It starts with available company filings, results, presentations and transcripts, separating reported facts, management guidance and model assumptions. Two independent scores summarize its conclusions; neither is a probability of success or an automatic trading instruction.

Methodology 1.2 · effective October 16, 2026.

Business strength

How strong is the company's ability to create long-term shareholder value despite competition and obstacles? The current stock price does not enter this assessment.

Area · 20% eachWhat AI checks
Competitive advantageMoat, switching costs, distribution and competitive position.
Demand durabilityProduct necessity, cyclicality, replacement risk and room for further development.
Management executionDelivery on promises, capital discipline and responses to problems; results take precedence over presentations.
Business economicsMargins, cash flow and per-share value after stock compensation and dilution.
Financial resilienceDebt, maturities, liquidity and the ability to withstand adversity without damaging shareholders.
  1. Assign a score from 1 to 10 in 0.1-point increments to each area, with a concrete piece of evidence and its period. Evidence fits the business model: for example, AFFO for a REIT, NII and credit quality for a BDC, or retention and per-share cash flow for software.
  2. Take the arithmetic average of all five scores and round it to one decimal place. Each has the same weight; there are no hidden bonuses or deductions.
  3. Name the main weakness the average may conceal, state confidence and describe observable developments that would change the score. If a material evidence gap prevents assessment of any area, the overall score is unassessed.

Price attractiveness: return versus risk

How favorable is the expected return–risk balance at a specific price over three years? A good business and an attractive stock price are assessed separately.

  1. Verify the stock or share class, price, currency, quote date and source. Record the assessment date separately and identify a closing price as such.
  2. Build a realistic base case and a relevant bear case with the same three-year horizon. State assumptions for growth, margins, financing, share count and future valuation. Do not count dilution or debt twice.
  3. Estimate the price in three years, dividends over that period and total return in both scenarios. Unless stated otherwise, dividends are not reinvested and returns exclude taxes, fees and currency conversion.
  4. Assign a score from 1 to 10 in 0.1-point increments according to whether the base-case return sufficiently compensates for possible losses and uncertainty. A bull case, a fall from the historical high or a high dividend yield alone does not justify a high score.
  5. State the main sensitive assumption, confidence and a price or business change that would require reassessment. Without a verified price or defensible scenarios, price attractiveness is unassessed.

Three-year return = (price in three years + dividends over three years) / starting price − 1.

When assessments are updated on the website

Company analyses are updated when relevant new evidence becomes available: results, guidance, changes in business economics, financing or risks. The scores can remain unchanged if the new evidence does not justify revising them.

We check the latest regular-session closing price daily. A rise or fall of at least 15% from the price used in the latest price attractiveness assessment triggers a reassessment. The time elapsed since that assessment does not affect the threshold.

A price move prompts a reassessment of price attractiveness, not business strength or an automatic buy or sell signal. Moves below the threshold and routine checks do not create public Updates. A qualifying price reassessment can retain the existing scenarios. The published scores keep their original dates and prices; the live chart does not recalculate them.

Reading the 1–10 scale

ScoreBusiness strengthPrice attractiveness
1–2Fundamentally threatened businessPrice requires exceptionally optimistic outcomes
3–4Weak position, many unresolved problemsReturns offer little compensation for risks
5–6Viable business with significant limitationsApproximately fair return–risk balance
7–8Strong business with proven advantagesAttractive price under sober assumptions
9–10Exceptionally strong and resilient positionExceptional opportunity with a substantial margin of safety

Confidence is stated separately for each score as high, medium or low. It reflects the consistency and completeness of evidence and reliance on uncertain assumptions; it is not an extra score or an automatic deduction. Scores of 9–10 should be rare and supported by convincing evidence.

Business strength changes with new business evidence. Price attractiveness changes with the quoted price and model assumptions. Every update retains the date, methodology version and reason for the change in the thesis history. Earlier scores are comparable only under the same methodology. Live market charts do not automatically recalculate the analysis.

How the website is built and updated

Důchodce investuje is a presentation layer for ongoing research using ChatGPT. Two connected parts sit behind the website: a Space containing research Pages and the website that prepares their content for readers.

The author selects the companies and designs the structure and rules. ChatGPT produces the analyses, assumptions, calculations, scenarios and scores using the latest GPT model available at the time of processing. Older analyses therefore do not automatically come from the same model as the newest ones.

1. Research starts in the Space

Each company followed has its own Page. It brings together business analysis, financial results, information from reports and earnings-call transcripts, valuation, risks and the evolution of the investment thesis.

Company Pages have their own update schedules, usually weekly. Processing follows a shared template and methodology: the standard 13 areas of analysis, rules for working with sources, and the procedure for assessing business strength and price attractiveness.

The aim is to make companies comparable without losing the differences between their business models.

2. The website checks what has changed

A plugin connects the Space and the website through the MCP interface. The website automation checks the versions of registered company Pages daily. It loads their full content only when a new version has been created.

It then compares the new version with the last successfully imported source. It distinguishes a substantive change in the analysis from simple reformatting or changes in wording.

A new Page version does not automatically trigger publication. The website imports relevant new evidence, or a price-only reassessment after a rise or fall of at least 15% from the price used in the latest price attractiveness assessment. Other checks remain private. See when assessments are updated.

Financial research is not repeated during this transfer. The prepared Page is the source of the analysis; the website layer is responsible for faithfully converting, translating and displaying its content.

3. One update reaches the whole website

Substantive changes are incorporated into both the Czech and the complete English company analysis. The source scores, currencies, reference-price dates, scenarios and material analytical uncertainties are preserved.

An update is saved as a single coordinated package. It appears in the company analysis, cards, every Company comparison view and the bilingual Updates feed. Dividend classification remains a separate decision by the author.

If an update is incomplete or fails validation, the last valid version stays on the website. Processing the same version again does not create another record.

4. History is preserved

Published versions and previous scores are kept in a persistent database. Each assessment retains its date, reference price and other related information. This will make it possible to track how business strength, price attractiveness and the investment thesis evolve.

Routine content updates do not require a new website deployment, and the stored history survives subsequent website changes.

The stock chart is a separate information element. Its current price does not automatically recalculate the scores: they always belong to a specific analysis and its dated reference price.

5. The daily commentary has its own publishing path

The daily investment and macroeconomic commentary is produced separately in the Page Denni_komentar, following its own editorial rules. The website automation imports the completed edition, prepares an English version and stores it as a separate article.

Each edition has its own date, URL, sources and publication time. The newest commentary appears at the top, while earlier editions remain in the archive. Corrections are recorded; a new commentary does not overwrite previous articles or change company analyses and their scores.

Automation helps keep research and its presentation aligned. It does not, by itself, guarantee that an investment conclusion is correct. Readers should always distinguish supported facts, model assumptions and uncertainties.

Explore the investment analyses