Methodology · updated October 21, 2026
Methodology: how we calculate company verdicts, fair value and scores
We start from company filings and exchange prices, calculate more than 300 metrics and run them through models tested on years of historical data. The result is a reading of the data on the day it was calculated, not personal investment advice.
- 300+metrics calculated for every company
- Dailyrecalculated after the market closes
- 12 monthsthe horizon we test the verdict on, using past data
How a company gets its verdict, in four steps
Every company goes through the same steps, under the same rules. No one picks a company's verdict by hand.
- We collect the data
Financial statements, prices, analyst estimates, insider trades and news, from market data providers and public sources.
- We calculate the metrics
Profitability, growth, debt, cash flow, valuation, dividends, risk and price trend. We check units, currencies and split adjustments.
- We run the models
The metrics are grouped into families. Each family answers one question: is it a good business, is it financially sound, is it growing, is it expensive, where is the price heading.
- We publish the reading
The verdict, fair value, scores and scenarios, with the calculation date, the price used and a plain-language explanation.
What the verdict means: from Very attractive to Very unattractive
The verdict shows how a company stands today, across all its data, compared with the other companies we cover. It is a comparison, not a price forecast and not a call to buy or sell.
- Very attractive
The data looks much better than for most companies: a solid business, a reasonable price for what it earns and signals that back each other up.
- Attractive
More arguments for than against, with a few points worth watching.
- Neutral
The arguments for and against balance out, or the data shows no clear direction.
- Unattractive
More reasons for caution than arguments for: a high price, weak results or financial risks.
- Very unattractive
Several serious problems at once, such as heavy debt, falling profit and a price that does not reflect them.
The reference horizon is 12 months. Over that period we checked, on past data, how companies in each rating went on to perform. Past results do not guarantee future results.
Next to the verdict you also see the confidence level, meaning how complete and consistent the data is. When there is not enough data for a fair reading, we give no verdict and state the reason instead.
The models we use: what goes into the calculation
Eight families of metrics, each with its own question. On every company page you can see the figures in each family and which way they point.
- Business quality
Profitability, return on capital, margins and how steady earnings are from year to year.
- Financial health and risk
Debt, liquidity, interest cover and well-known academic scores such as the Altman Z, Piotroski F and Beneish M.
- Growth
How sales, profit and cash flow have developed over several years, plus analyst estimates.
- Valuation
Multiples such as P/E, EV/EBITDA or P/B, compared with the company's sector where that comparison makes sense.
- Fair value
Several valuation models, including discounted cash flow (DCF), with assumptions about the cost of capital, growth and terminal value. It is an estimate, not a price target.
- Price trend
Moving averages, momentum, volatility and other technical signals, read on each stock's own chart.
- Sentiment
Analyst ratings and price targets, insider trades and short interest, where this is public.
- Accounting red flags
Automated checks for patterns that, in the past, came before accounting problems at some companies.
How they are combined
The families are combined by a rules-based model, the same for every company, calibrated and tested on years of historical data. On every page we publish what went into the calculation, which way each family points and where the signals agree or conflict.
What we do not publish
The weights, thresholds and the way we calibrated the models stay internal. They are the research behind the platform, as with any provider of financial analysis. We are open about the fact that they exist.
The role of AI
AI does not decide the verdict or the scores. We calculate them with fixed rules. AI only writes part of the explanatory text, starting from figures that have already been calculated.
Data sources and how often we update them
We use licensed market data providers and official public sources. The date and time of every calculation appear on the company page.
| Data | What we take | Updated |
|---|---|---|
| Financial statements | Balance sheet, income statement and cash flow, from the reports companies publish | After every new report |
| Exchange prices | Closing price and price history, adjusted for splits and dividends | Daily, after the close. Intraday prices may be delayed by up to 15 minutes |
| Cryptocurrencies | Prices from cryptocurrency exchanges | Close to real time |
| Analyst estimates | Ratings, price targets and revenue and earnings estimates, compiled by data providers | When new estimates come out |
| Insiders and shareholders | Trades by management and major shareholders, from public filings with regulators | When filed |
| Macroeconomic data | Inflation, interest rates and government bond yields, from official sources | Monthly or when published |
| News | Headlines and summaries from news publishers and company press releases, linked to the source | Continuously |
We do not independently check every figure we receive. We run automated checks (units, currencies, impossible values), but an error at the source can still reach our calculations. When we find one, we fix it and recalculate.
Price projections and scenarios
The scenarios show a range of possible prices in 6 and in 12 months, with the probability of each outcome. We get them from a statistical model trained on the history of thousands of companies, not from an opinion about any single company.
A 70% probability is not a promise. It means that in similar situations in the past, the outcome happened roughly seven times out of ten. The market can move outside any range, especially around rare events.
The stress scenarios show how far a stock could fall if the market dropped 20% or 40%, based on how it has moved relative to the market in the past. Fair value is an estimate too, and it changes as the data and the assumptions change.
News, Flash items and automated text
News comes from news publishers and company press releases. We select and group it automatically by company, topic and country, and show a short summary with a link to the original article. The rights to the articles stay with their publishers.
Some text on the site is written automatically with the help of AI: news summaries, company descriptions and the explanations around the figures. We generate it from verifiable data, but it can contain mistakes. When a piece of text and a figure disagree, the figure takes precedence.
Limits of the analysis
- We work with past data. The models learn from history, and markets can behave differently from how they did before.
- Data can be delayed or wrong at the source, especially for small companies and recent listings.
- We do not know your situation: your income, goals, time horizon or how much risk you can take. That is why nothing here is a personal recommendation.
- We cannot see everything. Unannounced events, lawsuits, management changes or political decisions can change a company quickly.
- When we do not have enough data, we give no figure. We state the reason on the page instead.
- Investing involves risk. You can lose some or all of the money you invest. Past returns do not guarantee future returns.
Independence and conflicts of interest
- We take no money from the companies we analyse and accept no payment for a rating or for changing one.
- We do not change a company's verdict or scores by hand. They are all calculated under the same rules.
- We earn from subscriptions, advertising and affiliate links to brokers. If you open an account through such a link, we may receive a commission at no extra cost to you. Affiliate links are labelled.
- Brokers have no influence on the analysis. They have no access to the models and cannot ask for a rating to be changed.
- If anyone on the team holds shares in a company we analyse, we say so in the disclosure on that company's page.
Who publishes the analysis and our status
The analysis is produced and published by einvestitii.ro. We are not an investment firm and we are not authorised or supervised by any financial regulator, including Romania's Financial Supervisory Authority (ASF). We do not provide investment services: we do not manage money, execute orders or give personal advice.
Our analysis follows the EU rules on investment recommendations (Regulation (EU) No 596/2014 on market abuse and Delegated Regulation (EU) 2016/958). That is why every analysis shows who produced it, when it was calculated, the price used, a link to this methodology and any conflicts of interest.
Rating history over the last 12 months
Every company page shows the history of its verdict: when it changed, from which rating to which, and how the price has moved since. The history is public, even when today's verdict is only available to subscribers.
We do not delete past ratings when a company turns out differently from what the verdict suggested. They are exactly what shows you how well the method has worked.
This is not investment advice
All content on this site is for information and education only. It is not an offer, a solicitation or a personal recommendation to buy, sell or hold any financial instrument.
Investment decisions are yours. Before you invest, think about your goals, your financial situation and how much risk you are willing to take, and if you need to, ask a licensed financial adviser.
Changes to the methodology
We improve the models over time. When we change how the verdict or the scores are calculated, we update this page and record the change here, with its date.
- First public version of the methodology.
Frequently asked questions about our methodology
How is a stock's verdict calculated?
The verdict combines families of metrics (business quality, financial health, growth, valuation, price trend and sentiment) through a rules-based model tested on years of historical data, over a 12-month horizon. We publish what goes into the calculation and which way each family points, but not the model's weights.
Is the verdict a buy recommendation?
No. It is a reading of the data on the day it was calculated, comparing the company with the other companies we cover. It does not take your situation into account and is not personal advice.
Where does the data come from?
From the financial reports companies publish, licensed market data providers, public filings with regulators and official sources for macroeconomic data. We recalculate every day after the market closes.
Why don't you publish the full formula?
The models are the result of our research. We publish what goes into the calculation, how to read each metric and how we tested the model, but the weights and calibration stay internal, as with any provider of financial analysis.
What is a stock's fair value?
It is an estimate of what a share is worth, drawn from several valuation models, including discounted cash flow. It depends on assumptions and changes when the data changes. It is not a price target and not a promise.
Are you a regulated investment adviser?
No. We are not an investment firm and do not provide investment services. We publish analysis for information, with the disclosures EU rules require for investment recommendations.