How our ratings are calculated
Allocaria shows three different stock scores and a crisis index. This page says briefly what each one shows and where its limits are. The detailed method is not published.
Three numbers, three jobs
Each answers a different question, so one stock can score high on one and low on another.
| Score | Scale | Where it appears | What it shows | Trained model? |
|---|---|---|---|---|
| Allocaria Score | 0–100 and a word | Home page, Screener and the stock pages in the account area | A score of a company’s latest published data, built with fixed rules | No |
| Opportunity score | 0–100 and a word | Top ideas and the allocation calculator in the account area | Orders stocks for a closer look by combining several facts about price, value and the business | Partly — one of its inputs is model-based |
| AI signal | Rank 0–100 (50 = middle) | Advanced view of the stock pages in the account area | Where a trained model’s prediction for the stock ranks among all ranked stocks | Yes |
How a score reads as a word
The Allocaria Score and the Opportunity score share the same four words. The AI signal is a rank, not a score, and has none.
| Score | Reads as |
|---|---|
| 85–100 | Strong |
| 60–84 | Good |
| 40–59 | Neutral |
| 0–39 | Weak |
A relative score on current evidence — not a probability, a price forecast or a guarantee.
Allocaria Score
One number from 0 to 100 that scores a company’s latest published data. The rules are fixed and the same for everyone.
What it is
- A number from 0 to 100, built from the company’s latest published data.
- The same number for every user, recomputed whenever the data is refreshed.
What it is not
- Not a recommendation or advice: it says nothing about what to do with a stock.
- Not a forecast of future returns.
- Not personal: it is the same for every user and ignores your goals, horizon, risk tolerance and holdings.
How to read it
- Compare scores with each other on the same date, not with the past.
- A high number is a standing among today’s stocks, not a verdict on any one of them.
- The words Strong, Good, Neutral and Weak only name the score ranges shown above.
Rules
- When too little data is available, the score is marked partial or not shown at all — never a guessed one.
- The rules that turn facts into a score are fixed and chosen by judgment, not fitted to past returns.
Opportunity score
Orders stocks for a closer look by combining several facts about price, value and the business. It is for display only and never sets a portfolio weight.
- A high score marks a stock for a closer look today. It is not a forecast and not an instruction.
- The same for every user: it ignores your goals, horizon, risk tolerance and holdings.
AI signal
The only part that comes from a trained model. It is shown on its own and never sets a portfolio weight.
A percentile rank from 0 to 100: where the model’s prediction for a stock sits among all the stocks it ranks. 50 is the middle by construction. It is a rank, not a probability of beating the market.
- The model is trained on company fundamentals and price behaviour. Its inputs and settings are not published.
- Its link to later returns is weak: across many stocks it tilts the odds slightly, but for any single stock the outcome stays close to a coin toss.
Fair value and the gap to it
An estimate of what a share is worth based on the company’s reported figures — not a price target and not a forecast.
- The gap is the distance between the price and this estimate. Very large gaps are shown only up to a limit, because the estimate is not trustworthy beyond it; a “≈” in front marks a low-reliability estimate.
- No estimate (shown as “—”) when inputs are missing or cannot be compared reliably, for example when the share price and the company’s filings are in different currencies.
Crisis Risk Index (CRI)
A fragility indicator from 0 to 100 built from macro and market signals. It is not a forecast of when a crash comes and not a probability. The live reading and each signal’s current value are on the Crisis page.
See the live index and its signals →
Levels
The index is read in five levels, from the lowest to the highest fragility:
- Calm
- Normal
- Watch
- High
- Crisis
- A high reading means conditions look fragile, not that a crash will follow. A low reading does not mean calm is certain.
- Inside the calculation model the index can lower the share of risky assets. That is drawdown control, not a return forecast and not an order.
Data and freshness
- Prices, fundamentals and macro statistics come from Nasdaq Data Link (Sharadar), EODHD, FRED (Federal Reserve Bank of St. Louis), the European Central Bank (euro reference exchange rates) and FINRA; the CAPE valuation series from Robert Shiller's public dataset; global news-volume data from GDELT; country indicators from the World Bank; crypto-market data from CoinGecko and Binance; public filings from SEC EDGAR (processed offline); fund facts from the issuers' public factsheets. We only request market data from them; nothing about you is sent.
- When the data pipeline falls behind, the Screener says so in a banner instead of showing old figures silently.
Known gaps
- Some inputs depend on licensed data sources. Where a source is switched off or missing, the score is computed from what remains and marked partial.
- Rankings cover companies listed today. That suits a current ranking, but the list is not used to test history, because it would leave out companies that disappeared.
Limits
- Historically, this kind of stock screen did not beat the index after costs. It helps decide what to study, not what to buy.
- A score compares stocks with each other today. It says nothing certain about where any single stock goes next.
- The rules behind each score are documented choices made by judgment; another reasonable choice would rank some stocks differently.
- Where a figure cannot be calculated, the product shows a dash and the reason, not a guess.
- Past results do not tell what comes next, and every score and estimate can be late, incomplete or wrong.
- Backtests are labelled as backtests; they are not live results.
- Nothing in Allocaria is personal investment advice. Ratings, plans and review items are information to help you decide; the decision is always yours. Some figures in your account, such as “Portfolio fit” on a stock’s page, are calculated from the holdings you entered; they are shown apart from the Allocaria Score and remain information, not a recommendation to you.
- A rating is a rank, not a probability or a price forecast.
- Past performance and backtests do not guarantee future results.
- Ratings are produced by Allocaria's own models. We accept no payment from any company to influence a rating, and hold no financial position in the securities we rate.