StakeBible publishes long-term price ranges for crypto-assets, explains how each one is built, and grades them against the real price when their dates arrive. This page sets out who is behind the site, how the work is done and where its limits are.
Who we are
StakeBible is operated by PREDICTIVE CAPITAL AI LABS S.R.L., a Romanian limited liability company registered with the National Trade Register Office (ONRC) under number J2026056423004 (tax identification code 55656798). StakeBible was founded by George-Eleodor.
The operator's details are also on our contact page.
What we publish, and what we don't
We publish:
- Long-term forecasts. Each one is a central path with a band around it, in 90-day steps up to about ten years ahead, with a confidence score where we can support one.
- Research on projects. A score from 0 to 1000, strengths, risks and red flags, with a link to the source wherever one can be verified.
- Market data and a planning tool. Prices for the coins and exchanges we track, and a tool that projects the investment plan you describe.
We do not publish:
- Investment advice. A forecast is a statistical estimate, not a recommendation to buy, sell or hold. We know nothing about your situation.
- Short-term signals. There are no daily targets and no trading calls. Our models read market cycles, and a cycle says nothing useful about next week.
- Forecasts for assets a cycle model cannot describe. Stablecoins, fiat currencies and tokenised real-world assets such as gold or shares are tracked for price but never forecast.
We never hold your funds, never ask for exchange keys and place no orders.
How the forecasts are made
Our price history comes from daily closing prices on four exchanges (Bitstamp, Gate.io, Binance and OKX), with CoinGecko's aggregate price as a fallback for coins none of them lists. Bitcoin is modelled on its own, as a long-term trend combined with its halving cycle; Ethereum and every other coin are forecast relative to Bitcoin. The width of each band comes from testing the models on data they had not seen, and every published forecast can be recomputed from our database and the parameters saved with its run. The full method, with its limits, is on our methodology page.
How we use AI
- Statistical models compute the forecasts. Every price, return and band on StakeBible comes from models fitted to the price history in our own database.
- Language models research projects. A language model with web search reads public sources about a project and returns a score from 0 to 1000, strengths and risks with their sources, and any red flags. Its answer format has no field for a price. The research text on coin pages is written by that model, not by an editor, and each page names the model and the date.
- The research score limits what we claim. It caps the confidence we publish for a coin, and a very low score marks a project as dead or abandoned, in which case we publish no forecast for it. It does not change the forecast itself, and it is not a reason to buy: in a blind test on coins whose outcome we already knew, it did not predict prices.
- Language models help in three other places. They judge which small coins enter the High Potential selection, read two dates from public sources (the next US Federal Reserve meeting and the estimated date of the next Bitcoin halving), and draft some of the site's translations.
- No language model writes a price. No price, return or band on StakeBible is written by one.
- Nothing about you reaches a language model. We send them questions about crypto projects and public data only.
More detail is in the methodology section on language models.
Editorial policy
- Independence. No project, exchange or advertiser pays to appear on StakeBible, to be scored or to enter the High Potential selection. We carry no advertising, no sponsored content and no affiliate links.
- How sources are chosen. Exchanges are read in a fixed order of priority, set by how deep and complete their price history is. Research relies on public sources found by web search. Where a cited source can be verified, the page links to it; where it cannot — some search results expire or point to a site's front page — the claim is marked as having no verifiable source, and red flags are summaries that carry no link of their own.
- Projects in the news. When we write about a project that is in the news, press coverage only tells us which project people are talking about. What we write is based on what the project itself published. A person decides whether each article is published.
- Language. Our pages are in English, French and Spanish. The research prose (strengths, risks, red flags and the High Potential dossiers) comes from web research and stays in English on French and Spanish pages, and those pages say so. It is rewritten at every research run, so a translation would always lag behind it.
Corrections policy
If you think something on StakeBible is wrong (a fact, a figure, a source or a translation), write to support@stakebible.com with the address of the page and what you believe is wrong. We check every report. When we find an error, we correct it, and when the correction changes the meaning of a page, we say so on that page, with the date.
We don't edit individual forecasts by hand. If a forecast rests on bad data, we fix the data at its source and the forecast is recomputed at the next model run.
Accuracy
Every forecast is graded when its date arrives: we compare it with the real closing price and record whether the price landed inside the band. Grades are kept per model version, so a new model never inherits an old one's record, and forecasts that cannot be graded (a coin delisted, a gap in the data) are counted rather than dropped. We publish no accuracy figure until graded forecasts exist. The current status is in the accuracy section of our methodology.
Contact
Write to support@stakebible.com. Our contact page explains what to include. How we handle personal data is set out in our privacy policy, and the rules for using the site are in our terms of use.