How We Score Trust in a Crypto Project, From 0 to 1000

Every coin on StakeBible carries a trust score from 0 to 1000. It decides whether we forecast the coin at all, and it limits how confident we let that forecast sound. This article explains where the number comes from and where it stops.

Why 0 to 1000

We cover thousands of coins. On a scale of 0 to 100, hundreds of them would share the same score, and a tie tells you nothing about which of two projects the research trusts more. A scale of 0 to 1000 keeps them apart.

How the score is produced

The score comes from grounded research: a language model that searches the web, reads public sources about the project and answers a fixed set of questions about it. It returns:

• a trust score from 0 to 1000, and how well documented it considers its own judgement;
• strengths and risks, each with the source it came from;
• red flags, when it finds any;
• a short rationale that ties them together.

We give it facts we compute ourselves: the coin's rank and market capitalisation, its returns and volatility from our own price history, and how long that history is. We never ask it for a price or a return. The format it answers in has no field for one, so it cannot write a number we could not reproduce from our own data.

Scores are added, not overwritten. When a project is researched again, the new score sits next to the old one, so we can always see what we said and when.

What the score controls

Eligibility. Below 150, the research has found an abandoned or dead project. We publish no forecast for it. In a blind test on coins whose outcome we already knew, every abandoned project landed below 150.

How confident we sound. The confidence we publish next to a forecast starts from the trust score and can only go down from there, never up. A project the research trusts at 600 cannot carry a forecast confidence above six out of ten, however tidy the model's band looks.

What it does not do

It does not predict the price. In the same blind test, a decentralised exchange whose token went on to lose more than 99.95% against Bitcoin scored 720, above four of the five winners. The project was alive and working; the token collapsed anyway. The research was asked to judge the project, not the chart, and that is what it did.

That is why the score is a gate and a cap, and never a signal. On StakeBible it never raises a forecast, never ranks coins by expected return, and never appears as a reason to buy.

It is not a safety rating either. It says nothing about the exchange where you hold a coin or about whether your funds are safe there.

Where you see it

The research behind the score, with its sources, is on each coin's page. The projects where it found the least to trust are listed in crypto red flags.