Most crypto lending sites rate almost everything between 4 and 5 stars. We did too, until recently. But when every lender scores well, the score stops carrying information and starts working as decoration.

So we reevaluated how we rated and scored every product on the site against a new rubric, including 11 ratings across lending and yield. Not one of them held its old number. We published the result anyway, because the alternative was leaving scores up that we could no longer defend. Here is exactly how the numbers are built, so you can check our work or disagree with it.

We do not grade on a curve. A 5 means a lender with essentially no unmitigated risk, not simply the best of a weak field.

What Each Score Means

A lender is measured against what a genuinely safe, transparent lending product would look like, not against its competitors.

A lender scoring in the 2s is not necessarily a fraud, and we will still tell you what it does well. It means the gap between what that company publishes and what you would need to know before handing over your Bitcoin is wide enough that we cannot recommend it.

The 6 Things We Score

Every lender is scored out of 100 points across six categories, then divided by 20 to give the star rating. The weights reflect what actually costs people money when a lender fails.

Collateral safety carries the most weight for a simple reason. In 2022, the lenders that destroyed their customers did not fail because their rates were uncompetitive. They failed because they had lent out collateral that customers believed was sitting untouched.

How We Score Interest and Yield Accounts

Lending and earning are different products with different ways of going wrong, so they get different rubrics. When you take a loan, you keep ownership of your Bitcoin and the question is what the lender may do with the collateral. When you deposit for yield, you hand the assets over completely, and three different questions matter.

1. Do your deposits get lent out to somebody else, or do they sit where you put them?
2. Where does the yield actually come from, and is that source disclosed and economically credible?
3. Is the advertised rate one a normal customer can get, or does it need a paid plan, a lockup, or a large holding of the company’s own token?

So asset safety replaces collateral safety at the same 30 points, and the liquidation mechanics that matter on a loan are replaced by a 20-point yield legitimacy axis. The other four categories carry the same weights as the lending rubric.

A yield of 10% funded by lending your coins to strangers is a different product from a yield of 3% paid out of Treasury interest, even when the marketing page shows them the same way. The rubric is built to separate them.

Two things we watch for in particular: 1) Rates quoted before the platform’s commission is deducted, which can take 25% to 30% of what you earn. 2) Rewards paid in a company’s own token rather than in the asset you deposited, which quietly hands you a second investment you did not ask for.

How We Handle What We Cannot Verify

Sometimes a fact we need is not published anywhere we can find. We handle that in a specific way rather than guessing.

• If we cannot verify a claim, we do not score it as a zero. A zero is a finding, and an absence of evidence is not a finding.
• We remove that category from both the points earned and the points available, so an unknown does not become a penalty.
• We say in the review which facts we could not establish.

The reverse also applies. We do not award points for claims a company makes about itself unless something independent backs them up. Phrases like “perfect track record” or “bank grade security” earn nothing on their own.

A lender that publishes nothing will not score well here, even if nothing bad has ever happened to it. Being unverifiable is itself a risk you take on as a borrower.

What Our Commercial Relationships Do to the Score

Nothing, and we can now prove it rather than just promising it. Some links on this site are affiliate links, meaning we earn a commission if you sign up. Those relationships are disclosed and they have no input into the rubric or the scores.

The clearest evidence is that after our re-rating, the only lender we currently earn commission from is the lowest rated lender on the site. We kept the relationship, because it is a real product some readers will still want, and we put the rating and the reasons directly beside it. A rating system that conveniently favoured whoever paid us would be worth nothing to you, and would not survive anyone checking it themselves.

How We Produce This Research

We use AI tools in our research and article drafting. We think you should know that, and we think the more useful thing to tell you is what happens afterward, because that is where the value actually comes from.

• Every factual claim is checked against a primary source, which usually means the lender’s own terms, help center, or pricing page rather than another publication’s summary of them.
• Every outbound link is checked before publishing. This is tedious and it is also how we have caught dead sources and wrong URLs that would otherwise have been published.
• A named human editor reviews the piece, and our editor regularly sends work back.
• When we get something wrong, we correct it on the page and say what changed rather than quietly editing the number.

That last point is not hypothetical. While building this rubric we found several dated claims from articles written in years past, which are now fixed and noted.

A tool does not make research accurate or inaccurate. Checking does. We would rather be judged on whether our facts hold up than on which software was open while we wrote.

One limitation worth stating plainly: We research these products through their documentation, their regulatory records, and the experiences users report publicly. Where we have not borrowed from a lender ourselves, we do not pretend otherwise, and you should weigh our findings accordingly.

Where the Evidence Comes From

We score from primary sources wherever they exist. In practice that means 5 kinds of evidence.

1. The lender’s own terms, help center articles, and pricing pages, quoted directly where the wording matters.
2. Attestation and proof of reserves reports, including who signed them and what they actually cover.
3. Regulator records, court filings, and enforcement actions.
4. Review platforms, weighted by volume rather than headline score.
5. Contemporary reporting from the period being described, rather than a company’s later account of itself.

Every review lists its sources at the bottom, and we check that each link still resolves before publishing.

When We Re-Check

Crypto lending terms move quickly, and a review that was accurate last year can be very wrong this year. Each review carries the month its figures were checked, near the bottom of the page above the sources.

We re-score a lender when its terms change materially, when a regulator acts, or at least once a year. If a score changes, the review changes with it, and we say what moved and why rather than quietly editing the number.

If you think we have a fact wrong, tell us. We would rather correct a review than defend it.