Crypto data is noisier than most readers assume. Exchange order books can be spoofed, reported trading volumes are routinely inflated by wash trading, and a single misconfigured oracle can send a headline price wildly off from what anyone actually paid. Our methodology exists so readers can see exactly how a number in an OBOL article got there, and so we have a standard to be held to when we get it wrong.
Price and market data
Prices and market figures on OBOL come from CoinGecko's aggregated market API, cached on our servers and refreshed on a short interval — the same source that powers the charts, the directory and the asset pages. We do not run our own price index, we do not aggregate venue order books, and we do not filter individual exchanges out of the aggregate. That is worth stating plainly: an aggregated price inherits whatever wash trading sits in the underlying venues, so treat any figure here as indicative rather than as a settlement-grade benchmark, and read the primary source before acting on a number.
On-chain verification
Where an article makes a claim about wallet activity, protocol TVL, exploit mechanics or token flows, that claim is checked against a block explorer or the protocol's own contracts rather than taken from a project dashboard. We do not operate indexing infrastructure of our own, so anything we cannot verify from public chain data or a primary document is either attributed to the source we took it from or left out.
How the ratings are scored
Every service in our ratings section is marked against a published rubric for its category. Each category has five or six criteria, each carries a stated weight, and the weights sum to 100. Analysts score each criterion from 0 to 10 against documented evidence, and the headline OBOL score is the weighted mean of those marks, rounded to one decimal place. Nobody types the headline number in by hand. If you disagree with a rating, you can see precisely which criterion produced it and argue with that instead of with a verdict.
The weights are not uniform across categories, because the risks are not. Security carries the most weight for exchanges and bridges, reserve quality for stablecoins, and true all-in cost for cards and instant exchangers. The rubric for each category, including what evidence moves each criterion, is printed at the top of that category's page.
Some categories go further and publish the source of every individual mark: the URL the evidence was taken from and the date it was checked. In those categories a criterion with nothing on the public record scores 0 and is labelled "not published" rather than estimated. Their rubrics are printed in full at the end of this page.
What a score means
9.0 and above is exceptional: best in category on the evidence we can verify. 8.0 to 8.9 is strong and recommendable to most readers with the stated caveats. 7.0 to 7.9 is solid for a specific use case and weaker outside it. 6.0 to 6.9 is mixed — usable, but there are better options for most people. Below 6.0 means real drawbacks, and below 5.0 means we would not put our own funds through it. Scores are comparable within a category and not across categories, because the rubrics differ.
What the evidence is, and what it is not
Ratings are built from public evidence: audit reports, enforcement filings, proof-of-reserve and attestation documents, published fee schedules, on-chain data, incident post-mortems and the protocol's own contracts. Where we depend on a third-party dashboard we say so in the review. This is desk research, not undercover testing: we do not audit anyone's books, and we cannot confirm what a company holds beyond what it publishes and what the chain shows. A rating is a documented judgement, not a certification.
Independence and conflicts
No company can pay for a rating, a review, a position in a ranking or a change to a score, and we accept no in-kind consideration from the services we cover. If a page ever carries an affiliate link it will be disclosed on that page and it will have no bearing on the rubric. Where OBOL staff hold a position in an asset or service under review, that is disclosed in the review itself.
Review cadence and corrections
Each category is re-scored quarterly, and immediately after any material event: an exploit, an enforcement action, a withdrawal halt, a licence change or a change to fees. Every review carries the month its evidence was last checked. When a score changes materially we note what moved and why. When we get something wrong, we correct the review and log the change rather than quietly editing the number.
Corrections to data
If a methodology error changes a figure that has already been published, we issue a correction on the original article and note the change in our corrections log. Persistent or structural data issues are written up separately so readers understand not just what changed but why our process missed it. Questions on sourcing or methodology can be sent to [email protected].
Rubric: Best DeFi strategy and protocol economics advisors
This is the complete rubric for the Best DeFi Strategy & Protocol Economics Advisors rating. Each firm is marked from 0 to 10 on every criterion below, and every mark on a review page carries the URL it was taken from and the date it was checked.
| Criterion | Weight | How the mark is set | Source |
|---|---|---|---|
| Published research and models | 25% | 0: nothing published. 5: a blog without models. 10: open models, simulations and papers. | Firm website, GitHub, arXiv, SSRN |
| Client-confirmed work | 20% | The share of the firm's named case studies that the client itself confirms, converted to 0–10. | Client websites, governance forums |
| Recent economics work | 20% | Date of the firm's most recent dated economics work for a client. 10: within 12 months. 7: within 24. 3: within 36. 0: older or undated. Blog posts and security-only audits do not count. | Client reports, governance forums, case studies |
| Conflict-of-interest disclosure | 15% | 10: a published policy, with payment in tokens disclosed. 5: partial disclosure. 0: nothing. | Firm website, terms |
| Named team | 10% | Named people whose profiles can be checked. | Firm website, LinkedIn |
| Transparent terms | 10% | Engagement formats, timelines or a price guide are published. | Firm website |
Formula.Score = Σ (mark × weight), rounded to one decimal place. A criterion with no published evidence receives 0 and is shown as “not published”. The score is calculated from the marks and is never entered by hand.
Equal scores. The firm with the higher published research and models mark ranks first.
Revision. A firm is re-scored when something material happens, not on a fixed calendar. The date of the last check is printed on every review, next to each mark.
Inclusion. Firms are included only when their website is live, they are operating, and their main service matches the category definition. A firm we cannot verify is held back in draft and does not appear in the rating. No firm can pay for inclusion, a mark or a position.
Last updated 22 September 2026