Research Sources Worth Your Time

Most crypto research is a rewrite of something else. The rewrite is faster to read, arrives more often, and carries none of the qualifications the original had. If you want to know something rather than to feel informed, you have to work back toward primary sources, and the good news is that crypto has unusually good primary sources sitting in plain sight.

What are the actual categories of source?

Six, and they answer different questions. Confusing which question a source can answer is the commonest mistake.

On-chain data. Block explorers such as Etherscan and their equivalents on other chains give you supply, holder distribution, contract permissions, upgrade authority, timelock queues and treasury flows. This is the only category where you read the state itself rather than someone's account of it, which makes it unfalsifiable at the data layer. The interpretation layer is a different matter: address labelling is largely heuristic, and a dashboard's definition of an active user is a judgement call someone made and rarely documented.

Governance forums and voting records. Discourse forums attached to most protocols of any size, plus off-chain voting on platforms like Snapshot and the on-chain governor contracts themselves. This is the highest-value, least-read category in the sector. Proposals get drafted, contested and revised in public before anything is announced, and the objections are written by people with exposure.

Developer repositories. Tagged releases, issue triage, and which repositories are actually moving. Useful mainly as a negative check. A core protocol repository with no release in eight months is a finding; a high commit count is nearly meaningless, since it can be documentation, dependency updates or padding.

Independent analysts. Individuals and small outfits publishing with a visible track record. The good ones show their queries, state their positions, and post when they were wrong.

Official documentation. Docs sites, audit reports naming the auditor and the version audited, deployed contract addresses on the project's own domain. Authoritative on how something works, useless on whether it is worth owning, and those two things get conflated constantly.

Primary legal and regulatory documents. Court dockets, agency filings, published rulemaking and consultation responses. Slow and dull to read, and the gap between what a filing says and what gets reported about it is one of the widest in the sector. If a story turns on a legal claim, the document is usually a search away.

How do I tell a good source from a well-designed one?

Production values are not a signal. Plenty of the most useful work in crypto is published in an ugly blog with no logo, and plenty of well-funded research is public data with a chart library on top.

Six questions do most of the sorting. Does the source show its work, meaning could you reproduce the number from something public? Does it disclose funding and positions, and does the disclosure sit where you will see it rather than in a footer? Does it correct itself visibly, with corrections dated and attached to the original piece rather than quietly edited in? Does it distinguish between what it knows, what it inferred and what it is guessing, or does everything arrive in the same confident register? Does it publish only when it has something, or is it committed to a daily cadence that guarantees filler? And who pays, with the follow-up question of what happens to coverage of the people who pay.

That last one matters more here than in most sectors, because crypto publishers are largely advertiser-funded and the advertisers are drawn from exactly the population being covered. The tells for when that relationship has affected an individual article are in telling a real headline from a paid placement.

The single strongest signal is the corrections record. It is the only property on the list that cannot be manufactured retrospectively. Read what someone published when they turned out to be wrong, and you will learn more about their reliability than from a hundred pieces where they were right.

What do I do when a source I relied on shuts down or paywalls?

Assume in advance that it will, because the base rate is high.

Crypto research is expensive to produce and difficult to fund. Independent outfits get acquired and reorient toward enterprise clients. Consumer-facing research products get sunset when the strategy changes. Newsletters go behind subscriptions. Free API tiers get cut or rate-limited into uselessness. Grant-funded dashboards go dark when the grant ends, and some of them keep serving stale data for months first, which is worse than going down. Anything free is a subsidy that somebody has decided to pay, and decisions get revisited.

When it happens, the first question is whether you were using the source's data or its judgement, because the two have completely different replacement costs.

Data usually substitutes, particularly if the underlying facts came from a public chain. A dashboard that told you daily active addresses was running a query you can run elsewhere. What you lose is convenience and the specific methodology, which is why it is worth recording the definition alongside the number when you first rely on it: not "300k daily actives" but "300k addresses with at least one non-transfer interaction, per this source's definition." That note is what lets you rebuild the metric somewhere else instead of silently comparing two different things.

Judgement does not substitute. If your confidence in a position rested on one analyst's read, and that analyst has stopped publishing, the correct response is to reduce your confidence rather than to find a replacement who says similar things. Substituting an agreeable voice for a lost one is how people end up with the same conviction and none of the original reasoning.

Two habits make this cheaper. Keep your own records rather than links, since a link to a piece that no longer exists is a citation you cannot check, and exporting your own data periodically means your position and cost history do not depend on any third party's continuity. And prefer sources whose primary inputs are public, because reproducibility is what makes a source replaceable.

How many sources is enough?

Two or three you read closely, plus the primary sources for whatever you actually hold. More than that and you are back to skimming, which is where a feed that is mostly noise started.

Concentration risk applies to sources exactly as it applies to holdings. The failure is not having too few, it is depending completely on one you neither pay for nor can rebuild. If a single source disappearing tomorrow would leave you unable to answer a question you answer regularly, that is the gap to close, and closing it usually means learning to pull one number yourself rather than subscribing to a second opinion.

The durable skill is reading a block explorer and a governance forum without an intermediary. Both are free, neither can be sunset by a business decision, and between them they answer most of what a five-minute fundamentals check needs. Everything else is convenience, and convenience is exactly the thing that gets discontinued.