I ran the query on a Tuesday afternoon and it returned in nine seconds.
Four hundred and twelve token generation events. Every one listed between Q3 2025 and Q1 2026. Every one still holding a live market capitalization above $40 million at the snapshot date. For each event I pulled seven fields that any serious allocator would treat as minimum viable disclosure: vesting schedule with named cliff dates, team wallet addresses, audit scope plus auditor identity, treasury multisig composition and signing threshold, liquidity lock terms, foundation jurisdiction, and ecosystem grant accounting.
Sixty-one percent of the sample had at least one of those fields empty. Not ambiguous. Not pending. Empty. The registry entry did not exist.
The most predictive variable in the dataset was not a number. It was a blank.
That is not a rhetorical flourish. When I joined the blank-field flag to 90-day post-listing price action, the subgroup carrying two or more missing disclosure fields underperformed the fully-disclosed subgroup by a median of 34 percentage points. The fully-disclosed cohort was not immune to drawdowns. It simply produced fewer catastrophic ones, and it recovered from the ones it produced.
I don't trust the narrative. I trust the immutable ledger. The ledger here was telling me about the things it had been structurally prevented from recording.
The crash wasn't the signal. The blank was.
Why Disclosure Infrastructure Exists and Blanks Persist Anyway
The tooling is not the bottleneck. That is the part most analysts get wrong.
Vesting schedules have had on-chain primitive support since Sablier and Hedgey standardized streaming contracts. Unlock trackers have scraped those contracts for years. Etherscan labeling has matured to the point where a diligent researcher can hand-trace a foundation wallet to a CEX deposit address in under ten minutes. Dune has enough decoded tables that a vesting cliff is a date_trunc away from being visible on a public dashboard.
So when a project's vesting field is blank, it is not because the data is technically unavailable. It is because someone decided not to publish a contract address, or not to label a wallet, or not to answer a form field that a listing platform left optional.
Disclosure in this industry is not a legal category. It is an emergent one โ a social contract enforced by allocator diligence, listing requirements, and reputational cost. That is exactly why blanks are informative. A regulated issuer cannot leave a prospectus field blank without triggering liability. A crypto foundation can leave anything blank, indefinitely, and suffer nothing but a Twitter thread.
I built the dataset in three passes. The first pass was purely mechanical: pull every token event from the listing registries, extract the seven fields, flag nulls. The second pass was label propagation โ take the contract addresses that were published and walk the funding graph outward two hops to see whether the unpublished addresses were discoverable anyway. The third pass was a manual audit of 60 randomly selected entries, because automated null-detection fails on fields that are filled with a placeholder string like "TBA" or "see docs" or a Notion link that 404s.
That last category matters more than the raw nulls. A field filled with a dead link is functionally identical to a blank field, but it is socially cheaper to publish. It reads as compliance without being compliance. Of the 39% apparently "complete" records, 11% contained at least one dead or circular reference. The true blank rate was closer to 65%.
The Evidence Chain
The vesting schedule that was never a schedule
Of the blank vesting fields, the on-chain funding graph revealed something consistent. Within 48 hours of the token generation event, the deployer wallet moved a tranche to a fresh address that had never interacted with the protocol before the listing date. That address then sat dormant.
Dormant is not the same as locked. A dormant address with no stream contract and no vesting wrapper is fully liquid. It just hasn't moved yet.
When I cross-referenced the timing of the first outbound transfer from those dormant addresses against price action, the correlation was uncomfortable. The median first-transfer date landed 11 days before the local top. Across 74 observable cases, that is not noise. That is a schedule. It just wasn't the schedule anyone published.
During DeFi Summer I spent a month modeling Uniswap V2 slippage and MEV extraction, and the lesson that stuck was structural: the interesting behavior is always in the transactions the protocol doesn't advertise, not the ones it does. Vesting is the same problem wearing a different hat. The published schedule is a narrative artifact. The transfer graph is the settlement layer.
The team wallets nobody labeled
This is where the industry's decentralization language diverges hardest from its custody reality.
I tracked 118 projects that describe themselves as "community-governed" or "DAO-managed." For 71 of them, the treasury multisig had at least one signer whose address resolved โ through two funding hops โ to a wallet that also received tokens from the original deployer. That is not decentralization. That is one cap table wearing three hats and calling it governance.
A DAO is a compliance shield when the signer set and the founder set are the same people. I've seen this pattern since 2017, when I was sixteen and manually tracing ETH from ICO wallets to exchange deposit addresses over six months. Sixty percent of the tokens in my sample got dumped by the people who sold the vision. Nothing about the mechanism changed in nine years. Only the vocabulary did.
The labeling gap is where the blank becomes load-bearing. If a treasury address is unlabeled, a tracker will not count it. If it is not counted, it does not appear in the "foundation holdings" column. If it does not appear in the column, the float calculation that every dashboard runs is wrong by however much that address holds.
I ran that correction on 22 projects. The median adjusted float was 19% higher than the published figure. Two of them were more than 40% higher. Every downstream valuation model built on the published number was pricing a different asset than the one that existed.
The audit that audited the wrong thing
An audit is a scope document with a logo attached. The industry treats it as a verdict.
Of the 412 events, 289 claimed an audit. Of those, I could retrieve a public report for 231. Of those 231, the report's stated scope covered the token contract and the staking module only โ not the governance executor, not the upgrade proxy, not the treasury withdrawal path โ in 166 cases.
That is 72% of audited projects with a material unaudited attack surface, and the unaudited surface is invariably the one holding the money. A staking contract is where users deposit. The upgrade proxy is where someone can redirect the deposit.
I've made this argument before and I'll keep making it: the presence of an audit is a marketing fact. The scope of the audit is a risk fact. When a diligence form asks "audited? yes/no," it is collecting the marketing fact and discarding the risk fact. A blank audit-scope field is not a missing document. It is a missing document about the part of the system that matters.
The treasury with no signers on record
Multisig composition was the second-most-blank field after grant accounting. Twenty-six percent of projects with a stated treasury did not publish the signer set or the threshold.
An unpublished threshold is not a neutral omission. It means an allocator cannot answer the only question that matters for custodial risk: how many people must collude to drain this, and do those people have a legal identity?
When I could reconstruct signer sets from on-chain proposal history, the distribution was grim. The median threshold across 94 reconstructable treasuries was 2-of-4. A 2-of-4 with two signers sharing an employer, a jurisdiction, and a funding source is functionally a 1-of-1 with extra steps.
The liquidity lock and the subsidy arithmetic
Every project in the sample claimed locked liquidity. Forty-two percent published the lock contract address. The rest published a screenshot, a tweet, or nothing.
A screenshot is not a contract. A contract with a 30-day unlock is not a lock. I found 19 cases where the lock contract existed, was verified, and contained a withdraw function callable by the deployer with no timelock. The liquidity was locked in the sense that a door with the key taped to it is locked.
This connects to something I have tracked since 2020: the incentive structure that produces these numbers. Liquidity mining APY is not yield. It is a project paying to rent a number on a dashboard, and the number leaves the moment the payment stops. Unlock a liquidity position early, and the rent evaporates along with the pool depth that justified the listing.
When the lock terms are blank, the subsidy has no end date. That is not a feature. That is an open liability on the protocol's balance sheet that nobody has been asked to mark.
The grant accounting that never closes
Grant and ecosystem fund accounting was blank in 78% of cases. This is the least glamorous field and the one with the clearest signal.
An ecosystem fund that publishes inflows but not outflows is a wallet with a narrative. I traced 31 such funds. The median outflow velocity was 0.7% of committed capital per quarter โ a rounding error against the grants announced publicly. Announced grants are a press release. Disbursed grants are a transaction. The gap between them is the story.
Correlation Is Not Causation, and That Matters Here
The lazy reading of everything above is "blank fields cause collapses." That is wrong, and it is the kind of wrong that gets people liquidated.
Blanks do not cause anything. They are a symptom of the same underlying condition that produces the collapse: an issuer that has calculated the cost of disclosure and found it higher than the cost of ambiguity.
The decision to leave a vesting field blank is not an oversight. It is a capital allocation decision made by someone who understands that publishing a cliff date invites selling pressure while hiding it merely risks a blog post. The blank and the eventual drawdown share a parent, which is a team optimizing for float management rather than for holders.
There are legitimate blanks too, and I want to be precise about them. A project with a genuinely adversarial regulatory posture may withhold foundation jurisdiction because naming a legal entity creates enforcement surface. That is a real tradeoff, and I don't think it should be scored the same as an unnamed vesting schedule. In my sample, roughly one in six blanks fell into this defensible category, and I flagged them separately. The gap in performance between "defensible blank" and "deliberate blank" was almost as large as the gap between "complete" and "blank."
There is a second counter-intuitive result. Complete disclosure does not correlate with outperformance; it correlates with lower tail risk. The fully-disclosed median return was essentially flat against the sample. What it lacked was the left tail. That distinction matters for anyone building a portfolio rather than picking a lottery ticket.
And there is a third thing, which is the part the industry keeps missing. The obsession with more data is itself a failure mode. Ten dashboards with the same three fields do not produce diligence. They produce the feeling of diligence. What produces diligence is knowing which fields are absent, why they are absent, and who benefits from the absence.
I watch this problem play out in Layer 2 as well. The competitive axis between OP Stack and ZK Stack has never really been the cryptography. It is who convinces more projects to deploy chains first. Distribution wins. And distribution competition creates a race to lower the disclosure bar, because the chain that asks fewer questions wins more deployments. Rollups are the venue, but the same incentive that produces blank vesting fields produces blank deployment standards.
The Signal to Watch Next Quarter
Here is what I would track, and what I am tracking.
Compute a disclosure density ratio for every new listing: published fields divided by seven, adjusted for dead links. Anything below 0.57 โ four of seven fields โ goes on a watchlist rather than a portfolio.
Then track the delta. A project that fills a blank field after listing is telling you something. A project that fills a blank field after a drawdown is telling you something different, and it is usually the more expensive lesson.
The number I am most interested in for Q2 is not price. It is the blank rate itself. If the aggregateblank rate falls below 50% for new listings, the market is pricing disclosure correctly and the edge in this dataset closes. If it rises above 70%, we are in a period where nobody is reading the form, and the marginal value of reading it goes up, not down.
Data doesn't lie. It goes missing, and someone decided where.
The question worth sitting with is not whether your favorite project disclosed its cliff schedule. It is what the missing schedule was designed to let you not see โ and whether you would have bought anyway if the field had been filled in.