Someone in a group chat I am in has been averaging down on the same token for over a year now. The project's last blog post is old enough that the roadmap in it promises an ecosystem fund that never made a single grant. The Discord has maybe a dozen active users and one of them is the moderation bot. And still the token trades every day, respectable volume, spreads tight enough that his fills look completely normal. He reads that as a pulse. I keep trying to explain that trading activity is roughly the last thing to stop when a project dies, and that by the time volume actually goes to zero, the answer was knowable years earlier.
The mechanics of why are worth understanding, because they explain why zombie tokens are everywhere. An AMM pool needs no permission and no team involvement to keep quoting prices. Arb bots will keep that pool aligned with whatever centralized listings remain, indefinitely, on autopilot. Exchanges delist slowly, because delisting is admin work with no revenue attached. So a token can keep printing candles for years with zero human beings working on the underlying project. Price and volume carry almost no information about whether the thing is alive, which means you have to look at everything except the chart.
The six checks I run
None of these is decisive on its own, and each one has a known way of lying to you. Together they are hard to fake, because faking all six costs about as much effort as actually running the project.
- Commit history. Find the core protocol repos rather than the docs or website repo, which teams update out of habit long after real work stops. Roughly six months without a substantive commit is a serious flag. Two caveats. Dependency-bump bots can make a dead repo look busy, so read what the commits actually change. And some teams take development private, so check whether tagged releases still appear even when daily commits do not.
- Governance. Look for the last proposal that passed and was actually executed on-chain, rather than the last forum post. Two or more quarters with no proposal originating from anyone connected to the core team is a flag. Dead projects tend to have governance forums where the newest threads are holders asking where the team went, with no official reply anywhere in them.
- Treasury outflows. The strongest single check, because payroll does not lie. A live project has a salary rhythm, regular outflows from the treasury or team multisig to a consistent set of addresses, or to an exchange for fiat conversion, typically monthly or biweekly. Two bad patterns to watch for. The outflows stop entirely, which means nobody is being paid, so nobody is working. Or there is one large sweep to an exchange followed by quiet, which usually means the team cashed out on the way to the exits.
- Social cadence. Ignore follower counts, they are a lagging artifact of the good times. Look at the gap since the last substantive post and at what the posts contain. Ninety days of silence from an account that used to post weekly is a flag. So is a feed that has degraded into pure retweets and vague "big things coming" posts, which usually means one part-time community manager is the last person on the payroll, if anyone is.
- Team departures. Search the project on LinkedIn and look at who lists it as a past position. Founders keep the title on their profile for years, so ignore them. Engineers update within a few months of actually leaving, because they need the profile to get hired somewhere else. If most of the identifiable engineering hires now show a different current employer, development has stopped no matter what the official channels say.
- TVL and usage decay. Separate mechanical decline from real exit. If TVL fell because the token price fell, that is just arithmetic. Measure usage in units that do not depend on price: unique active addresses, transaction counts on the core contracts, protocol revenue, TVL denominated in the deposited asset. A steady bleed in those, with no visible response from the team, means the users who understood the project best left first.
Dormant is a real category, and it looks different
I want to be fair to hibernating projects, because they exist and a few of them have come back and rewarded the people who held through the quiet stretch. Teams cut marketing to zero in bad markets and keep building. The distinction shows up in which checks fail. A dormant project goes silent on social and lets governance idle, but the treasury keeps paying salaries and the commits keep landing, slower maybe, but substantive. A dead project fails across the board at the same time.
My rough scoring, and I hold it loosely. Zero to two flags, dormancy is plausible, keep watching. Three or four flags, assume dead unless you can find specific contrary evidence, like a funded team wallet that is still paying developers. Five or six flags, it is done, and anything you hold is a lottery ticket on someone eventually reviving the brand, which does happen occasionally, but you should price it as exactly that and nothing more.
There is also a softer tell I trust more than I probably should. Live teams dispute things. Post something wrong about an active project and someone with a team tag corrects you within a day. Dead projects have a distinctive non-response, where someone asks "is this dead" in the Discord and the question just sits there for weeks. Silence in the face of that specific question is usually an answer in itself.
Run the checklist before you average down
The whole exercise takes about an hour for a project you already hold, and the order matters a little. I start with the treasury, because on-chain flows cannot be spun, then commits, then LinkedIn, and only then the softer social checks, so the vibes get calibrated by the hard data instead of the other way around.
The most common cope I hear is that the team is quiet because they are heads down building. Heads-down teams still pay salaries and still push code, and both of those are publicly visible, one on-chain and one on GitHub. Quiet plus payroll plus commits really is heads down, and I have seen it resolve well. Quiet plus nothing else is gone.
And on averaging down specifically. Buying more of a dead project lowers your average entry into an asset whose only realistic exit is a nostalgia pump you cannot schedule. If the checklist says dead and you still want exposure to the revival scenario, fine, but size it like the lottery ticket it is and stop treating it as a position you are actively managing. I keep a dated note with the six checks for anything I hold that has gone quiet, and if a project fails the list twice, roughly six months apart, I harvest the loss and stop reopening the question.