Universe membership is a measurement boundary, not an endorsement: a coin's presence on our board says it met published liquidity and age rules, nothing more. Nothing here is investment advice. Decisions are yours.
Every ranking hides its most consequential decision in the guest list: which assets were allowed to compete at all. Most platforms treat universe selection as backstage machinery. We publish ours as rules with reasons, because a board whose membership criteria are secret can be gamed by insiders, argued by projects, and never audited by readers, and all three failure modes are routine in this market.
Two universes, on purpose
This site runs two deliberately different universes, and conflating them is the original sin our data work exists to prevent. The monitor universe answers "what does today's liquid market look like": the top 100 assets by 30-day median dollar volume, refreshed monthly, current listings by definition. The research universe answers "what was the market at every historical date": reconstructed point-in-time from dead-inclusive data, top 150 by the same liquidity measure as of each date, so a 2021 backtest competes against 2021's actual roster including every name that later died. The monitor universe would be survivorship malpractice if used for backtesting; the research universe would be pointless overhead for a live board. Each is fit for exactly one purpose, and the split is documented wherever either is used, because the dead names that fill the crypto graveyard appear in only one of the two.
The rules, each with its reason
Liquidity: 30-day median dollar volume, vetted venues only. Median because averages are one wash-spike from capture; vetted venues because wash trading makes reported volume from unaudited exchanges majority-fiction. Liquidity ranks the universe because it is the property that determines whether a reader could actually transact, and because it is the hardest metric to fake on venues with consequences.
Stablecoins excluded. A momentum rank on an asset engineered not to move is noise wearing a row. Their one analytically interesting behavior, depegging, is a risk event we cover in research, not a trend the board should rank.
Wrapped, pegged, and staked-derivative assets excluded. They duplicate their underlying: ranking both an asset and its wrapper double-counts one trend and wastes a board slot that could carry information.
Exchange leveraged tokens excluded. Their daily-rebalance decay mechanics make every daily formula on the board, momentum especially, systematically misleading. No fair measurement of them exists at our cadence.
Listing age of at least 90 days. The first weeks after a listing are dominated by allocation unlocks, airdrop rotation, and promotional volume, the launch mechanics this market is famous for. Momentum arithmetic needs history to mean anything, and the manufactured-volume problem is at maximum intensity exactly when a token most wants onto screens. Ninety days is a published, arguable choice; that it is published and arguable is the point.
Top 100 for the monitor, top 150 for research. The monitor cut keeps the board legible and its data auditable venue by venue. Research runs wider because strategies must be tested against the candidates they would actually have faced, including the marginal ones.
The edge-case ledger
Real markets produce cases rules must handle explicitly, so ours are written down. Rebrands and ticker changes: the asset keeps its identity and history under the new symbol, continuity documented. Chain migrations and token swaps: treated as the same asset when the swap is one-to-one and holder-continuous, otherwise as a delisting plus a new listing, the LUNA case being the canonical split. Stablecoin depegs: stablecoins stay excluded, but a depeg event enters research data as what it is, a market-wide risk shock. Delistings from our vetted venues: the asset exits the monitor at the next monthly refresh, and in research data its history ends there rather than being deleted, because removing the corpse is exactly the bias we built this pipeline to avoid. Every edge case resolves toward the same principle: identity follows holders, history is never rewritten, and exits are recorded rather than erased.
Change policy and the anti-gaming argument
Universe rules change rarely, loudly, and forward-only: announced with effective dates, logged quarterly in the universe update series, never applied retroactively, and visible in the attestation chain because membership is part of the daily snapshot that commit-reveal attestation locks in. Publishing selection rules invites one obvious objection: does it not help projects game their way onto the board? Our answer is that the rules were chosen so that gaming them requires becoming the real thing. Sustained median liquidity on vetted venues for months is not a hack; it is a market. Meanwhile the alternative, secret criteria, produces the familiar pathologies: listing-by-relationship, quiet removals, and boards whose membership is a negotiation. A gameable-only-by-honesty rule set, published, is the best defense we know, and it has the side benefit that when someone disagrees with a boundary, the argument can be had in public with the rule on the table. That is what methodology is for. Decisions are yours.