Selection & Monitoring

What this page is not

Observation eligibility is not a rating, not a recommendation and not a safety statement. It says only that a market carries enough verified evidence to be worth the cost of watching. It is separate from Melega Score (project quality, melega-score-v1) and completely separate from sponsored visibility, which can never buy admission, a score or a ranking position.

Policy version space-observation-v1. A newly discovered pool is admitted to observation when it presents at least USD 2,000 of verified pool liquidity, inclusive. Market capitalisation, fully diluted value, protocol TVL and raw concentrated-liquidity parameters are not substitutes and are never used for this test.

The measurement must come from an implemented market-measurement writer and must be no older than 60 minutes. A pool-existence record with no measured market columns, discovery time, import time and the current clock are not measurement evidence.

Exact scope today: this filter is bound in one place — the registration of new EVM pools into the pool-indexer queue. A candidate rejected there never becomes a queued pool, so it costs no per-pool log indexing, backfill or candle work. Pools already registered, and the other refresh and enrichment lanes, are not filtered by this change; reconciling them is separate, pending work.

Missing, blank, non-numeric, infinite or negative liquidity, absent provenance, and invalid, future or stale timestamps all produce a named deferral — never a zero, never "unsafe", never a rating. A provider error, timeout, omission or partial response is degraded data, not a drained pool.

Deferral is not terminal. The identity stays in the canonical backlog and is reconsidered whenever its evidence improves. Technical outcomes (unsupported chain, unsupported protocol, invalid address) remain explicit and separate.

There is no universal 24-hour volume requirement for admission: a genuinely new pool has no history yet. A qualifying pool with measured-zero or unknown volume can enter bounded observation. That grants nothing else — it does not create history and does not qualify a project for the Emerging section, which keeps its own stricter, unchanged criteria (authentic inception within 30 days, at least USD 10,000 liquidity, at least USD 1,000 of 24-hour volume, fresh observation, canonical deduplication).

The intended contract: once admitted, a pool stays durably tracked and refreshed even after it drops out of top or filtered lists — a filtered list finds entrants, it does not define the monitored set. A verified, fresh fall below the floor removes the pool from positive selection at detection and suspends ordinary heavy work after a bounded confirmation; a verified drain or verified critical risk excludes immediately. Reasons and history are kept, and re-entry requires new qualifying evidence rather than continuous polling of excluded rows. One weak pool does not condemn a project that has another qualifying market.

Today this lifecycle is specified, not deployed. The admission boundary above is the only part bound in code. Reconciling already-registered legacy identities against this policy is also pending.

Target cadences under consideration are roughly 5 minutes for new or fast-changing markets, 15 minutes for ordinary ones and 30 minutes for quiet ones. These are targets awaiting capacity validation, not current service levels. No cadence is claimed as operating today. Every market figure on Melega Space is published with its own observation age.

Planned direct coverage, certified per DEX, chain and version separately: Uniswap, PancakeSwap, PumpSwap, Meteora (DLMM and DAMM V2 counted separately), Raydium, Orca, Aerodrome, Curve, QuickSwap and LFJ. Fluid and Manifest are candidates. Melega's native DEX is one source among these, not the boundary of the universe and not a ranking privilege. Every entry above is planned until its specific deployment is certified; presence in this list is not a claim that the lane is live.

The intended routing is direct first: authorized DEX APIs, subgraphs and native paths, read in batches and incrementally, with aggregated sources used only to fill genuine gaps, as bounded failover and for small comparison samples. Direct adapters are not yet built, so this routing describes intended next work rather than how data is obtained today. Two aggregator brands that share one origin are treated as one source of evidence, not two independent confirmations.

Rankings and Melega Score stay organic and separate from paid placements. Sponsored placements exist and are clearly labelled as such; sponsorship can never buy observation admission, a score or an organic rank.