Liquidity Score Changelog: v4 archive

Impact notes

• Bunni is now excluded during crawl intake and DeFiLlama pool processing instead of being treated as a live DEX venue
• Retained-pool filters and challenger snapshots ignore Bunni even if stale rows or unexpected inputs survive earlier gates
• Liquidity scores, dexPriceUsd, and downstream DEX cross-checks no longer count Bunni TVL, pool counts, or protocol medians

Impact notes

• Sub-threshold direct API pools now still block later exact-address staged duplicates from re-entering scoring with incompatible TVL semantics
• GeckoTerminal and CoinGecko discovery rows can no longer inflate liquidity by resurrecting the same exact direct pool after the direct row was excluded from scoring
• Balancer stablecoin pools now dedupe correctly against Balancer direct API even when DefiLlama labels the row as generic `balancer-v3` without stable subtype metadata

Impact notes

• Pools that are skipped as duplicates or dropped by retained-pool quality filters can no longer keep influencing dex_prices
• dexPriceUsd, price_sources_json, and downstream dexPriceCheck consumers now reflect the same retained pool surface used for challenger publication and UI liquidity detail
• High-TVL discovery rows that never survive retained-pool admission can no longer manufacture near-peg DEX aggregates for depegged assets

Impact notes

• Meteora DLMM pools now enter the direct-API merge path with measured TVL, volume, balances, and fee data
• PancakeSwap V3 pools now add protocol-native primary coverage across BSC and supported EVM chains through official Graph subgraphs
• Aerodrome Slipstream and Velodrome Slipstream pools now contribute pool-state TVL, balances, fees, and DEX-price observations via the on-chain Sugar view contracts
• Direct-source precedence over overlapping DeFiLlama rows now requires measured non-zero 24h volume, so Slipstream pool-state rows expand coverage without displacing stronger DL rows when volume telemetry is absent
• New concentrated-liquidity quality buckets now score PancakeSwap and Slipstream fee tiers consistently with existing Uni V3 logic

Impact notes

• GeckoTerminal and CoinGecko Onchain token crawls now read multiple bounded pages instead of stopping after page 1
• DexScreener and CoinGecko tickers fallback now trigger for weak partial coverage, not only coins with zero pools or no DEX price
• Fallback orderbook rows now preserve explicit synthetic/decayed/provenance flags instead of masquerading as organic USDC pools
• Coverage confidence now incorporates protocol breadth, source-family breadth, measured balance share, measured price share, and synthetic or decayed TVL share
• Direct-API pools default to a shorter maturity assumption and Fluid reserve normalization now marks whether balances were safely measured
• Shared secondary-pool contribution logic centralizes GT/CG/staged/fallback aggregate handling to reduce drift across merge paths

Impact notes

• Direct API and staged/fallback pools resolve tracked assets by chain+address first, with chain-scoped symbol fallback only when unique
• Cross-source pool dedupe now uses exact pool ids first and derived token-shape matches only when they are unique on both sides
• Repeated sightings of the same physical pool across direct API, staged, and fallback sources now collapse before dex_prices weighting
• Depeg challenger inputs publish from the full retained pool set instead of the visible top-pools subset
• Fluid pools with missing token decimals now fall back to neutral balance rather than using unsafe raw reserve units

Impact notes

• Fluid top-pool rows now populate Balance and Detail when the official DexReservesResolver is deployed on that chain
• Fluid pool quality now uses measured balance health on resolver-backed EVM chains, rather than a hardcoded neutral 1.0
• Fluid fee detail now comes from the on-chain pool config (`1% = 10_000`), normalized to basis-point badges in the UI
• BSC and Plasma Fluid pools remain on neutral-balance fallback until Fluid ships the same resolver path there

Impact notes

• Direct-API Balancer, Raydium, and Orca pools now populate top-pool balance bars and detail badges
• Measured direct-API balance ratios now feed balance-weighted aggregates, stress, and effective TVL instead of assuming 1.0
• Balancer weighted pools normalize balance health against target token weights rather than raw reserve symmetry
• Orca vault balances are normalized from raw token units before balance-health calculation
• Top-pool fee tiers now serialize as actual basis points (for example 1bp, 5bp, 30bp) across UniV3, CG-onchain, and direct APIs

Impact notes

• Raydium lower-case poolType contract fix restores live Solana pool coverage
• Orca now paginates via cursor.next with retry/backoff and a below-threshold stop, instead of truncating after page 1
• Direct API pools are fingerprint-deduped and preferred over overlapping DeFiLlama pools before score computation
• Direct API sources merge before staged/DexScreener/CG-ticker fallbacks, preventing lower-confidence sources from claiming the same pool first
• direct_api-only rows now classify as primary coverage (confidence 1.0) instead of fallback coverage
• Fluid volume now uses one-sided USD-normalized pool volume instead of double-counting raw token legs
• Balancer intake is limited to supported stable/weighted pool families on mapped chains only

Impact notes

• Volume activity now uses log-scale (33.3*log10(V/T/0.005)); median score rises from 5 to ~35
• Cross-chain component removed; TVL Depth raised to 35%, Pool Quality to 22.5%
• Durability: organic 15% (sqrt curve), TVL stability 35%, volume consistency 25%, maturity 25%
• Locked liquidity sub-component removed from durability (no reliable data source)