feat: badge redesign + chain x region scoping, per-chain bench pages

benchmarks/evm-quote-latency.yml

@@ -78,6 +78,10 @@ dimensions:
  - { value: eu-west, label: EU West }
  - { value: sgp, label: Singapore }
 
+# Exact per-cell (chain x region) rankings for scoped badges. One query,
+# one sample per (provider, chain, region) cell.
+rank_matrix_query: histogram_quantile(0.50, sum by (provider, chain, region, le) (rate(evm_swap_quote_latency_ms_bucket[24h])))
+
 providers:
  - slug: mobula
  name: Mobula

benchmarks/oracle-deviation.yml

@@ -69,7 +69,7 @@ methodology:
  - "Chainlink round age. The on-chain `updatedAt` timestamp from `latestRoundData()` is published separately as `ocb_oracle_last_round_age_seconds{source=\"chainlink\", pair}`. Chainlink updates only on deviation (typically 0.25-0.5% for blue chips) or heartbeat (~1 h), so a 30-minute age on a quiet ETH minute is normal; a 2-hour age on SOL during a volatile minute is the actual signal."
  - "USDT ≈ USD assumption. Binance only quotes USDT pairs (BTCUSDT, etc.). We treat USDT as ≈ USD ± 10 bps drift, which is acceptable for a bench whose alert floor is ≥ 10 bps. A real USDT depeg would surface as Binance drifting from the other three sources for *every* pair simultaneously, exactly what we want this bench to flag, not hide."
  - "MATIC → POL migration. Polygon migrated MATIC → POL 1:1 on Sep 4 2024. The Chainlink mainnet feed contract is still named MATIC/USD but on-chain `description()` confirms it tracks the POL token; Pyth renamed the feed to POL/USD; Coinbase delisted MATIC-USD and only lists POL-USD; Binance kept MATICUSDT as a frozen historical pair AND lists POLUSDT. We point Binance at POLUSDT so all four sources track the same underlying asset. Bench label is kept as `pair=\"MATIC/USD\"` for query continuity."
- - "Time-aligned deviation (canonical headline). For every pair of sources (a, b) we anchor on the more recent of their two SourceTSs (Chainlink's on-chain `updatedAt` for Chainlink, fetch time for the continuously-updating sources) and look up the older source's price in a 30-minute per-source rolling history at the anchor moment. The result is published as `ocb_oracle_deviation_at_oracle_ts_pct` and feeds the leaderboard headline `ocb_oracle_max_deviation_pct`. This eliminates the artifact where Chainlink's heartbeat lag inflates 'deviation' by the market's drift between its updates — a researcher grading oracle quality wants to compare Chainlink's price against the market at Chainlink's own updatedAt, not at the harness's fetch instant. The legacy fetch-time gauge (`ocb_oracle_deviation_pct`, aliased as `ocb_oracle_deviation_at_fetch_ts_pct`) is preserved unchanged for backward compatibility. Alignment misses (no history sample within ±10s of the anchor) are counted as `ocb_oracle_alignment_miss_total`. Methodology recommendation from Coinpaprika data team review; matches the convention published by Chaos Labs and Risk DAO oracle risk reports."
+ - "Time-aligned deviation (canonical headline). For every source pair we anchor on the more recent SourceTS (Chainlink's on-chain `updatedAt`, fetch time for continuous sources) and look up the older source's price in a 30-minute rolling history at that moment. Published as `ocb_oracle_deviation_at_oracle_ts_pct`, feeding `ocb_oracle_max_deviation_pct`. Removes Chainlink heartbeat-lag inflating deviation; alignment misses count as `ocb_oracle_alignment_miss_total`."
  - "Excluded by design. Redstone (push-pull, no continuous gauge to scrape without integration contract), Uniswap V3 TWAP (per-pool integration + derivation of same CEX prints), DIA (smaller footprint than the four kept), and aggregator-of-aggregators (CoinGecko, CMC, DefiLlama), re-aggregating already-aggregated data adds latency and hides per-source disagreement."
 
 findings:

benchmarks/rpc-capabilities.yml

@@ -124,6 +124,12 @@ dimensions:
  - { value: eu-west, label: EU-West }
  - { value: sgp, label: Singapore }
 
+# One instant query feeding the exact per-cell (chain x region) rankings
+# behind scoped badges and leadership claims. Cross-region averages hide
+# region-restricted leaders (dRPC wins every chain from sgp only), so
+# badge scoping reads these cells instead of the aggregate ranks.
+rank_matrix_query: avg by (provider, chain, region) (quantile_over_time(0.50, rpc_latency_milliseconds[24h]))
+
 # Real metrics emitted by the rpc-capabilities harness:
 # rpc_latency_milliseconds{provider, chain} gauge
 # rpc_latency_milliseconds_histogram{provider, chain} histogram

benchmarks/solana-dex-quote-latency.yml

@@ -76,6 +76,10 @@ dimensions:
  - { value: eu-west, label: EU-West }
  - { value: sgp, label: Singapore }
 
+# Exact per-region rankings for scoped badges. One query, one sample per
+# (provider, region) cell.
+rank_matrix_query: histogram_quantile(0.50, sum by (provider, region, le) (rate(solana_quote_latency_ms_bucket[24h])))
+
 faq:
  - q: "Which Solana DEX has the fastest quote API right now?"
  a: "{{best_name}} currently returns quotes the fastest at {{best_p50}} (p50, 24 h) across {{count}} measured providers. The leaderboard re-sorts every 60 seconds against fresh samples and rotates the target token each tick so no provider can serve from edge cache. The answer reflects 24 hours of measured latency across us-east, eu-west and sgp, not a marketing-page claim."

benchmarks/stablecoin-peg.yml

@@ -58,12 +58,12 @@ methodology:
  - "Sources per stable, USDC: Binance USDC/USDT (USDT-anchored, secondary), Kraken USDCUSD, Bitstamp usdcusd. USDT: Kraken USDTUSD, Bitstamp usdtusd. FDUSD: Binance FDUSDUSDT (USDT-anchored, secondary). USDe: Binance USDEUSDT (USDT-anchored, secondary). DAI: Curve 3pool `get_dy` forward (USDC → DAI) and reverse (DAI → USDC)."
  - "Cadence: 5 seconds for CEX REST tickers, 12 seconds for Curve `get_dy` on-chain (matches Ethereum block time). Per-venue samples are grouped into 60-second buckets for the liquidity-weighted median aggregation."
  - "Aggregation: per-minute liquidity-weighted median across USD-quoted venues (`Quote = QuoteUSD`). Weight is 24 h USD volume estimate. The aggregated price drives `peg_deviation_bps` (per-minute median, retained for backward compatibility) and `peg_deviation_bps_histogram`."
- - "Primary headline: `peg_deviation_worst_bps` — the per-minute MAX |price − $1.00| across every venue sample in the 60-second bucket, in basis points. The leaderboard ranks stables on `quantile_over_time(0.99, peg_deviation_worst_bps[24h])`. Switching from per-minute median to per-minute max surfaces the sub-minute depeg wicks (e.g. a 5-second print to $0.92 on one venue) that the median would smooth into invisibility — matching the OHLC convention used by every TradFi reference rate. The companion gauges `peg_minute_min_bps`, `peg_minute_open_bps` and `peg_minute_close_bps` expose the full bar shape so consumers can reconstruct the within-minute distribution."
+ - "Primary headline: `peg_deviation_worst_bps`, the per-minute MAX |price - $1.00| across all venue samples in the 60-second bucket, in basis points. The leaderboard ranks stables on `quantile_over_time(0.99, peg_deviation_worst_bps[24h])`. Per-minute max surfaces sub-minute depeg wicks that a median would smooth away, matching the OHLC convention of TradFi reference rates. Gauges `peg_minute_min_bps`, `peg_minute_open_bps`, `peg_minute_close_bps` expose the bar shape."
  - "Cross-venue gap (OCB-only number): `max(price across USD-quoted venues), min(price)`, per minute, in basis points. Surfaces the windows where Coinbase quotes $1.0003 while Kraken quotes $0.9978. Stored as `peg_cross_venue_gap_bps`."
  - "USDT-anchored secondary metric: Binance USDC/USDT, FDUSD/USDT and USDE/USDT are exposed on `peg_deviation_usdt_anchored_bps{venue}` separately so the USD-anchored primary leaderboard is not contaminated by USDT's own peg deviation."
  - "Time outside band: total seconds in the trailing 24 h during which the per-minute aggregated price fell outside [0.995, 1.005] (±50 bps). Split into `peg_time_below_peg_24h_seconds` (< 0.995) and `peg_time_above_peg_24h_seconds` (> 1.005) because Circle redemption only clears above-peg, so the direction tells you which failure mode is active."
  - "Depeg event flag: binary `peg_depeg_event_flag` set to 1 when the per-minute aggregated price has been outside [0.97, 1.03] for ≥5 consecutive minutes; cleared after 30 minutes back inside. Conservative so it does not flap during normal stress events."
- - "Outlier rule (multi-venue consensus): a single sample more than 2% off peg is kept only when at least one OTHER venue has also been outside the same band in the same direction within the last 30 seconds. Single-venue glitches (one CEX returns a stale or fat-finger print while every other venue is at $1.00) are dropped as `dropped_isolated`. Multi-venue confirmation (Kraken AND Bitstamp both at $0.87 = real depeg) is kept as `kept_corroborated` so the percentile metric surfaces the event. Replaces the previous flat 20% drop / 10% cap which would have clipped USDC at $0.87 during the March 2023 SVB depeg to $0.90, erasing the event in the percentile metric. Sanity floor: anything more than 50% off peg is treated as a parser bug and dropped regardless. Methodology recommendation from Coinpaprika data team review."
+ - "Outlier rule (multi-venue consensus): a sample more than 2% off peg is kept only when at least one other venue was outside the same band in the same direction within the last 30 seconds. Isolated single-venue glitches are dropped as `dropped_isolated`; corroborated moves (Kraken AND Bitstamp both at $0.87) are kept as `kept_corroborated`, so a real depeg like USDC during SVB in March 2023 survives into the percentile metric. Anything more than 50% off peg is treated as a parser bug and dropped."
  - "Excluded by design: aggregator-only prices (CoinGecko, Coinmarketcap, DefiLlama) because they are themselves liquidity-weighted medians of the venues we already poll directly. Algo-stables that have already failed (UST, USDR) are out of scope; the bench tracks live, currently-redeemable stables."
 
 findings:

src/app/about/page.tsx

@@ -52,7 +52,29 @@ export default function AboutPage() {
  13 live benchmarks. ~150 (provider × chain) probe pairs. Every metric is queryable on the public Prometheus and reproducible from the harness source.
  </p>
 
- <SectionRule label="How it works" number="iii" />
+ <SectionRule label="Independence and disclosure" number="iii" />
+ <p className="text-base leading-relaxed text-ink-soft">
+ OpenChainBench is funded by <a className="lnk" href="https://mobula.io" rel="noopener">Mobula</a> because we needed honest infrastructure benchmark data internally and saw no neutral source existed. The site is open-sourced because the industry needs it.
+ </p>
+ <p className="mt-4 text-base leading-relaxed text-ink-soft">
+ <strong>Mobula competes in five live benchmarks</strong>: aggregator head lag, bridge fee, bridge quote latency, network coverage and metadata coverage. Mobula currently leads aggregator head lag and bridge quote latency. This is a real conflict of interest and we want it visible above the fold, not buried.
+ </p>
+ <p className="mt-4 text-base leading-relaxed text-ink-soft">
+ Three safeguards stand against the conflict:
+ </p>
+ <ul className="mt-3 space-y-3 text-base leading-relaxed text-ink-soft">
+ <li className="flex gap-3"><span className="text-ink-faint">·</span><span><strong>Open Prometheus data</strong>. Every number on the site is a literal <code>quantile_over_time</code> query. Anyone can hit our <Link className="lnk" href="/api/citable">/api/citable</Link> or <code>/api/stat/&lt;slug&gt;</code> endpoint and re-derive the leaderboard with their own aggregation. If Mobula were inflating its rank, the raw data would show it.</span></li>
+ <li className="flex gap-3"><span className="text-ink-faint">·</span><span><strong>Open harness source</strong>. Every harness is on <a className="lnk" href="https://github.com/ChainBench/OpenChainBench/tree/main/harnesses" rel="noopener">GitHub under harnesses/</a>. Clone, run <code>docker compose up</code>, your <code>/metrics</code> endpoint emits the same numbers ours does within 30 seconds.</span></li>
+ <li className="flex gap-3"><span className="text-ink-faint">·</span><span><strong>Public methodology review</strong>. We invite external review and ship the fixes publicly. In June 2026 the Coinpaprika data team flagged four issues on the stablecoin peg and oracle deviation benches. We shipped three of them in pull requests <a className="lnk" href="https://github.com/ChainBench/OpenChainBench/pull/349" rel="noopener">#349</a>, <a className="lnk" href="https://github.com/ChainBench/OpenChainBench/pull/352" rel="noopener">#352</a> and <a className="lnk" href="https://github.com/ChainBench/OpenChainBench/pull/353" rel="noopener">#353</a> within twenty-four hours and pushed back on the fourth with citations to CME, Chainlink and CoinGecko convention.</span></li>
+ </ul>
+ <p className="mt-4 text-base leading-relaxed text-ink-soft">
+ We do not run a paid tier. We do not sell ranking slots. We do not take provider sponsorship in exchange for inclusion. There is no token. If you spot any deviation from this policy, file a <a className="lnk" href="https://github.com/ChainBench/OpenChainBench/security/advisories/new" rel="noopener">private security advisory</a> and we will treat it as the integrity incident it would be.
+ </p>
+ <p className="mt-4 text-base leading-relaxed text-ink-soft">
+ Hosting and infrastructure costs are paid by Mobula. The site runs on Vercel; harnesses run on Railway. We are open to grant funding that preserves editorial independence, but we will not accept funding from any party we benchmark.
+ </p>
+
+ <SectionRule label="How it works" number="iv" />
  <p className="text-base leading-relaxed text-ink-soft">
  Every benchmark is a YAML spec plus a harness. The spec describes what to measure, which providers, which Prometheus queries hold the numbers; the harness runs continuously on Railway and exposes those metrics. A single shared Prometheus scrapes every harness; the site queries Prometheus directly and re-renders every minute. Every provider is rendered with equal visual weight. readers do their own ranking.
  </p>
@@ -62,15 +84,15 @@ export default function AboutPage() {
  walks through the steps. New providers, new metrics, new chains. all welcome via pull request.
  </p>
 
- <SectionRule label="What you can do" number="iv" />
+ <SectionRule label="What you can do" number="v" />
  <ul className="space-y-4 text-base leading-relaxed text-ink-soft">
  <li className="flex gap-3"><span className="text-ink-faint">·</span><span><strong>Read</strong> the{" "}<Link className="lnk" href="/#latest">live benchmarks</Link>{" "}.</span></li>
  <li className="flex gap-3"><span className="text-ink-faint">·</span><span><strong>Reproduce</strong> any number. the{" "}<Link className="lnk" href="/methodology">methodology</Link>{" "}page tells you how.</span></li>
  <li className="flex gap-3"><span className="text-ink-faint">·</span><span><strong>Contribute</strong> a harness via{" "}<a className="lnk" href="https://github.com/ChainBench/OpenChainBench">GitHub</a>. Pull requests for new providers and new benchmarks are reviewed openly.</span></li>
  <li className="flex gap-3"><span className="text-ink-faint">·</span><span><strong>Discuss</strong> ideas in{" "}<a className="lnk" href="https://github.com/ChainBench/OpenChainBench/discussions">GitHub Discussions</a>{" "}or follow{" "}<a className="lnk" href="https://x.com/OpenChainBench">@OpenChainBench</a>.</span></li>
  </ul>
 
- <SectionRule label="Contact" number="v" />
+ <SectionRule label="Contact" number="vi" />
  <p className="text-base leading-relaxed text-ink-soft">
  See a number you can&apos;t reproduce? File a{" "}
  <a className="lnk" href="https://github.com/ChainBench/OpenChainBench/issues/new/choose">