{
  "schemaVersion": "vultax-public-research-data-v1",
  "articleUrl": "https://vultax.com/research/crypto-exchange-feed-latency-september-2026",
  "datasets": [
    {
      "title": "Observed latency by venue",
      "columns": [
        {
          "name": "Venue",
          "unit": null
        },
        {
          "name": "Median (p50)",
          "unit": null
        },
        {
          "name": "95th percentile (p95)",
          "unit": null
        },
        {
          "name": "Rank by p95",
          "unit": null
        }
      ],
      "rows": [
        [
          "Coinbase",
          "35.0 ms",
          "288.5 ms",
          "1"
        ],
        [
          "Bitrue",
          "69.5 ms",
          "311.6 ms",
          "2"
        ],
        [
          "Bybit",
          "72.8 ms",
          "331.1 ms",
          "8"
        ],
        [
          "WhiteBIT",
          "72.8 ms",
          "330.9 ms",
          "7"
        ],
        [
          "Gemini",
          "75.4 ms",
          "361.8 ms",
          "13"
        ],
        [
          "Binance",
          "76.8 ms",
          "328.1 ms",
          "5"
        ],
        [
          "Gate.io",
          "77.2 ms",
          "333.5 ms",
          "10"
        ],
        [
          "BitMEX",
          "78.0 ms",
          "338.9 ms",
          "12"
        ],
        [
          "Poloniex",
          "79.8 ms",
          "338.6 ms",
          "11"
        ],
        [
          "OKX",
          "80.0 ms",
          "326.5 ms",
          "4"
        ],
        [
          "HTX",
          "80.4 ms",
          "324.3 ms",
          "3"
        ],
        [
          "Bitget",
          "81.0 ms",
          "330.7 ms",
          "6"
        ],
        [
          "KuCoin",
          "81.0 ms",
          "332.5 ms",
          "9"
        ],
        [
          "Bitstamp",
          "82.9 ms",
          "406.0 ms",
          "15"
        ],
        [
          "Upbit",
          "92.8 ms",
          "589.7 ms",
          "16"
        ],
        [
          "BTSE",
          "111.5 ms",
          "383.5 ms",
          "14"
        ]
      ],
      "measurementContext": "Order-book snapshot age at capture, one-hour sample, September 2026. Ordered by median; rank 1 is the shortest tail. Coinbase over 64,647 snapshots, BTSE over 27,418.",
      "sources": [
        {
          "label": "Vultax order-book snapshot collection",
          "url": "https://vultax.com/methodology",
          "note": "Per-snapshot latency recorded at capture time, minute-level coverage per venue, and the explicit-gap standard"
        },
        {
          "label": "Glassnode — Centralized exchange latency monitor",
          "url": "https://latency.glassnode.com/cex",
          "note": "Independent WebSocket and FIX round-trip probes to major venues from multiple regions; measures a different quantity from snapshot age"
        },
        {
          "label": "Vultax editorial policy",
          "url": "https://vultax.com/editorial-policy",
          "note": "Sourcing standards and AI disclosure"
        },
        {
          "label": "Yahoo Finance — AWS outage takes down Coinbase (20 Oct 2025)",
          "url": "https://finance.yahoo.com/news/aws-outage-takes-down-coinbase-104129926.html",
          "note": "Coinbase unreachable during an AWS outage while Binance, Kraken and OKX stayed online; 'all funds are safe'."
        },
        {
          "label": "Coinbase — A postmortem of our May 7, 2026 outage",
          "url": "https://www.coinbase.com/blog/a-postmortem-of-our-may-7-2026-outage",
          "note": "The venue's own account of the May 2026 outage traced to an AWS thermal event; roughly eight hours down and about twelve more to full recovery, by the postmortem's timeline."
        },
        {
          "label": "CoinAPI — Why Real-Time Crypto Data Is Harder Than It Looks",
          "url": "https://www.coinapi.io/blog/why-real-time-crypto-data-is-harder-than-it-looks",
          "note": "Independent account of connections that stay open but stop delivering updates, silent data gaps and out-of-order updates"
        }
      ],
      "measurement": {
        "kind": "observed",
        "temporalCoverage": null,
        "population": "16 centralised crypto exchanges with order-book snapshot feeds",
        "methodology": "Per-snapshot latency recorded at capture time; median and 95th percentile per venue over a one-hour sample, with no gap filling."
      },
      "id": "section-observed-median-latency-by-venue-table",
      "kind": "published-table",
      "description": "Order-book feed latency on 16 crypto exchanges: medians from 35.0 ms (Coinbase) to 111.5 ms (BTSE), a 95th percentile of 288–590 ms on every venue, and data in every minute of the day. Plus the coverage check that catches a feed that stays connected and stores almost nothing.",
      "articleUrl": "https://vultax.com/research/crypto-exchange-feed-latency-september-2026",
      "sectionUrl": "https://vultax.com/research/crypto-exchange-feed-latency-september-2026#section-observed-median-latency-by-venue",
      "publishedAt": "2026-09-02",
      "modifiedAt": "2026-09-16",
      "version": "4bc31c4a175d13330512012423cdaab01bdcd1fb230f0da8c6e8ef85d6547992",
      "isLive": false,
      "limitations": "Published research snapshots and calculations; these exports do not contain current quotes or live account data. Observation windows and populations are stated in the source captions and methodology. Missing metadata is unknown, not zero. Measurements describe observed data-feed behaviour from Vultax collection infrastructure over the stated windows. They are not a statement about any exchange's internal systems or conduct, and nothing here is financial advice.",
      "downloads": {
        "json": "https://vultax.com/research/crypto-exchange-feed-latency-september-2026/data.json?dataset=section-observed-median-latency-by-venue-table",
        "csv": "https://vultax.com/research/crypto-exchange-feed-latency-september-2026/data.csv?dataset=section-observed-median-latency-by-venue-table"
      }
    }
  ]
}
