Dequan — Browser Latency Test

Measures network + browser overhead on each leg of the live trading pipeline against your nearest Cloudflare PoP and our NYC bare-metal box. Run the same test in Chrome, Brave, Opera, Edge, Firefox and export the JSON to compare.

Rounds: Warmup:

Idle.

Environment

Per-leg latency (ms)

Leg	Min	p50	p75	p95	Max	Mean	Failed

Tip: A clean browser profile with no extensions matters more than browser choice. Ad-blockers and privacy extensions intercept every fetch() and add 1–10 ms per request — that swamps the real engine difference (<5 ms) between Chromium browsers. Use a fresh incognito/private window or a dedicated test profile.

Notes on what is measured

cf-edge GET — HTTPS GET to auth.dequan.xyz/healthz (Cloudflare Worker, your closest CF PoP). Tests browser→CF leg only.
cf-edge POST 1KB — same endpoint with 1 KB body. Catches request-body buffering differences (upload cost).
trading-api GET — HTTPS GET to dequantrade-ws.dequan.xyz/healthz (NYC box via CF tunnel). Closest match to the real trade hop.
ws-handshake — full WebSocket open (TCP+TLS+HTTP upgrade). HTTP/3 vs H2 differences show up here.

All tests run sequentially with a 20 ms inter-round gap so we don't measure connection-pool contention. Connection reuse (keep-alive, H2 multiplexing, H3) is still active and measured naturally.

Raw JSON

// Run a test to populate this