B-book execution lag vs. artificial latency on news: How are you guys isolating the two?
I've been running a semi-automated intraday strategy across three different two-phase evaluation firms over the last six months. They all white-label through similar grey-label setups, though their liquidity providers supposedly differ. Lately, I'm noticing a distinct drift in execution quality during pre-NY session liquidity transitions, specifically around the 08:30 EST releases.
I expect slippage during high-impact data—that's just the nature of thin books. But what I'm trying to diagnose is the systemic 180ms to 420ms round-trip delay that persists for up to five minutes after the initial volatility spike settles. On my LMAX and Interactive Brokers feeds, the spread compresses back to baseline within 45 seconds, yet the prop terminals are still serving delayed fills and artificially wide synthetic spreads. It smells less like LP congestion and more like deliberate virtual dealer plug-in throttling to prevent news-straddling or latency arbitrage.
For those of you scaling payouts on five-figure accounts, how are you benchmark-testing your broker-integration latency? Are you maintaining external FIX engine logs to compare timestamp gaps against what the MT5 terminal reports, or is it a losing battle trying to hold these platforms to institutional infrastructure standards? If it's a virtual dealer desk issue, switching firms won't do much if they all license the same backend bridge.
The cleanest way to isolate this is by comparing your broker execution timestamps against the raw LMAX or Saxo feed latency for the exact same millisecond window. If the market spread normalized within 150ms but your fill was held for 800ms, that is almost certainly artificial B-book delay rather than genuine network queueing during the news rush.