Bollinger %B on ratio of USO to SPY

Another interesting relationship.  The ratio of USO over SPY seems to follow a downtrend, probably due to USO's negative bias being based on rolling futures, and its expense ratio.

However, extremes in this ratio seem to indicate suitable entry points in USO. I am using Bollinger bands to visualize them and a %B limit of zero as a cut off.  In the last 3 years, there have been seven instances where %B has gone at or below zero, including now. The idea would be to accumulate positions every week when %B < 0.  This may cause to appear to average down if %B stays under zero for more than a week, but the previous six times have all been followed by rallies in USO.

There are also interesting gaps in the USO drop, which may suggest a bottom is near. Of course what I marked as exhaustion gap may turn out to be a measuring gap.  I don't know for sure how much these gaps can be trusted.