It never ends

Here's the first look at the STOF efficiency. This is a leg of the ELREAL trigger that requires 2 out of 4 scintillator planes to fire, one in the first plane and one in the second (S1X or S1Y, and S2X or S2Y).

For x<1, the efficiency is practically flat. You see 2 bands here, one at .999 and one at .998 because the engine only prints this to 3 decimal places in the scaler file.
For x>1, there is a decrease in the stof efficiency and also an increased scatter.
Even if one averages over the scatter, it's lower on average for x>1.

An update

Here's a plot of STOF efficiency (same as above) compared to what the predictor gives (the 2-D) predictor. This is using the probability of a given track producing a "hit" in each plane.