Okay, here, I think I have every plot imaginable having to do with predicted 3/4 efficiency. First, for comparison, I replayed every 10th run with the scaler file efficiency requiring 3 other planes to fire (unlike before).
Second, I added an array to the engine to keep track of the 3/4 efficiency for just electron events with a cut on the Cherenkov and the shower counter.
So, when I refer to the engine efficiency, it's the first thing I mentioned, and the electron engine efficiency is the second.
Next, I used the elaborate 3/4 efficiency model (the one that projects the track to all the planes and looks at which paddle was hit and where), to make a more compact look-up table for predicting efficiencies as a function of just hsdelta. I made 2 such tables: one using all the events, and one using electron events, once again, cutting on the Cherenkov and shower counter.
Then, I ran a kumac that looked at every ten runs (the same ones I replayed in the previous part), and used the lookup table to predict an overall 3/4 efficiency and one just using electron events.
The results are below. ONe thing to note: The predictor seems to underpredict the 3/4 efficiency pretty consistenly, but I think that's due to the transition from the detailed model to just one based on delta. If I use the elaborate model, it reproduced the "correct" 3/4 efficiency that the engine gives.

First, here are the 2 efficiencies (overall and one with just electron events) from the engine. The electron efficiency is usually higher.

Next, here's the same plot, but with the predicted efficiencies instead.

Here, we have the two efficienies using just electron events side by side:

And this one is for all events

And finally, here's the ratio of the engine efficiency to the predicted one, using electrons:


Just, for reference, this is what the reference run looks like, with just electrons:

And zooming in...