REW EQ Optimization moves predicted result further off target curve

Hi,
I'm on REW 5.31.1. Using the EQ function for a subwoofer with a target curve in the range of 16-160 Hz. (no add room curve, crossover cutoff 1000, LF cutoff 0), Flatness = 1, overall gain = 0 and individual gain = 12.

I see the EQ function updating the predicted response and the result appears to be almost on top of the target curve. Then it says "Optimizing Gains and something" and then the predicted curve is further off target. It initially looks to be well within the 1 dB flatness, but after it is about 1.6 off target in one place.

What is it that is being optimized and why does this result in a target that is further off target?
I can manually go adjust the filter gains to get it closer but I assume that would mess up the thing that was being optimized.

I can send a video of that helps.

Files as requested. The predicted EQ result is above the flatness target from about 89 and 160 Hz range Seems to be manually correctable with just lower filter gains.

Thanks John. Your responsiveness is world class.

So does the fix end up matching the target curve more closely than it does now? Or does it just not look like it is better before the optimization pass?

- I took your response as "The Q's were not within limits, so the optimization fixed that, but then the gains were not readjusted after fixing the Q's?
- Again, it looks like I can correct the predicted to better match the target manually by just lowering the gains on some of the filters with the Q's where they ended up - so I think that the program should be able to do that as well.

.

The picture worth a thousand words!
Will be interested to see what happens on my satellites as well.
Excellent!!