To summarize your experience, you first went higher than you might want, then later went significantly low. The low means at that time there was too much insulin for the carbs that had hit your blood stream. The earlier high means the insulin didn’t get active soon enough to counteract the initial glucose from eating.
So how to deal with these opposite-direction pressures? With traditional insulin like Humalog or Novolog, I’d approach it like this. To reduce the tendency to crash later, either reduce the insulin or slow it down. I’d try modifying the dose instead of 5 + 2 over 2 hours, maybe 4 + 3 over 4 hours, or 3.5 + 3.5. But that would mean an even greater initial BG spike. So to deal with that, you need to speed up the initial activation of the insulin. One way to do that is to give the insulin earlier before eating. Try 20 or 30 minutes earlier maybe even further before that. The idea is to get your BG heading down just in time for the initial BG spike to cancel that out and turn it into a mild rise. That’s a delicate balancing act and you’ll have to run the experiment a few times to adjust the spike downward and the crash upward, tweaking the insulin amount, the split between initial and square wave, the length of the square wave, and the pre-bolus time. Use your CGM to watch for the BG to be heading down before you eat (“wait for the turn” i.e., the turn downwards in the graph.) If the CGM graph looks like you’re heading for a low at any point, don’t hesitate to take a few glucose to correct that. With a belly full of pizza the glucose won’t be as fast as usual, but it still should be able to keep you out of trouble. Do you want me to say more about adjusting those 4 knobs (the amount, the split, the length, and the pre-bolus time)?
Another approach is to use a small dose of faster insulin to handle the initial rise, in combination with the main long square-wave that’s supposed to catch the expected late rise from slow-digesting pizza. Afrezza is fast enough to do this. Fiasp may or may not be.
I have a seriously hard time with Pizza, even when I assume 50-60 carbs/slice. I have only ever gotten it right once, and have been chasing that result ever since. The one time that was perfect was a 45/55 % bolus split over 70 minutes, followed up with an additional 12% 3 hours later, so the entire bolus was 112% of normal over a 5.5 to 6 hour period.
If it was us, unless your 4U bolus was IM, we would think the drop was from previous basal/ bolus. The reason is that we start dropping slightly at 25 minutes, but we don’t see a sharp downturn until 45 minutes.
But, for you, it could be one or the other depending upon how long it takes for your insulin to activate. On the whole, I’d be inclined to think it is a previous bolus/ basal, but I could be wrong: possibly you activate very quickly.
A good idea, imho, would be to test that: for a series of, say 10 boluses, test when you see when you “turn the corner.” For us, as mentioned above, it is about 45 minutes. We use this info a lot for planning or understanding.