While I have moved (still in hot Florida) I now have a better situation to actually test this. I put an IoTaWatt on my power, so I have minute by minute power draw on the A/C, as well as I can computer control my thermostat.
I have been running variations of two situations:
1) A steady 74F thermostat setting all day, and
2) A cold-soak setting that starts at 70 or 71 at 1am, then begins raising to 77F at mid-day, and drops back to 75 in the evening.
I have run a variety of times, so as to try to drive the majority of the A/C load into times when the outside air is cooler.
It is very easy to avoid the mid day, it only takes a few degrees increase in the thermostat (though it does become noticeable as it heats). Here's what the data look like:
The filled green (air handler) and yellow stacked (AC Compressor) are watts. The red line is OAT and the blue inside, with the yellow the thermostat setting.
You can see in the wee hours it keeps the temperature quite low, and coasts up. In this case from about 8:30am to 5pm it didn't run the A/C at all, with the inside temp peaking around 77-78. But then... it turns on and takes quite a while to get the temperature back under control.
I have also tried this delaying the rise in set point until later in the morning (as the OAT is pretty cool still), which can slightly delay the evening run time, but it is not enough.
As a typical comparison, this day required 19.01 kWh for the managed day, and 15.59 kWh for a steady 74F setting. The OAT both days was about the same (and same 90F high each day). On advantage this time of year is weather here is very repeatable. I compared with attic temperature, the managed day reached 105.9 and the day I set to 74F it was 106.8 (I use the attic as a rough indicator of radiant heat load); while different it would imply a bigger demand on the lower kWh day.
I've gone through 3-4 iterations of a day each way, with different configurations. In every case, the steady setting came out better in terms of power usage, some larger, with this example being pretty average.
Now that's not to say there is not a magic arrangement, and it might vary depending on efficiency of the outside unit (mine is one year old 16 Seer lennox).
But as that famous TV series said, at least for me: Myth Busted!
PS. The explanatory solution is certainly correct from @Harper - Reinstate Monica is correct, but I offer this as a specific answer from actual data.