Trainer accuracy - just a question

I have a question that has been touched on previously but I just wanted to clarify one point.
This can be a can of worms so I’m hoping it will get less action in this forum.

This pertains to wheel on set ups.
I know the speed and distance calculated in game will be different from the speed and distance that show up on a cycle computer because of hills, drafting etc.

But, if I ride out of the draft, on a flat road in Zwift, should the speed in game be the same as what my cycle computer shows?
I have accurately measured my wheels circumference.
The cycle computer reads 15 mph and my game speed is 20-22.
Both devices are using the same data from my speed sensor but arriving a two different answers.

Is this a valid test?
I do not want to start a thread of wheel on vs direct drive or smart trainer vs Z Power.
Just a question.
If anyone has ridden in a similar situation and noted agreement or disagreement between their 2 devices, that would be useful to know.

From my experience of riding on Zwift and a lot of it on my own, it over calculates speed so you can’t compare what your speed sensor says compared to Zwift. As far as I know your wheel speed plays no part in calculating speed on Zwift apart from being used for zpower to estimate your speed based on a known power curve for your trainer.

Speed in Zwift is a function of the power you are producing with your trainer, not the speed of your wheel. You can produce the same power at different speeds on your trainer by adjusting the trainer’s resistance, so although the speed of your trainer is a factor in how much power you are producing, the speed of your trainer is only indirectly related to your in-game speed.

Edit to add: this is true for classic trainers as well. Although you can’t adjust their resistance, the resistance of supported trainers at varying speeds is known and is used to calculate power output at those speeds. So the speed of a classic trainer is still only indirectly related to in-game speed.