OK, here are some results.
Now in the model, heating is arranged with simple electronic thermostat, which senses current indoor temperature and provides a variable heating power using a steep P controller (Google "heat anticipator" for explanation of the classic mechanical PWM device), capping max power to 10kW.
Again, at the top/blue is the calculated "required" heating/cooling power if steady 21degC would be desired. This info is not used in calculation, just for reference. Top/orange is the solar irradiance power. Again, second graph is outdoor temp. Third graph is simulated room temp. Fourth graph is cumulative output energy.
Blue is low thermal capacity house (3kWh/K), orange is high thermal capacity house (15kWh/K). Insulation (i.e., average heat flow through envelope) is the same.
First, bear with me, without hourly cutting of power
see thermo1.png
As you can see, the cumulative energy drifts apart whenever there is overheating due to solar irradiance. Higher thermal capacity house shows reduced consumption right after such sunny days as it stores the energy. This is even with a "dumb" thermostat which does not understand the thermal capacity. Two month consumption, 1015.8kWh vs. 976.53kWh in favor of higher thermal capacity house.
Now let's cut the power to both houses. Let's cut it for 12 hours per day, say between 6am to 6pm when going to work and, in very non-Finnish way, eat out after work.
see thermo2.png
Two month consumption is now to 948.45 and 948.56 kWh, basically the same result. As you can see, the average temperature of the low-C house dropped, which also dropped its consumption, but not very much. With high-C house, consumption also dropped, but not as much; but it's still as good.
You asked to "start early enough to get it there". Now this expectation comes from the mindset of low thermal capacity. I didn't implement such logic in the model, but as you can see the high thermal house keeps decent nice indoor temperature without such trickery. Here's a zoom-in on transient response:
see thermo3.png
So while you can see that high-C house heats up slower, it starts from higher temperature. You decide which one is better, especially if you miss-schedule and fail to predict your living patterns.
This example data was not cherry picked. I'm sure I could find examples of both high-C building showing significantly better result (now the difference was marginal), but also could find data to prove your point where low-C building shows better result.