What I mean is this. This is about the price of electricity, not the volume sold. Two prices are used to represent the whole price structure: baseload price and peak price. If you choose this representation, you want those two data points to be maximally separated. You achieve this by choosing for peak load the highest peaks only: working-day peaks, when power companies have to throw in their most expensive, rarely used resources.

Now the interesting thing in this comparison is not the prices themselves but their development over time (in response to the spread of renewables). Then it doesn't matter precisely how you define peak load, as long as you do it consistently over time. Then the IWR method is just as good -- just a different convention. What I don't buy is their claim that it is better.