If you look at your interesting paper, nearly every very large event, > 1000 nT, occurred with sunspot counts of around 100 or less.
I actually didn't see sunspot count presented as part of that data. However, if you look at page five of this paper, it does have a nice plot of severe geomagnetic storms versus sunspot cycle. That data directly contradicts your hypothesis, as in four out of five cases the largest storms occurred at or near the peak of active cycles. The second largest storm occurred right at the peak of Cycle 22, when sunspot counts were around 200.
I'd be interested in any papers you can reference that came to the conclusion that the largest storms are likelier during weak cycles and/or low activity. It is a contention of the paper that I referenced above that the peak for solar storm activity comes somewhat after cycle maximum, however that says nothing about the distribution of the largest storms. Also I think this paper is referencing CIR storms as opposed to CME storms.