Empirically, I think you still want below average temps, though the variance is high. 2016-2017 sticks out like a sore thumb.
I sorted the winters in your sig based on snowfall. (I estimate the 2019-2020 snowfall...CON had 46 but you prob had at least 15-20" more I'm betting)
If we sorted by precip, it would have a higher correlation I'm sure. (i.e., temps matter, but less than precip)