I'd rather it look good in the ultra-long range than bad. I look at it this way: given the known model biases towards cold in the long range, if the long range looks good, it might turn out good, it might turn out bad, it might turn out average. If the long range looks warm, that is indicative of a strong warm signal. Of course I'm just referring to consistent runs that move forward in time. Plenty of times the ultra-long ranges just flip back and forth randomly; just NWP noise.