Good points on ensembles, however they are the much better long range tool than a single Op run, especially 1 or 2 weeks out.
To your point, ensembles often have different camps that might skew the reality of the outcome. The EPS has 50 ensemble members, so for example, you might have 25 cold & 25 warm, which could disguise the true eventual reality. However, once you get a good majority of members that agree, then you can see the likely direction of the pattern.