Computer modeling has improved forecasts immensely over the last several decades. In the last 10 years the improvement has been remarkable. Many of our computer models are run out to 10 days or 15 days. While the specifics are frequently wrong on a 10 or 15 day forecast the general trends can be used to predict whether the temperature will be above normal or below normal, whether the pattern will be stormy or quiet, etc.
10/28 ECMWF Forecast Courtesy: WSI
This mornings weather porn winner goes to the European model. The model is our most accurate (by almost every standard of verification) and updates twice a day. The run from last night develops a hurricane over the Caribbean, moves it over south Florida, and then up the coast as a nor’easter with a rain storm changing to snow in southern New England. Still this forecast is 7-8 days out so take it with a grain of salt. The odds of this verifying are slim to none!
Incidentally one of our other computer models, the GFS, develops a hurricane over the Caribbean around the same time but sends it way out to sea. It does develop a separate storm/nor’easter that drops some snow in northern New England.
There’s no question there will be a lot to watch by the end of next week as the weather pattern becomes active. It’s too early to start diving into specifics but if I were you I wouldn’t get too excited about the blizzacane just yet.
Way Too Much Weather 
Your forecast of 30-45 days away from meaningful snow is basically climatology! I will say that there is virtually no correlation between a warm October and November snowfall so I’d be careful extrapolating the current and past weather pattern into a forecast.
The post was intended to be about the problems with long range modeling – certainly not that snow was likely in any location in CT.