Apologies for the year confusion but some archives work on the year of the first month of winter e.g EarthSat. Using my running 10 year Dec CET 4.3 and 30 year Dec CET 4.8, then even if you use the previous years only 1976 and 2009 are below normal and the average of all years is close to normal. 2, 5.2, 6.2, 3.1, 4.8. Factor in sunspots and 1976 and 2009 were bang in the middle of very low sunspot periods which is almost certainly a factor as well so then these Dec CETs are probably not cold at all. Once you get a normal or warmer than normal Dec then getting a strongly cold winter becomes increasingly difficult because you need at least one month to cancel out Dec anomalies and this is why there is not a very clear stats relationship between -ve AO and -ve NAO winters. Similar story with all the other teleconnections as well.