As you know JB and I use an analog method that identifies years with similar expected and antecedent conditions. They prove very useful for predicting the future. The only year since 1998 since I have been doing this which has significantly departed for the seasons was 2001/02 winter when a strong second solar max in terms of solar flux and UV warmed low and mioddle latitude atmospheres and shrunk the polar vortex.

I presented this at NWS regional winter workshops in the fall of 2002 and was invited the present that to CPC in Maryland. High ultraviolet produces warming through ozone chemistry, a finding that was speculated by Labitzke and van Loon (1997) and  confirmed by Hansen's super modeler Drew Shindell (1999) who included ozone chemistry and flux/UV in a special climate model and got improved verficiation retrospectively..They all found a tendency for the warmth to work its way down from the high atmosphere where to the middle troposphere where it affected the jet stream and flow patterns.

The peak in flux/UV in 2001/02 winter was very clear (from September 2001 to March 2002.

The resultant warming in the low and middle latiudes was likewise very clear. This is the 500mb height in February. This looks more like the southern hemispehre with a tight polar vortec in blue - trapping real cold air in the polarregions and a warm ring in low and middle latitudes.

While I was there I made a case for an analog approach that used PDO, AMO, QBO, ENSO, solar and other factors. I suggested the winter of 2002/03 would because of the new regime in the Paciic would be a throwback El Nino more like those in the 1950s to 1970s with cold in the east and warm west and lots of snow. I was told that they felt it was a 'slam dunk' likelihood that 2002/03 would be a classic warm north El Nino almost like 1994/95,1997/98. It was indeed a cold winter with 5 successive below normal months in the northeast and heavy snows. CPC at he time had abandoned use of their climate models and were focusing on ENSO and trends.

The CPC has moved towards a more reasonable approach with better results. Huug van den Dool is one of their statistical superstars. he has created a number of usefrul statistical models - a soil moisture model that generally does well in late spring and summer (though not this year) and a constructed analog model that looks at sea surface temperature anomaly patterns globally and matches them to past year. He then weights them based on similarity. This approach brings good results becaiuse it sees PDO, AMO, ENSO, IOD patterns and sequences. It does not consider QBO, solar and volcanism which can throw it off as in 2001/02

This is the list of weights for years since 1956. The higher the number, the greater the similarity. Negative numbers are opposite.

The best matches as bold. Note the data before 1956 is not available. We would expect 1955 and 1950 to be matches as well. Here is the winter depcition for the constructed analog. It is very similar to the WeatherBell forecast this year.

.  .

Here is a composite of the top years and adding 1950 and 1955, similar w/r ENSO, PDO, AMO.

Here is the winter temperatures for the same years.

Re; Comment below: Link to difference between 1999 (which might seem to be an analog) and 2011.