I found my reference to your work...it was the outlook for not last winter, but the previous one.
Perhaps one of the most skilled forecasters on a forum replete with talented meteorologists and hobbyists alike, in americanwx.com, "StormchaserChuck" devised a formula roughly a decade ago that predicts the mean aggregate state of the NAO for the ensuing winter using the SSTs in an area of the north Atlantic. We strongly endorse this methodology as the most accurate predictor available for the mean state of the winter NAO.
The following methodology is a wonderful illustration of the delayed feedback between sea and air that represents the very essence of the elaborate system of atmospheric oscillations that we so often reference.
"In 2006 on a site called easternuswx, elaborate research was done with North Atlantic SSTs, showing high lagging predictive value for following Winter's NAO/AO. The correlation factor was higher than 0.4, and there was advantage over decadal cycles. Meaning, it would predict years that reversed the decadal trend. The index was very accurate in predicting the +NAO for the 2006-2007 Winter, and got much attention after a topic called "This will be the warmest Winter on record for the US" (It was 7). Since then the index has performed wonderfully after the fact:
2016-17: Strong - NAO signal/ Weak -NAO Winter .. Yes
2015-16: +NAO signal/ +NAO Winter ... Yes
2014-15: Strong - NAO signal / +NAO Winter ... No
2013-14: Strong +NAO signal / +NAO Winter ... Yes
2012-13: slight -NAO signal / strong -NAO Winter ... Yes
2011-2012 neutral NAO signal / +NAO Winter ... even
2010-2011: Strong -NAO signal/ Strong -NAO Winter ... Strong Yes
2009-2010: Strong -NAO signal / Strong - NAO Winter ... Yes
2008-2009: weak -NAO signal / weak -NAO Winter ... Yes
2007-2008: -NAO signal / +NAO Winter ... No
2006-2007: +NAO signal / weak +NAO Winter ... Yes
2005-2006: strong +NAO signal / +NAO Winter ... Yes
9-1-2
This index as a value forecasting tool is proven high, although the scientific fundamentals are a bit weaker in my opinion.
Since so many people have asked me about this in the last few years, I thought it a good idea to make a post. The index is a measurement of May 1 - Sept 30 SSTs in the North Atlantic. It's correlated to following November-March NAO/AO (+6 month lag).
The index is a composite of 2 areas in the North Atlantic (blue box - red box). When blue box is cold SSTs, negative NAO Winter. When red box is warm SSTs, negative NAO Winter. For compare, and red box is 65% value of blue box anomaly (so -1 blue +0.65 red is same thing). Visa versa.
The index this year is strong -NAO. Top 10-15 negative index of all predictive years going back to 1948 (69)".
Needless to say, we believe that the case for a predominately -NAO phase this winter is a strong one, and this means that it is more likely than not that the arctic oscillation will also be in its negative mode more often than not. But we can look far above the sea surface of the equatorial Pacific to find more definitive indicators.