Winter '17 / '18 - North Atlantic Oscillation Analog Years

To date ... 1990 is the leading analog year for the North Atlantic Oscillation (NAO) in the run-up to the coming winter.

The average NAO for Winter '17 / '18 would turn out positive should this come to pass.  The analog forecast reaches winter's NAO/s minimum in DEC and its maximum in FEB before declining at the start of meteorological spring.

One problem.
Winter '90 / '91 ENSO/s signal was warm La Nada (+0.4°C ENSO Region 3.4 temperature anomaly) yet a weak La Nina appears in the cards for this year and unlike this year ... the corresponding PDO analog year was negative (this year it's positive; trending negative) and QBO was west-positive (this year it's east-negative and trending same).

And herein lies the problem with analogs.
Analog years don't exist in a vacuum.
'94 / '95 moderate El Nino; cool PDO trending warmer; QBO east trending west [REJECTED]
'54 / '55  moderate La Nina; cool PDO trending cooler; QBO east trending west [REJECTED]
'04 / '05 weak El Nino; PDO cool trending warmer;  QBO west trending west [REJECTED]
/73 / '74 strong La Nina; cool PDO, QBO west [REJECTED]

Other teleconnection indexes frequently confound the potential predictive value of a particular index.

---

Other correlation-type analysis techniques can also provide insight into the likely state of important winter teleconnection indexes.

Analysis of the North Atlantic Oscillation (NAO) index's period-of-record reveals a statistically significant association (Chi-square Test for Independence; alpha = 0.05) between its sign for the month of NOV and the sign of the average NAO index during a subsequent meteorological winter (D-J-F).

Period-of-record correlation analysis finds if NOV/s NAO is positive (negative) ... then there's a 73% (54%) probability the winter's average NAO/s sign will also be positive (negative).  The analysis false alarm rate is 27% (46%).