I would exercise caution about using summertime NAO as a means to predict the wintertime NAO. I looked back at the NAO index from 1989-2018 and of those 30 years, just 13 had a pos/pos or neg/neg relationship, which is good for an r-value of just 0.09. Aka no correlation.
Additionally, there have been 9 summers prior to 2019 (1993, 1998, 2000, 2008, 2009, 2011, 2012, 2015, 2016) where the NAO index has averaged less than -1. Among those years, a -NAO in DJF was observed in only 2 of the subsequent winters.