What I really, really want to do and have been wanting to do for nearly the past decade is take the daily values of the NAO, AO, PNA (and whatever other indices daily values are available for) and create like a weekly or bi-weekly dataset. I know looking at height fields is the way to go, but the raw numbers do provide some value...one of the values is you can quickly see transitions then view that period visually. Doing so though is not as easy as just adding up two weeks of values and dividing by 14. I'm thinking of emailing someone at CPC for perhaps some input.
Anyways with the seasonal forecasting discussion, I don't think it gets the respect it deserves. In reality, seasonal forecasting can be more accurate than forecasting 3-days out. Utilizing all the data available, understanding climo, and how the atmosphere sorta of works can provide tremendous insight into how the atmosphere may evolve moving forward...and there is tremendous success in doing so. Only issue is when someone forecasts a "great" pattern 7-weeks out and that pattern doesn't produce snow it's called a bust...even though the pattern verified. It's one thing to forecast patterns, but that pattern producing is dependent on numerous other factors which aren't necessarily tied into the overall pattern.