Disappointing to see a trough. If it actually rains moderately all week, fine. But just for a few sprinkles and no thunderstorms. No bueno.

In my own research I've noticed some big differences between using 51-10 and 91-20. The thing is, those 13/14 of the analogs included in 51-10 are part of that average. It's rounded. Sometimes trends start happening, especially in shorter time frames, and being outside of the "base period average" can give a little skewed results. I always try to use the base where most of my analogs are located within, otherwise you may just be showing "51-10 vs 91-20".

The whaler is off trimmed

I would think base state for the anomaly period would be more important than number of analogs within certain years...but again, all signs point basically the same way so it doesn't matter all that much which base period you go with in this case. It's only if you look just at the past 10-15 years that there is a strong dry signal - but that is independent of ENSO over that period. And obviously very few analogs.

Yeah the AIFS is my #1 LR forecasting tool now, consistently outperforms other ensemble forecasting systems, including the EPS.