Good observation.  Since 2000, ENSO has shifted to lower variability, meaning we don't see the bigger amplitude events that we saw in the 80s/90s.  The models in general have more difficulty forecasting weaker events and capturing the correct onset time.  All those cases of misses/false alarms during OND since 2000, occured with the weaker events (< +/-1degC).  The jury is out on whether this weaker ENSO variability is associated with climate change and it is very difficult to determine with the short lenght of the historical record.  There are certainly papers that project weaker variability but there are also papers that suggest strong variability.