In my opinion, filtered indexes are one of the unsung heroes of 2008+. They can significantly reduce IO when the column is part of an oft-used predicate.

As previously mentioned, relying on autostats to maintain the statistics will only result in threshold changes at the TABLE level, not the column level. For slowly changing, large, historical tables, statistics will only be an issue over time. More than likely, if you have a defrag plan on your indexes, the stats on the index will be updated by the index rebuild process.

Thanks for re-enforcing my belief patterns.

We use enterprise almost exclusively but the STANDARD edition point is worth the reminder. PARTITIONING is enterprise+.