Last month's New Home Sales data from the Census Bureau was a real downer.  It showed sharp declines in sales and a puzzlingly abrupt jump in inventory levels.  One thing we used to point out about this data series was its immense margins of error.  These could frequently result in major revisions that ended up painting completely different pictures compared to the initial release. For whatever reason, the size of those revisions remained stable enough in recent years that our coverage hasn't felt the need to point it out as regularly.  Today doesn't necessarily bring that old school feeling back with a vengeance, but there was moderately big revision that brought last month's 591k reading up to 629k (for the month of April). The bigger news is the extent to which May's new home sales crushed forecasts.  The median forecast saw sales coming in at 588k--a far cry from the actual reading of 696k.  When taken in conjunction with last month's positive revision, it completely changes the tone of the long-term trend from one of "volatile reversal" to "gradually leveling off." Granted, big margins of error cut both ways and there's no guarantee today's 696k will remain intact by the time it's revised in a month.  That said, it could go higher or lower.   As always, keep in mind that the new home market is significantly smaller than the existing home market.  That one still conveys more of a "volatile reversal," albeit to levels that are in line with pre-pandemic highs in 2016-2018.