Which Receiver Stats Actually Repeat, Ranked
Somebody has measured the year-to-year persistence of nearly every receiver metric in common use, and the results reorder most of what gets cited in player analysis. The stat you would most want to be stable is near the bottom.
The table
The top of the list: slot rate at 0.75, targets per game at 0.70, fantasy points per game and receptions per game both at 0.68, yards per game at 0.67, average depth of target at 0.65, targets per route run at 0.64.
The middle: open score 0.59, first downs per route run 0.58, average depth of completion 0.57, yards per route run 0.56, yards after catch per reception 0.48.
The bottom: touchdown rate 0.19, EPA per target 0.18, drop rate 0.14, contested catch rate 0.02, route rate 0.01.
What the top of the list has in common
Role. Slot rate is the stickiest thing about a receiver because where he lines up is a coaching decision that rarely changes, and targets per game is next because a role in the offense persists.
That is a slightly deflating finding. The most predictable things about a receiver are facts about how his team uses him rather than measures of how well he plays.
The source is careful about what follows from this. Stickiness is not the same as predictive value for future fantasy points. A metric can repeat reliably and still not forecast production, and slot rate is exactly that: highly stable and not a performance measure at all.
What the bottom of the list destroys
Drop rate at 0.14 and contested catch rate at 0.02 are the two figures worth sitting with, because both get used as character assessments. A receiver gets labelled as having bad hands or being a contested-catch specialist, and the data says neither label persists from one season to the next.
The source puts it plainly: there is no correlation year over year for these. A receiver who dropped a lot of passes last year is not meaningfully more likely to drop them this year.
Touchdown rate at 0.19 belongs in the same category and explains why touchdown-driven seasons regress so reliably.
How to read a receiver differently
Lead with the role measures, because they are the ones that will still be true. Treat efficiency measures as descriptions of what happened rather than predictions. Treat drops, contested catches and touchdown rate as noise unless the sample is very large.
Applied to any single game, including a week that wins a fantasy matchup, the components divide cleanly. Whatever came from targets and alignment is likely to recur. Whatever came from touchdowns and contested catches is the part that was borrowed.