true.
but the case-in-point is that people make decisions on players based on what they did in the previous season.
so if a player had an usually high-or-low a2 season, we may be able to correct that number to a more reasonable level.


I'm thinking of setting up some benchmarks of expectations for types of players:

Example:
BEST CASE: Player likely for Assist Increase? (Reason: He had an unlucky, low A2 season last year)

0%-20% A2: Regardless of player type, player has too few second assists and should be flagged as VERY LIKELY for overall assist increase. (ex. Corey Perry, Drew Stafford)
20%-30% A2: Passing should be player's top offensive skill. (If true, increase player's forecasted assist total significantly from original estimate)
30%-40% A2: Player should be primarily a passer. (If true, increase player's forecasted assist total mildly from original estimate)

40%-50% A2: Common ratio of second assists

50%-60% A2: Player should be primarily a scorer. (If true, decrease player's forecasted assist total mildly from original estimate)
60%-70% A2: Scoring should be player's top offensive skill. (If true, decrease player's forecasted assist total significantly from original estimate)
70%-100% A2: Regardless of player type, player has too many second assists and should be flagged as risk for overall assist decrease. (ex. Dan Boyle)

WORST CASE: Player likely for assist decrease? (Reason: He had a lucky, high A2 season last year)