"Last year, my model was high on the Cleveland Cavaliers (ranked second in the Eastern Conference) and the Oklahoma City Thunder (the highest overall projection) and appropriately liked the Brooklyn Nets and the Portland Trail Blazers to be more competitive than expected while putting the Philadelphia 76ers and the Phoenix Suns lower than conventional wisdom (but still not as poor as they actually finished, entirely out of the play-in tournament for both teams)."
"To project teams, I start with player ratings based on a combination of my SCHOENE stats-based projections and luck-adjusted regularized adjusted plus-minus (RAPM) provided by Krishna Narsu that covers the past three seasons. For each team, I project games played based on those missed to injury over the past three years and current absences then subjectively guess at the distribution of playing time."
The model rated the Cleveland Cavaliers second in the Eastern Conference and the Oklahoma City Thunder highest overall last season, also favoring the Brooklyn Nets and Portland Trail Blazers over expectations while ranking the Philadelphia 76ers and Phoenix Suns lower. The recalibrated forecast achieved the second-lowest root mean squared error and the strongest correlation to actual wins and losses among projections scored at the APBRmetrics forum. Player ratings combine SCHOENE projections with luck-adjusted regularized adjusted plus-minus (RAPM) spanning three seasons. Games-played projections use three-year injury history and current absences, with subjective playing-time distributions producing expected win totals comparable to ESPN BET over/under lines.
Read at ESPN.com
Unable to calculate read time
Collection
[
|
...
]