Michael Sweetney: Big Mike’s Numbers and the Analysts Who Love Them
The foundation of the statistical analysis revolution in sports is the fact that subjective impressions are not sufficient measures of a player performance. Objective measurements, usually in the form of statistics, are needed to properly determine value. Using too much subjective impression will either overvalue or undervalue a player. By the basis of their objectivity, statistical analysts (statheads) are supposed to be immune to the rank subjective posturing that afflicts most general managers and sports writers. That statheads are impartial observers is itself a hypothesis, which like all scientific hypotheses must be tested against the evidence. For that end, let us consider the stathead commentary on our favorite misused Knick of the past three years, Michael Sweetney, a.k.a. Big Mike.
Just the very fact that I appropriately used the word “favorite” to describe Sweetney is telling in as much as it is accurate. First, take a great player like Lebron James. His talent is so obvious and properly reflected by the scorecard statistics that there is little in the way of evaluation a more advanced statistical analysis provides. On the other hand, Sweetney is widely viewed as a toad: short, fat, and slow. Therefore, statheads like you or me love Big Mike because it gives us a chance to prove our hypothesis: “Subjective impression is insufficient to gauge player worth, so we need objective measurements.” Big Mike validates our scientific enterprise because we “know” he’s a productive player, even if nobody else can see past his limitations.
In a sense too, we statheads are rooting for an underdog, seeing in Sweetney his inner prince .
Accordingly, statheads are willing to look past Sweetney’s warts: he is a poor open court player, draws too many fouls, and does not rotate well on defense. These are all real concerns in the current ecology of the NBA which favors quick perimeter players. But staheads still stare at his steadfastly efficient production as a scorer and rebounder and insist he has value.
Last season, while we were ruing the Knicks’ poor usage of Sweetney, not much was being said of the undervaluing of their best player, Stephon Marbury. That statheads would ignore Marbury’s Top-3 point guard PER (just a hair behind the league MVP Steve Nash) to complain that he “dominates the ball too much” is a curious case of selective judgement. Compare the two: Sweetney is a statistical monster, who upsets aesthetically, and Marbury is a statistical monster, who upset aesthetically. But statheads have been much more vocal in support of Sweetney than for Marbury.
The reason for this asymmetrical commentary is strictly subjective “liking” of a player (which admittedly was the motivation for why I wrote my first piece on Marbury). This author was outright flabbergasted at the subjective criticism levied against Stephon Marbury by statheads in the face of his outstanding statistical performance. As statheads we laugh at labeling a productive player like Sweetney as useless for being lumbering and oafish. However, we then turn around and bemoaned Marbury’s inability to improve teammate performance, even if we should know better. By our own advanced metric standards of Plus/Minus, Marbury made the Knicks 12 points better per 48 minutes, easily ranking him as a league leader in that category.
By our own standards, the criticism of Marbury’s cancerous effect on team play is completely unjustified.
An update on Sweetney’s performance demonstrates another limitation on statistics: They are for the most part reactive. They tell us what happened in the past, but even our informed opinions on the future are still educated guesses. Statheads expected Sweetney’s performance to steadily improve, thrusting him into the Top-10 Power Forward plateau. Unfortunately, much to our chagrin, he has regressed, now posting a PER as slightly below league average.
This PER depression is largely due to a dramatic plummet in TS%. Sweetney was a monster low-post scorer last season, but his Field-Goal percentage has sunk inversely to his weight. Sweetney’s foul rate was expected to decrease as he got older and saw more regular minutes, but that hasn’t happened either. One promising indicator is his turnover rate declined with increase usage, though that is tempered greatly by his lowered shooting efficiency. In all, we should take Sweetney’s unique player card and file it into our database in order to improve our models and hypothesis. The regression is especially alarming because Sweetney is short for a frontcourt player and those performers have historically had shorter (no pun intended) careers with quicker peaks. At this stage it might only be wishful thinking, and not statistical indication, to believe he will ever move into an elite tier of power forwards.
Statistical analysis does greatly improve the evaluation of player performance, but like any other science it must maintain its discipline to be both credible and effective. For that matter, we cannot only point fingers at the subjective media for filling their columns with mindless ruminations: we must also be vigilant in policing ourselves. There should be no rooting in the press-box, nor in the regression model.
Of course, we can in our own time take off our stat thinking hats too and place Sweetney’s framed player card atop our mantle, remembering fondly how on those horrifically bad Knicks teams sometimes the only entertainment was his periodic hip checking of seven footers out of the lane.