There is a mythical quality to the perfect sports statistic: a single column which can be the arbiter of any school yard arguments about which player is superior. A single column that combines all of a players contributions, positive or negative, and produces a rating based on their accomplishments. The digestible nature of all-in-one ratings is what has made them a hit with fans across all sports, most notably in the MLB with Wins Above Replacement (WAR) and Quarterback Rating (QBR) in the NFL. The NBA's foray into all-in-one ratings was led by writer and analyst, John Hollinger, with Player Efficiency Rating (PER). Hollinger's goal was to create a scale that measured relative performance between players. With PER totals scaled relative to 15 as the average, totals from 25-35 suggest the player is an MVP candidate. PER is based on the traditional box score statistics: points, rebounds, assists, steals, and blocks, with turnovers and personal fouls detracting from the formula. PER wanted to be the solution to any basketball argument, but ultimately wasn't able to be represent the game well enough.
| Player | PER | BPM |
|---|---|---|
| Giannis Antetokounmpo | 30.90 | 10.80 |
| James Harden | 30.60 | 11.70 |
| Anthony Davis | 30.30 | 8.50 |
| Nikola Jokić | 26.30 | 9.50 |
| Karl-Anthony Towns | 26.30 | 6.80 |
| Joel Embiid | 26.10 | 4.10 |
| Kawhi Leonard | 25.80 | 5.00 |
| LeBron James | 25.60 | 8.10 |
| Nikola Vučević | 25.50 | 6.40 |
| Boban Marjanović | 24.90 | 1.40 |
PER has a few notable flaws that have caused basketball analysts to innovate upon it. For an “all-in-one” rating, PER does not accurately represent defensive contribution. This is due to the box score variables that compose the rating not capturing a complete defensive picture. Steals, defensive rebounds, and blocks are the only variables used in PER that indicate defensive performance, and while they are notable occurrences on defense; they don't necessarily correlate with the most important part of defense: preventing the opposing team from scoring.
MVP winner Giannis Antetokounmpo tops the PER chart in 2018-19.
Players prioritizing blocks, rebounds, and steals are more willing to be out of position or gamble. A relevant example of this would be Russell Westbrook. Westbrook's deadly combination of athleticism, height, motor, and passing ability has enabled him to stuff the box scores with dazzling figures. Westbrook is only the 2nd player in NBA history to average double digits in 3 categories over the course of a season: points, rebounds, and assists. Critics argue that Westbrook's box totals is a result of him being out of position and a desire to pursue personal accomplishments over team success. Westbrook has been known to defer easy baskets in favor of an unnecessary pass, or grab rebounds in transition when that role is typically filled by a center. According to John Schuhmann, a writer for NBA.com, in 2017, Westbrook led the league in uncontested rebounds with 8.6 per game. The Thunder's offensive strategy was based on starting transition with the ball in his hands. Westbrook's on the court playstyle provides support to the notion that box score statistics do not perfectly represent on the court performance.
Russell Westbrook is a gifted rebounder, but could that tenacity be better spent on other aspects of defense?
GIF Source: SBNation
The problem of unreliable box score causation is not exclusive to PER, but more broadly applies to any rating derived from traditional box score statistics. Even a more modern approach, such as Box Plus/Minus (BPM), which uses advanced box score statistics are able to better assign causality, still fail to accurately evaluate defensive performance. The contributions of strong defensive players are unaccounted for and their rankings are devalued in the models. Players which specialize in defense such as Ron Artest, Tony Allen, or Bruce Bowen all qualify as average, despite earning NBA All-Defense first team honors in their respective years.
So far, it may seem as though box score statistics are a relic of a less data driven and analytical NBA, but that doesn't meant they don't have their uses. In fact, being a relic of the past may actually be considered a positive. This is due to the intersection of two unrelated threads: the time variant nature of basketball rules and the "GOAT" discussion.
"GOAT" is a very common acronym used in sports media and by sports fans which stands for "Greatest of All Time". In sports fandom, there is constant desire to evaluate, and re-evaluate, where great players stand in the historical context of the sport. While these discussions serve as decent watercooler banter or an inoffensive sports radio topic; it is impossible to have objective answer due to changes in strategy, training, opportunity, and stat tracking that have occured over time.
The box score, as it is known today, was only implemented in 1974, when scorekeepers began tracking steals, blocks, offensive rebounds, and turnovers. The three point line, an integral part of the modern game, was only introduced in 1979. How do you compare the best shooters of the pre-3pt era to current players? You could measure a shooters performance relative to league average, but that still doesn't account of evolving offensive philosophy and training. Larry Bird is widely considered one of the best shooters of all time, yet he only averaged 1.9 3pt shots over the course of his career. Stephen Curry, another claimant for that title, has averaged 8.2 3pt shots per game through his career so far. These changes are what makes comparing players from different basketball eras difficult, especially relative to a sport like baseball which has seen very little change in strategy or scorekeeping philosophy over time.
Wins Above Replacement measures how many wins can be calculated for a baseball player like Ty Cobb, who played from 1905-1928. For the nba, there are too many blind spots to calculate an accurate evaluation of player like Bill Russell, who played from 1956-1968. The approximation that Basketball Reference uses is a heavily modified version of Bill James' Win Share system. Since very little surviving footage exists from the early days of baseball and basketball, the scorekeepers' boxscores are the only connection to a past era.
Left: Bill Russell and famed coach Red Auerbach. These two have 27 championship rings combined.
Right: Ty Cobb and Honus Wagner were two of baseball's earliest stars.
Although NBA analysts on the cutting edge have abandoned all-in-one ratings in favor of much more detailed optical and play-by-play tracking, box score based models are not obsolete. All-in-ones still have the power to bridge gaps between generations and represent events that have been lost to the sands of time. While each rating has its own set of quirks and flaws, these only create opportunities for basketball enthusiasts to improve upon the past mistakes of analysts. In my next article, "All-in-one Basketball Ratings Don't Work, But There is Room for Improvement" I create my own all-in-one basketball rating and explore the strengths on weaknesses of the model.
