From the GM’s perspective, analytics play an important role in player evaluation and roster construction. Whether you’re considering potential draft picks, free agent acquisitions, or trade options, the numbers help you identify your team needs and who might be able to fill them.
But analytics certainly aren’t everything. I find that analytics are great for describing what happens on the court but often fall short in explaining why. For example, the numbers may show that a particular player is an outstanding finisher around the rim, but they won’t tell you whether that’s the result of having the strength to overpower opponents, the quickness to blow by them, the size to shoot over them, the explosiveness to rise above them, or the craftiness and touch to finish around them. And the numbers won’t tell you to what extent that player’s apparent finishing ability is merely the result of having skilled teammates who draw defensive attention and set him up for easy layups and putbacks.
When evaluating a prospect and trying to decide whether his production will carry over to the next level, the reason why that player is successful makes a big difference—and you need traditional scouting in order to make that determination. Therefore, basketball analytics complement traditional scouting rather than replacing it. The data enables scouts to focus on more targeted questions, like why a player excels in particular areas as opposed to more broadly identifying and describing a player’s strengths and weaknesses. In turn, a scout’s report may lead to a deeper dive into the numbers or a different interpretation of them. It’s a constant back-and-forth.
Basketball is simply too complex and cooperative a sport for statistical analysis to be conclusive in and of itself. I believe this explains why we don’t hear of a “stats versus scouts” conflict nearly as much in basketball front offices as compared to baseball. When baseball teams started hiring analysts, they fired scouts. Basketball teams didn’t fire scouts when they started hiring analysts. If anything, there are more traditional scouts now. You can’t meaningfully evaluate how a basketball player’s production will translate to a new environment without watching the player and understanding the context in which that production occurs. And given how much basketball relies on teamwork, it’s especially important to see how a player functions as a member of the team, from communication to execution of team concepts and court awareness, or basketball IQ.
The parties most critical of basketball analytics tend to be accomplished former players and the media, as we might expect, since these parties have reason to be somewhat dismissive of new methods. Established members of the sports media have an interest in preserving their positions as authoritative experts, and any outside source of knowledge is in some way a threat to that authority. Accomplished former players have an interest in preserving their legacies. Arguably the most outspoken critic, Charles Barkley, falls into both categories.
Interestingly, Charles Barkley the player actually fares quite well by some of the newer advanced stats, often even better than he was perceived during his playing career. By Box Plus-Minus (BPM) he rates as the league’s best or 2nd best player 6 years in a row (typically #2 behind only MJ), with 10 years in the Top 5. He similarly has 9 years in the Top 5 by Player Efficiency Rating (PER), compared to just 4 years in the Top 5 by MVP voting. Barkley seems to criticize analysts’ emphasis on shot locations and efficiency, though largely by virtue of getting to the basket and the free throw line so often, Barkley led the league in scoring efficiency (True Shooting %) 4 years in a row.
Barkley’s exceptional standing according to these newer metrics in part results from offensive stats being more accurate and reliable than individual defensive stats. Barkley was perceived as a great scorer and rebounder, and the advanced stats certainly bear that out. He’s #8 all-time by scoring efficiency and #17 by total rebound rate. His relative weakness came on the defensive end, at least according to perception, though neither BPM nor PER docks him there. Rather, by BPM he ranks 90th all-time defensively (+1.8 points per 100 possessions), likely as a result of his high steal and defensive rebound rates.
Even using the current play-by-play and player tracking data, individual defense seems to be much more uncertain and context-dependent than individual offense. If, for example, a player is a high-volume, high-percentage outside shooter on his current team, that ability should translate fairly well to a new environment as long as he occupies a similar role. If, however, that player rates well in defending the pick-and-roll, I wouldn’t be so confident that those numbers will transfer to another team. Defensive metrics are often more a reflection of the team’s overall defensive ability and scheme than a fair representation of individual performance. Garrett Temple and Robert Covington are examples of players who’ve rated poorly defensively according to Synergy stats even though they’re active, athletic defenders who greatly improved their team defenses. Their individual numbers suffered because their teammates didn’t rotate or protect the rim, but that doesn’t make them poor defenders.
Coaches also use analytics to gauge what’s working and what isn’t over a particular period of time, as well as to evaluate opponents’ strengths, weaknesses, and tendencies. Maybe the stats show that your team’s getting crushed on the offensive glass or badly outscored with a particular lineup pairing on the floor. It’s then up to the coaching staff to figure out why that’s happening and adjust accordingly. The coach might decide to put a different lineup on the floor, change tactics, or even keep everything the same, depending on his or her assessment of the situation. Analytics assist the coaching staff’s preparation and identification of issues, but the coaches still have to determine how meaningful that information is and how best to use it.
Leave a Reply