Oct 28, 2014; New Orleans, LA, USA; New Orleans Pelicans forward Anthony Davis (23) dunks over Orlando Magic guard Evan Fournier (10) during the second half of a game at the Smoothie King Center. The Pelicans defeated the Magic 101-84. Mandatory Credit: Derick E. Hingle-USA TODAY Sports
So Anthony Davis had a night on opening day, throwing up 26 and 17 with 9 blocks (more on that in a moment) and 3 steals for good measure against the outmanned Orlando Magic. First, nobody has put up a 25+/15+/7+ line since Dwight Howard in 2010 (and then Howard again on the same date in 2009.) Second, I’m already taking a victory lap for declaring him the third best player in the NBA for this season, and until more games happen you can’t prove me wrong.
More interesting was a mini-debate about Davis’ 9 blocks. Depending on who was doing the counting, between one and four of them were “phantom” blocks. Fortuitous scorekeeping is not exactly a new phenomenon, (see the mini-controversy surrounding “The Nick Van Exel Assist Game“). Part of the reason I don’t completely trust the eFG splits between assisted and unassisted shots is the specter of the Hometown assist which would make assisted shots appear more efficient, as the more “generous” the crediting of an official assist, the less likely SportVU would be to categorize it as an “assist chance” if the shot missed.
This factor isn’t limited to assists. Other subjective stars (meaning blocks and steals though occasionally rebounds as well) can be inconsistently awarded, especially for a highly marketable star like Davis. Last season Connor Peterson of Georgetown Sports Analysis took a deep dive into the difference in Davis’ shot blocking prowess at home versus on the road:
"When the Pelicans are playing on the road, Anthony Davis blocks 2.2 shots per game. Of the 48 blocks Davis has recorded on the road, Davis has only had five questionable blocks, which means that about 10.6% of his blocks on the road are questionable. This error rate is not necessarily a cause for concern, for blocks are not always clear unless the statistician is physically on the court, which he is not. Furthermore, it can be hard to determine if a shot was blocked or simply altered, which does not count as a block. Adjusting his stats to exclude questionable blocks, Davis is averaging 1.9 blocks per game on the road.At home though, his numbers are far different. Davis is averaging 4.1 blocks per game at home. This alone is a red flag, but could potentially be attributed to the fact that Davis might just play better at home. However, out of Davis’ 86 blocks at home, 19 of them fall into the questionable category. In this case, 23.2% of Davis’s blocks at home are not actually true blocks, much higher than the 10.6% on the road. If we adjust his stats to account for questionable blocks, Davis is only averaging 3.3 blocks per game at home. Adjusting his total number of blocks gives a grand total of 2.6 blocks per game, a half of a block lower than his current average."
Peterson went on to find that though this phenomenon wasn’t unique to Davis, the Pellies were perhaps most guilty of this type of “enhancement.”
2013-14 Teams with highest BPG – Chart courtesy of
This isn’t an attempt to take a run at official scorers, who by and large do a great job. Rather, it in some way explains why people have been so desirous of moving beyond the box score stats in evaluating players. If the input stats are garbage, you get garbage out, as the acronym goes. Especially in the case of evaluating defense where slightly variations in relatively rare accomplishments, like blocks and steals, might have an outsized and often misplaced influence on box score-based analysis of defensive ability, there has been a desire to do better.
In many ways, this explains the rise of complex statistical modeling such as ESPN’s Real Plus/Minus and other Adjusted Plus/Minus-style metrics. Something is happening on the floor. The traditional counting stats aren’t picking it up, so more involved techniques are required to try to parse out those effects, leading to efforts to measure a players’ holistic impact while on the floor, controlling for teammates and opponents (the basis for any APM model).
The advent of detailed tracking and charting such as offered by SportVU provides the possibility of “a better box score.” Among all the differences between baseball and basketball analytics is how relatively good the very basic baseball stats are at recording the important events in the game. Though it takes some data-entry and coding legwork, generating a frightfully useful stat like BABIP (batting average on balls in play) requires only the scoresheets from the relevant games. The historical record of basketball contains no such treasure trove of information. We are left to guess how many blocks Bill Russell would have recorded had the stat been kept in his day.
To some degree, new tracking data eliminates much of this problem of lost data going forward. As one analyst with an NBA franchise told me last season: “[APM-style metrics] are great if you’re trying to infer something for which there is no data. They’re absolutely useless if you have some or all of that data. Player tracking identifies specific defensive stuff, which obviates the usefulness of APM.”
While teams have access to a much greater range of SportVU data than is in the public domain, more data is still better than less data. We are no longer tied to the false binary of “did Anthony Davis block that shot, or no?” In the following sequence he was awarded two blocks. Thee first one might have been generous, or he might have gotten a finger to Elfrid Payton’s flip:
The question is, does it really matter? It’s fair to say whether or not he “blocked” the first attempt, Davis altered it.
With SportVU in place there is data much more finely-tuned than simply counting shots blocked versus not. Over a decently large number of shots, a defender’s overall impact can be measured. Last year, opponents shot 48.8% at the rim on shots Davis contested. Does it make any sense to arbitrarily award him more credit for blocking beyond that attempt having a 0.0% chance of going in? Further, moving away from just using blocks alleviates the problem that all blocked shots aren’t created equal. Blocking a midrange jumper is generally both surprising and spectacular, but the NBA as a whole shot 39.4% on midrange shots last year, while completing 60.8% at the rim — blocking a layup is vastly more impactful for a defense than blocking a an 18-footer.
But by focusing strictly on Davis effect around the rim, his actual defensive impact becomes somewhat more clear. By a number of metrics, Davis wasn’t a great defender last year. He was perfectly decent for a second-year player, and all indications were, and continue to be, that with his physical gifts, his defensive ceiling is stratospheric. He’s not there yet however. That 48.8 FG% allowed at the rim was only a touch better than league average for bigs (right around 50% for all NBA big men), but more importantly whether through a New Orleans scheme that had him chasing ball-handlers well beyond the three-point arc in pick-and-roll coverage or through his own lack of recognition and rotation (or most likely a combination of both) he only contested 28% of opponents shots at the rim, well below league average of around 38%. This was reflected in the team’s defense where per NBA.com, New Orleans allowed the sixth largest proportion of opponents shots at the rim, while having the fifth worst percentage allowed, a percentage which was actually worse with Davis in the game.
This snapshot at one area of defense doesn’t prove anything beyond that reference to just shot blocking doesn’t tell nearly the entire story of Davis’s interior presence last season. Nor does the question of whether he blocked 5, 7 or 9 shots in the opener say much about how he performed. At least not when compared to the fact that he allowed 45.5% shooting while defending 39.3% of available attempts, figures which if they held up over the full season would make him one of the more impactful rim protectors in the NBA.