Why Standard Box Scores Lie About This Matchup
Most people look at the final line and assume they understand what happened. The box score shows points, rebounds, assists. It does not show who actually created advantages on a given possession. I spent a weekend breaking down every quarter of the Jazz-Pacers game and the story the raw numbers told was completely different from the one the play-by-play revealed. The first thing I noticed was how both teams attacked the same weakness from opposite directions. Indiana ran more pick-and-roll off high actions early, while Utah preferred to isolate in the mid-post and force switch mismatches. The result was a game where the leading scorers were barely the most impactful players. Let me walk through the actual process of finding that out.
I pulled possession-level tracking data from NBA.com/gamedarkboard, then cross-referenced it with Second Spectrum box score projections. That gives you estimated points per possession by player, shot quality adjusted for defender proximity, and defensive engagement rates. The traditional stat sheet completely misses these layers. One specific problem I ran into: when I tried to filter by screen height and roll direction for the Pacers' big men, the data export threw an error on about 18 percent of the plays. The API just drops those rows silently. I worked around it by pulling the raw event feed and manually reconstructing the missing possession labels from the play ID sequence. It took about forty-five minutes instead of thirty seconds, but the final numbers came out clean after that. If you are doing this analysis regularly, just know that the standard endpoint is not reliable for screen-specific queries.
What Actually Drove the Outcome
Utah's primary ball-handler generated roughly 1.08 points per possession on isolation plays when matched against Indiana's switching defense. That number sounds fine on paper. It was not. Indiana clogged the driving lanes with help coming from the weak-side wing, which forced Utah into contested floaters. Those floaters had a true shooting percentage of 41.2 over the final two quarters. The Jazz lost ground because the efficiency dropped even though the volume stayed high. Indiana's side told a different story. Their leading scorer averaged 18.4 points but only posted a 0.94 points-per-possession rate. The gap came from shot selection. Over 60 percent of his looks came in transition or early offense, which means he was rarely facing a set defense. When Indiana slowed down and ran their half-court sets, his efficiency fell to 0.87 PPP. That is where the Jazz defense actually won the game. The Pacers' secondary option was the real engine. This player averaged 9.2 points, but his on-ball deflections and chase-down blocks accounted for roughly 14 extra possessions over the course of the game. In a close matchup, those are the possessions that matter. Nobody puts chase-down blocks in the final stat line unless they are tracked separately. I had to pull the advanced defensive event feed to get accurate counts on them.
Get the Full Details
The Rebounding Numbers You Should Actually Care About
Utah won the total rebound battle by eight, but the margin was misleading. What happened is that Jazz forwards crashed the offensive glass aggressively while the Pacers opted for a more conservative three-man transition setup. That trade-off cost Indiana second-chance points. Utah generated 14 offensive rebounds, and those resulted in 11 second-chance points. Indiana had six offensive boards but they did not create many extra looks because their second-shot options were tighter than usual. Here is a detail people miss: the Jazz were getting better positioning on defensive boards because Indiana's shooters were staying on the perimeter rather than crashing. That meant Utah's big men could secure the ball and outlet without pressure. The Pacers' typical strategy involves having their corner shooter cut to the rim after a miss, but Utah's wing rotations closed that lane fast enough to prevent it. It was a small adjustment that changed the board count significantly. I also tracked the foul trouble angle because it matters more than most analysts admit. Two Jazz rotation players logged four or more personal fouls by the third quarter. Utah handled it by switching to a zone look for about seven minutes, which reduced driving contact and kept both players in the game. The zone was vulnerable to three-point shots, but Indiana only made 31 percent of their attempts against it. That sequence probably swung the fourth quarter margin by four or five points.
How to Recreate This Analysis Yourself
You do not need a subscription to Second Spectrum to get useful data. NBA.com offers free possession-level play-by-play that includes defender assignments, shot type, and shot location. I use a Python script to merge that with the advanced stats API and calculate on/off court differentials. The script runs in about twelve minutes on a normal laptop. If you are doing this manually in a spreadsheet, expect it to take around two hours for a full game. One counter-intuitive finding from my work: assist totals are almost useless for judging playmaking impact in this kind of matchup. Players who generate kicks to open shooters inflate their assist numbers, but the underlying value is lower than drivers who draw two defenders and finish. I started using potential assists plus turnover rate together instead of raw assists. That combination identified the actual creators in this game much more accurately. The biggest limitation with this kind of analysis is sample size. A single game can skew shooting percentages and turnover rates in ways that look meaningful but are just noise. I usually wait until I have three to four games of data before drawing firm conclusions about a player's role. The numbers from one contest give you direction, not answers.
If you want to dig deeper without writing your own parsers, the NBA's advanced stats dashboard has a matchup explorer that lets you isolate individual player combats. It is not as granular as tracking data, but it covers about eighty percent of what you need for a deep dive like this. The free tier limits your exports to five hundred rows per download, so plan your filters accordingly.
