(This post series is the first contribution to 8 Points, 9 Seconds by Kevin Hetrick. He has previously written for the ESPN TrueHoop Network site Cavs the Blog as a “draft expert.” Originally from Ohio, he has lived in Indianapolis for about 10 years and recently took a keen interest into what makes the Pacers tick.)
Looking at the Pacers, it is easy to see the makings of a good team. They are young, have cap flexibility and play at a fast pace. They should continue to be a team on the rise. With George Hill in tow, they have six players under contract who most NBA analysts would consider viable members of a quality eight-man rotation (Danny Granger, Hill, Darren Collison, Paul George, Tyler Hansbrough and Roy Hibbert). Most are 25 or younger and Granger is the senior-citizen of the bunch at 28.
This core can likely keep the Pacers as a 45 to 50 win team for several years.
In this post series, we will be analyzing exactly how the team can improve on its 2010-11 performance. But to start, we won’t be looking at the Pacers. We will first look at some of metrics that we’ll be using throughout the analysis.
The goal in every basketball game is to score more points than the opponent. It’s a simple game. And the best way to measure the effectiveness of a team at scoring and defending is offensive and defensive ratings, aka, points per 100 possessions (pts/100). This in and of itself doesn’t tell you how a team scored or defended effectively — it just tells you whether or not they did.
To find out the how, we look at a few other metrics. Specifically, we’re going to look at how well these other metrics have correlated with pts/100. (Correlations, and all trend data in this article series, are based on leaguewide stats from the past three seasons). The higher the correlation, the more likely it is that an improvement in the statistic will lead to improved rating. (A perfect positive correlation is 1.0, a perfect “negative correction” is -1.0 and something found to be independent is 0.) For instance, making a team shoot a low FG% is a better way to become a better defensive team than blocking a lot of shots is. Thus, defensive FG% has a correlation much closer to 1.0.
By far, the highest correlations to offensive rating are the shooting percentage statistics. Over the last three years, effective field goal percentage (eFG%, which is basically FG% that accounts for the fact that three-pointers are worth three points) has correlation of 0.864, true shooting percentage (eFG% that also accounts for FT%) has correlation of 0.859, and good ol’ fashioned field goal percentage has correlation of 0.771.
This isn’t really revelatory; if your team shoots well, they will likely be a good offense. The most interesting aspect may be that eFG% had better correlation than true shooting (this is true with defensive rating also). Logically, many consider TS% to be a better stat since it incorporates the ability to get to the free throw line (and make those freebies). However, teams were (very slightly) more likely to be effective with increases in eFG%.
Something more interesting is where the best offenses shoot from. Of all shooting ranges tracked by Hoopdata.com, percentage of field goals taken as threes had the best correlation with offensive rating at 0.411. This is not shocking; leaguewide, the eFG% for three-pointers is higher than any shot besides those at the rim. So, teams that create good looks from three-point range are more likely to have a good offense than teams that can’t. As this suggests, three-point shooting percentage also has good correlation (0.627).
More surprising is that percentage of shots taken at the rim had negative correlation with offensive rating. The negative correlation was small enough to effectively be independent (-0.083), but a reasonable assumption is that teams shooting more shots at the rim would fare better. This wasn’t the case. Average FG% at the rim has been 60%–64% compared to an average eFG% of around 50%; but oddly, teams shooting more often at the rim are slightly more likely to have worse offenses. Increasing a team’s proportion of shots from 3–9 ft, 10–15 ft, and 16–23 ft is more likely to correspond with decreased offensive efficiency (-0.097, -0.169 and -0.287, respectively).
Assist-related stats also offer some insights. Percentage of possessions including an assist had reasonably high correlation with offensive rating (0.474), but this is tied to other factors. Obviously, teams that don’t turn the ball over and make shots at a high percentage are more likely to have possessions with an assist.
This can be seen in the fact that the percentage of field goals that were assisted only had a correlation of 0.182. An interesting aspect of this is that increased percentages of field goals assisted from 3–9 ft, 10–15, 16–23 ft and three-point range had small negative correlations (from -0.097 to -0.198). Only percentage of assists on field goals made at the rim had a positive correlation (0.253).
These numbers reflect the importance of players who can create shots for themselves. If a team is relying on the offensive system to create a bucket from anywhere outside of three feet, it is likely to result in worse offense. The correlation of percentage of assists for various shooting ranges may also provide some insight into the importance of strong guards and wings, as opposed to big men. This conclusion is based on assisting for easy opportunities at the rim is the only area of assisting that appears to support good offense.
As for other statistical areas, reducing turnovers and getting to the free-throw line (in terms of FTAs per FGAs) had decent correlations of 0.427 and 0.326, respectively. Actual free-throw shooting percentage had minimal positive correlation of 0.169. Pace was completely independent from offensive efficiency (-0.021 correlation). And offensive rebounding rate (ORR) was independent from offensive rating with correlation of 0.025. This doesn’t mean that offensive rebounds aren’t important — just that a great offensive rebounding team is as likely to be a good offense as a bad offense.
In summary, improving a team’s offense is most easily performed by finding players that can create and make shots, first for themselves and secondarily as shots at the rim for teammates. Creating good three-point attempts is important. Reducing turnovers is more likely to result in improved offense than getting to the free throw line. Building a strong offensive rebounding team should be a GM’s last concern.
The other side of the ball not-so-surprisingly has some similar results. Like on offense, forcing opponents to miss shots is most important on defense. The correlation is even higher here, however. Opponent eFG% has correlation of 0.92, TS% is at 0.917, and FG% is 0.906.
One trend that is more pronounced is how well certain shots are defended. Opponent’s FG% at the rim and from three-point range both had very high correlations of 0.712 and 0.706. Opponent FG% from 16–23 feet was important (0.55 correlation), while defending better from mid-range was less so (a correlation of only 0.21).
The range of correlations for shot locations on offense was tightly packed from 0.35 to 0.65; shooting well from anywhere was comparably likely to result in quality offense. The wider range of correlations defensively likely highlights the importance of contesting shots in the paint and closing-out on shooters.
Unlike with offense, there was minimal correlation between reducing three-point attempts and better defense (0.157). But there was decent correlation for forcing opponents to shoot long twos and defensive rating. Percentage of opponent’s shots from 10–15 feet had correlation of 0.515 and percentage of shots from 16–23 feet was at 0.378. In sum, it was more important to ensure opponents shot from this range than to make them shoot a below-average percentage from there.
Defensive rebounding was a lot more likely to result in good defense than offensive rebounding was for offense. The correlation of rebounding to defensive efficiency was 0.649. Percentage of blocks and defensive plays (block, steals, charges) had relatively low correlation with defensive rating (0.314 and 0.37, respectively) while reducing free-throw attempts and forcing turnovers had even lower correlation (0.247 and 0.177).
Finally, passing stats were more indicative of defensive performance than for offense. On defense, lower rates of assisted field goals had positive correlation with defensive efficiency for all shooting ranges. Additionally the positive correlation on defense is higher than the negative correlation for offense. Reducing the percentage of assisted field goals has correlation ranging from 0.118 (16–23 feet) to 0.426 (at rim). On offense, it seems that is important to have players who can create their own shots; and on defense, it appears vital to ensure that opponents are forced to create their own shots. Make of that what you will.
To summarize, on defense it is important to contest shots at the rim, close out on shooters, and control the defensive boards. The correlation of defensive efficiency to FG% at the rim and on threes, and to lower assist rates; demonstrates the importance of the defensive system and quality rotations for defensive efficiency. More traditional individual and team stats like blocks, fouls and turnovers have lower correlation to defensive efficiency than stats reflecting good team defense (shot locations, assist rates). Good one-on-one defense is important, but good rotations, consistent effort and keeping the opponent acting as five individuals instead of as a cohesive unit are more important.
Tags: Stat Talk