The Virtuous Circle Of Offense and Defense and the NBA Finals

Feb 26, 2015; Cleveland, OH, USA; Cleveland Cavaliers guard Kyrie Irving (2) dribbles against Golden State Warriors guard Stephen Curry (30) in the fourth quarter at Quicken Loans Arena. Mandatory Credit: David Richard-USA TODAY Sports
Feb 26, 2015; Cleveland, OH, USA; Cleveland Cavaliers guard Kyrie Irving (2) dribbles against Golden State Warriors guard Stephen Curry (30) in the fourth quarter at Quicken Loans Arena. Mandatory Credit: David Richard-USA TODAY Sports /
facebooktwitterreddit
Feb 26, 2015; Cleveland, OH, USA; Cleveland Cavaliers guard Kyrie Irving (2) dribbles against Golden State Warriors guard Stephen Curry (30) in the fourth quarter at Quicken Loans Arena. Mandatory Credit: David Richard-USA TODAY Sports
Feb 26, 2015; Cleveland, OH, USA; Cleveland Cavaliers guard Kyrie Irving (2) dribbles against Golden State Warriors guard Stephen Curry (30) in the fourth quarter at Quicken Loans Arena. Mandatory Credit: David Richard-USA TODAY Sports /

A trope common amongst NBA analysts is to suggest a team’s “offense is keyed by their defense” or something along those lines. Quite often, this is a backhanded compliment for an excellent defensive squad which, to put it mildly, struggles to score on the other end. The late model Bulls and Grizzlies are prime examples of this sort of team. On the other hand, some teams and players of supposedly see their defense improves when they see the ball go in the hoop on offense.

The notion of offense playing into defense and vice versa finds its way into all sorts of debates about personnel and game strategy. One of the key points against the “hack-a” strategy of intentionally fouling a weak free throw shooter is that it forces the fouling team to play against a set defense every single time. On the other hand, shooting a lot of three pointers is said to be dangerous because it produces long rebounds which lead to more fast breaks for the opposition. While the mechanism of this criticism is probably incorrect, as the defense appears about as likely to rebound a three pointer as they are a midrange jump shot, the warning against too many threes might have some merit. While a 33.3% three point shooter and a 50% two point shooter will have identical eFG%, the guy shooting twos will make more shots, meaning the opposition has to inbound the ball and play against a fully prepared defense. While this is balanced[1. And perhaps even overbalanced, a topic for future research!] against the increased offensive rebound opportunities it’s part of what makes a three-point heavy strategy so high variance.

Inspired by looking at just how large the

effect of steals on the ensuing offensive possession

, I decided to look at just how much the start of an offensive play or possession[2. Often times, “play” and “possession” are used interchangeably, the key point of differentiation usually concerns offensive rebounds. Typically, an o-board starts a new

play

but not a new possession for most metrics which measure by possession.] changed the expectation of the next shot based on how the ball transferred from the old offensive team to the team formerly defending. The results largely align with common sense, stops lead to score and so on. The chart graph below illustrates leaguewide eFG%[5. It is important to remember for the entirety of this analysis that we are dealing only with plays ending in FGA. Turnovers and shooting fouls are extremely important to determine the precise outcome of a given play type, but for the purposes of a first pass through the data, the most important factor, shooting, is a good place to start.] on shots directly following various occurrences[3. Much as with the steals analysis, an intervening event in the play-by-play was assumed to result in a deadball play rather than a continuation of the post-steal, post-rebound, etc.]:

“Deadball” includes shots coming after non-shooting fouls, deadball turnovers such as traveling out of bounds, or charges, substitutions[4. Indicating the ball was tipped out of bounds with no change in possession], team rebounds and other similar events. Interesting to note it doesn’t appear to make much difference if a free throw is missed or made, the defense is likely set either way.[6. DOWN WITH HACK-A-DUDE!] Further, forcing the now-offense to take the ball out of bounds from the referee or directly from a made basket appears to achieve much the same purpose for the defense. In fact, on plays where possession changes, shots can be divided into two broad types – those where the defense is “set” and “live ball” plays:

For sake of context, the 6.3% gap in eFG% between attacking set defense and playing without a stoppage[11. Something to look into in the near future is removing plays shots taken longer than about 8 seconds after a defensive rebound or steal from the “live” category as at least in the case of steals, the advantage gained from transition appears to dissipate around that time frame.] is larger than the gap between Golden State’s league leading mark and Denver’s 25th best rate. The improved efficiency found through what Kevin McHale called “playing downhill” following Game 2 of the Western Conference Finals isn’t hard to see, and helps explain why eschewing a timeout in such a situation is often the correct percentage play even if it ends in ugliness at times.

Of course, this gap isn’t constant between teams. Here’s how well each team shot in each broad category of possessions during the 2014/15 regular season:

Applying this frame of analysis[10. In cursory form, I imagine there will be much more to come along these lines in the offseason as not only teams but individually players can be examined in this way, with a quick example being Corey Brewer, who had 24.5% eFG better shooting in liveball (61.5% eFG) situations as compared to against set defense (37.0%). Not exactly the ideal candidate for floor time as part of “hack-a.”] to the impending NBA Finals, it tells a the story of the teams’ disparate styles of success reasonably well. Unsurprisingly, given their season-long dominance, Golden State looks good, with the best shooting against set defense (by far) combined with the third-best in liveball situations. In fact, they are one of only three teams in the past five seasons[7. As far back as I’ve looked so far. Sue me.] to score at a higher rate against set defense than their own scrambled defense allows.[8. The other two were the 2012/13 championship Heat…and the 2012/13 Oklahoma City Thunder who might easily have been the best team in the league that year without Pat Beverley’s Olynyk-like intervention.] As a comparison think of what it means to make more free throws than the opposition attempts, as it is to my mind a similar measurement of dominance.

Meanwhile, the Cavaliers have the fifth widest split in eFG% between liveball and set situations. Getting stops really does help their offense a great deal!

In graphical form, here are the respective offenses and defenses lined up by which team is on offense:

Cle v GSW comps
Cle v GSW comps /

Based on regular season play at least,[9. Knowingly a somewhat apples-to-oranges comparison given the different composition and performance of the Cavs at present, but small heavily biased sample sizes based on games version a few discrete and non-overlapping opponents doesn’t seem like a better solution.] a bad defensive possession or two for Cleveland could easily transform into one of Golden State’s trademark scoring spurts, with transition baskets forcing Cleveland to fight against Golden State’s superb (43.1% eFG allowed) half court defense, leading to more liveball possessions for the Dubs and so on. While the feedback loop[10. Hat tip Danny Leroux] exists in the opposite direction, is not quite as strong – Golden State will be able to break up Cleveland runs more easily with their ability to score against even dug-in half-court defenders.


As a quick addendum, offensive rebounds are something of a special case. Interestingly, Cleveland wasn’t especially effective in terms of shooting directly from an offensive rebound this season, with an eFG% of only 50.4%. However, especially through the efforts of Tristan Thompson, the Cavs have made something of a living under the rim, with Thompson and Timofey Mozgov 1st and 3rd respectively in Offensive Rebound Chase % among players with at least 200 playoff minutes. To some degree, Cleveland’s efficiency on these shots doesn’t really matter, as they are truly “extra” to the normal your turn-my turn flow of a basketball game. However, they aren’t without a cost. This season saw a small but noticeable correlation between Chase% on higher allowed eFG% off of defensive rebounds:

Chase v defensive transition
2014/15 NBA Regular Season /

In other words, if Cleveland crashes the boards, but Golden State rebounds, the Warriors’ likelihood of success on their ensuing possession only rises.