Stat Stats: Do the Cavs lack a Killer Instinct?

Stat, from the latin statim for “immediately”,

Stats, as in statistics.

The impetus for the first ever “Stat Stats” post is a comment from Raoul in today’s Recap.

Do you have any stats showing the Cavs have a bigger “maddening ability to take the foot off the gas and let their opponents back into games” than any other team?

News flash: The other team is also trying to win! Have you heard the expression “Everyone makes a run in the NBA”? As long as anyone has played basketball, most games are a series of back and forth runs. When the other team makes a run, that does always mean someone took their foot off the gas. That is the way basketball goes.

Certainly, anyone who has ever played basketball can appreciate that:

1.) As one team increases its lead in a game, its sense of urgency naturally decreases as a function of the lead.

2.) As one team gets further in the hole, its sense of urgency naturally increases as a function of a lead.

3.) As one team starts to get blown out, its incentives may switch from trying to win to trying not to get embarrassed.

4.) As one team starts to, ahem, blow out another team, its incentives may switch from trying to continue blowing out versus trying to rest its stars or avoid injuries.

Any way you slice this, there are natural forces that tend to push NBA game scores towards somewhat even affairs.  Another way to think about it is, in games that regularly feature combined scoring around 200 points, a 10 point differential in the final score is considered a blowout.  Or, playing roughly 10% better than your opponent over 48 minutes is considered dominance of sorts.  I can best describe this phenomenon by a running joke I have with a buddy who doesn’t follow basketball much.  Whenever I invite him over to watch he lets me know he’ll come by for the last five minutes when the game is sure to be tied.

I do get the sense this season that the Cavs have trouble slamming the door on opponents, more so than they should at least.  So I decided to investigate a little bit.

I found two imperfect ways to look at this.  The first is easy to digest.  I subtracted each team’s Point Differential (or, average score at the end of the game) with that team’s Average Largest Lead.  This is probably a more meaningful comparison between good teams that win most of their games.  As for terrible teams, the worst they can do is have a largest lead of 0 (right at tipoff).  It can’t get any worse and most big leads by bad teams happen very early and are partially random.

The second way to look at this is to arbitrarily define a “team is comfortably ahead” threshold and compare the plus/minus of each team after it has exceeded that threshold.  Or, how much quote unquote killer instinct does your team have?  When you are up 10, do you hang on, build a lead, bleed it away, or quickly succumb to boredom?  NBA.com provides in-game splits based on point differential.  So I chose being ahead at least six points (a 2 possession game), and compiled the plus/minus for each of the top teams in the league when that condition is met.

Be….sure….to…drink…your….RAPM?

First Approach Results: The Cavs have an average largest lead of 15.5 points.  Their point differential is 6.0.  That’s a drop of 9.5, or the 6th smallest drop behind: LAC, TOR, GSW, OKC, SAS.  That smells right, considering the Cavs are 4th in SRS.  Nothing too interesting here.

Second Approach Results:

SAS:  +3.5 PPG
GSW:-0.4 PPG
OKC: -5.3 PPG
CLE: -8.9 PPG
BOS: -10.1 PPG
TOR: -14.1 PPG
LAC: -15.3 PPG

This is a little more interesting.  The Cavs get outscored at a rate of -8.9 points per game in situations when they are ahead by at least six points.  I compared them to the other top six teams in the NBA (according to SRS).  As you can see above, the Thunder bleed points away at a rate of -5.3 per game, while the Warriors basically play to a draw (-0.4).  The Spurs, meanwhile, outscore opponents by 3.5 points per game anytime they are already ahead by at least 6 points.  Clearly, they are the ultimate Killer Instinct team.

Looking at the results, it’s a little bit hard to grasp what exactly is happening.  How can the Warriors, with a point differential of +12.5 have a slightly negative point differential when ahead by six points???   Well, it has to do with the way you are classifying the events.  I’ve provided an example:

Imagine a team up 5-0.  Then, that team hits a 3 to go up 8-0.  (Only now is the greater than or equal too +6 condition valid, so put a mental marker at +8.) After a wild sequence of runs in both directions, the opposing team claws their way back to a six point differential, at say, 33-27.  At this point, the winning team has been ahead by at least 6 points for almost the entire game, and yet, their plus/minus would register at MINUS TWO because they didn’t begin to be “ahead by at least 6” until after the 3 pointer to put them at +8.  (6-8 = -2) Now, the opposing team hits ANOTHER TRIPLE to bring the game within one score at 33-30.  This shot was attempted while the winning team was ahead by at least 6, so it counts in the plus/minus aggregation.  Therefore, the winning team has a plus/minus of MINUS FIVE (3-8 = -5) during all moments of the game when they were ahead at least six points.  See how this is kind of weird?  It is more intuitive to think: “If a team creates enough of a lead to get above the +6 threshold and eventually loses it to get out of the +6 threshold then its plus/minus throughout those sequences evens out to zero.  But that’s not the way it works as you can see in the example.

The crowd is getting restless.  The winning team’s superstar steals the ball and throws down a dunk, making the score 35-30.  Then Delly (he’s here now) steals the inbounds pass and drains a no-hesitation 3 to put the winning team up 38-30.  One more stop + another bucket will put the game out of reach.  Delly fronts the 10’7” MonSTAR, forcing the turnover.  He finds J.R. Smith who takes a triple pump, stanky-leg, falling out of bounds 3-pointer.  BANG!  The good guys go up 41-30 and the game ends because Chuck Norris rushes the court and roundhouse kicks all the MonSTARS back from whence they came.  Now, the winning team didn’t go back UP by more than six until AFTER Delly’s no-hesitation 3.  So only the stanky-leg triple counts in the plus/minus when the winning team was up at least 6.  So, even though the final score, 41-30, contributed to a +11 for the winning team, that team actually posted a MINUS TWO when leading by at least 6 points.  Hopefully that sheds some light on the values.  Here’s a graph to demonstrate the story I just made up.

Neither of these methods are great.  Ideally, you would want to isolate garbage time, which is going to factor in greatly here.  I don’t have a way to do that (yet).  But I suppose Raoul has an appropriate perspective that all teams give up big leads.  That said, the numbers here are congruent with the theory that the Cavs lack a killer instinct.  But those numbers don’t prove that the Cavs are any better or worse than similarly ranked teams when it comes to SRS.  (Raoul’s point, I think.)  That said, the lack of the killer instinct would contribute to a lower SRS….so…I’m gonna go out on a limb and say that, yeah, I’ve found nothing to change my belief that the Cavs do in fact lack a Killer Instinct (although we haven’t proven anything).

When this phenomenon happens (the whole, taking the foot off the gas pedal), and against whom is probably more valuable information than what I’ve presented here.  But, despite the theme of the title, this took me forever so act like it’s interesting please.

Editor’s Update: I replaced the old graph with the proper one.  Also, here is a list of the percentage of time a team is ahead by at least 6 points:

SAS:  50%
GSW:56%
OKC: 38%
CLE: 41%
BOS: 37%
TOR: 32%
LAC: 35%