2018-19 Duke: It's How You Finish
The Blue Devils had a knack for playing incredibly well in the final minutes of games. What was their secret?
This is the second in my 4 part series about the 2018-19 Duke Blue Devils. Part 3 will be released on Monday, Nov 23. Visit my archive for past articles on other teams.
The 2018-19 Duke Blue Devils sure knew how to finish a game strong. They were no slouch in the opening 35 minutes of game, posting an adjusted margin of +34 points per 100 possessions (excluding garbage time). But when they got to the final 5 minutes, they took it to another level. They put up an adjusted margin of +47 per 100 possessions, which is even more impresive when you consider that most of their games against weak opponents are excluded since I’m filtering out garbage time.
Duke was outscored in these final stretches only 4 times all season, and 2 of those were games they still won. For comparison, they were outscored over the first 35 minutes in 5 games, and lost 4 of them. In Duke’s 6 losses, they were -36 in the opening 35 minutes but only -1 in the final 5 minutes. The only loss where they were outscored by more than 1 point in the final 5 minutes was a -5 against Syracuse, and that took overtime and the absence of both Cam Reddish and Tre Jones.
So, what was Duke’s secret to late game success?
All it takes is one thing you’re really good at
So, what was driving this improvement late in games? Did Duke’s offense improve? Let’s look at the key stats there…
In the last 5 minutes, Duke had a 53% eFG%; it was 54% in the opening 35
They turned the ball over on 17% of late possessions vs 18% for the rest of game
Duke collected 38% of their own misses in the final 5 minutes, and 35% during the opening 35
All in all, the Blue Devils scored 1.13 points per possession in the final 5 minutes and 1.15 during the opening 35
OK, so something on defense drove the improvement. Maybe it was forcing turnovers?
Duke forced turnovers on 14% of opponent possessions in the last 5 minutes, compared to 20% during the opening 35
That wasn’t it. Defensive rebounding?
Duke snagged 72% of opponent misses in the final 5 minutes, compared to 71% in the opening 35
OK, maybe opponents shot a lot worse?
Duke’s opponents had a 38% eFG% in the closing 5 minutes compared to 47% earlier
Duke allowed 0.86 points per possession during the final 5 minutes, vs 0.93 earlier
THERE IT IS! The key was that opponents didn’t hit any shots.
Suffocating half court defense
Sometimes opponents shoot worse because they get fewer fast breaks, but that didn’t happen here; opponents took about 30% of their shots in transition early and late in games. There is some data that indicates that Duke turned up the pressure in halfcourt defense, however.
If we just look at possessions longer than 10 seconds (to filter out fast breaks), opponents had an effective FG% of 47% in the first 35 minutes and 37% afterwards. Duke forced opponents to use at least 20 seconds in the shot clock on 24% of their possessions in the first 35 minutes, but this jumped to 31% later in the game as Duke turned up the pressure. So, the Blue Devils didn’t force as many turnovers, but they kept opponents from getting quality shots. Duke also forced more teams into isolation-scoring late in games. 54% of opponent’s makes were assisted during the opening 35 minutes, but this plummeted to only 41% when the game got close to the end.
The effect of this shows up in shot charts for Duke’s opponents. Here’s what opponent’s shot charts looked like during the opening 35 minutes, on halfcourt opportunities (>10 seconds into shot clock):
And here’s the same chart for the final 5 minutes of games:
Opponents turned hopeless offensively in halfcourt situations late in games. Instead of hitting a solid 60% of their shots at the rim, they hit a poor 48%. Instead of hitting a passable 32% of threes, opponents hit a hopeless 18% late. Opponents always struggled to hit 2 point jumpers, and got even worse late in games (29%!).
A good match against late shot clock play types
A big part of why Duke was so much more effective late in the shot clock relates to the type of plays opponents run late in the shot clock. According to Synergy, Duke’s opponents had 167 possessions ending int he final 4 seconds of the shot clock. Only 7 of those were post ups. Late shot clock possessions tend to be isolation, pick and roll, or passes to spot up shooters.
Per Synergy data, these types of late shot clock plays are exactly what Duke was good at defending. Duke was graded in the 87th percentile defending Spot Ups, 92nd on P&R Ball Handlers, and 93rd on Isolation, but was only in the 62nd percentile defending Post Ups.
Much of Duke’s success in these areas was due to their perimeter defenders, who combined quickness and length. Zion, Cam Reddish, and Tre Jones all were in the top 180 nationally in steal rate, and those three along with RJ Barrett committed fouls at very low rates. The only game Duke lost when they had a lead at the 5 minute mark was January 14th vs Syracuse; Cam Reddish did not play this game, and Tre Jones left with a shoulder injury early. Instead, Duke was forced to play Alex O’Connell and Jack White for much of the final 5 minutes and overtime. Syracuse put them in isolation or pick and rolls frequently, and had more success than any other team scoring in the halfcourt late against Duke.
Duke was an excellent team in the 2018-19 season, but turned into an absolute juggernaut down the stretch of games. When Duke had their star players available, their halfcourt defense was like a vise for opponents. With a quartet of smart, athletic players who almost never fouled out (only 2 among their top 4 players all season!) Duke used their defense to finish off many of their wins.
In 4 of Duke’s 6 losses, they were missing at least 1 of their 4 star freshmen. The most notable stretch was the 6 game stretch without Zion Williamson after his shoe practically disintegrated. In my next article, I’ll look at how Duke played without Zion, and how Zion and the team adjusted upon his return.