A Closer Look at Close Losses
Locked on Sonics: Late-Game Misery, Research Style
Sonics Beat
Get Sonics News Delivered to You!
Sonics News Archive
Kevin Pelton, SUPERSONICS.COM | March 20, 2007
In last Saturday's 99-98 loss to the Golden State Warriors, the Seattle SuperSonics saw a seven-point lead slip away in the final four minutes and a five-point lead evaporate in the final two minutes. They saw Rashard Lewis, an 87.2% free-throw shooter this season, split two attempts at the line with 7.2 seconds left. They saw Baron Davis hit a pull-up jumper from just inside the top of the key with 2.3 seconds to play. Finally, the Sonics were unable to pass the ball in to their first two options, Ray Allen and Lewis, leaving them with a contested three at the buzzer by Damien Wilkins that found only rim.

"Itís identical to whatís happened to us the last couple of games and itís hard to put a finger on what we need to do to get better."
Terrence Vaccaro/NBAE/Getty
Just another night in a crazy season for the Sonics - their 18th loss of the year by six points or less, many of those defeats not decided until the final buzzer.

"I donít have an answer for you," Allen said after the game. "Itís identical to whatís happened to us the last couple of games and itís hard to put a finger on what we need to do to get better. Weíve had our chances. Hopefully weíve got guys who are paying attention to this and learning from it."

Saturday's loss stretched the Sonics losing streak to five games (it was snapped the next night when they won 95-77 in Portland). Four of the five games were decided by four points or less. At Philadelphia, Allen had two shots to tie or give the Sonics the lead in the final minute, while Wilkins also missed a potential tying attempt. At Toronto, the Sonics forced overtime with an Allen buzzer-beater, but could not contain T.J. Ford in the extra session. Hosting Detroit, Allen missed three times in the final 1:02 with the Pistons leading by two.

Since Opening Night, when the Sonics saw a seven-point lead slip away down the stretch against Portland, this has been a season of heartbreak.

"I can remember very vividly that one trip where we went to Detroit, Milwaukee and Memphis and should have won all three of those games," recalled Sonics Coach Bob Hill.

What has made the close losses especially frustrating is how they have held back a potential playoff push. Saturday's game was not only a loss for the Sonics but also a win for Golden State. The Warriors hold the eighth seed in the Western Conference, five games ahead of the Sonics. Change a few close losses into wins and the Sonics would be in the thick of the race.

Instead, players, coaches and observers have been left trying to figure out why the Sonics have lost so many close games. Several reasons have been suggested, most notably the team's relative youth, but most are somewhat unsatisfying because they should have applied equally to last year's Sonics squad, which went 12-9 (.571) in games decided by five points or less.

"I think in some of the cases this year, the execution of the plays late has been pretty good," Hill said. "The reading of the defenses in some cases hasn't been very good. I think a lot of them is born from inexperience. We've had games where we've executed perfect and got the shot we wanted and it just didn't go on. We've had games where the first option before the jumpshot option was open and we just didn't recognize it. We've had games where we've recognized the slip guy and thrown the ball out-of-bounds. It's been a plethora of reasons, but overall I've been pleased with the execution of the plays late."

The conventional wisdom is that good teams win the close games. Statistical analysis of the NBA and other sports has favored point differential, which barely distinguishes between a one-point win and a one-point loss. The reality probably lies somewhere in between good teams winning the close games and their outcomes being random.

(Click to enlarge)
Using data from Basketball-Reference.com, the chart at right shows winning percentage in games decided by at least six points plotted against winning percentage in games decided by five or fewer (a technical definition of a close game that is not perfect, but good enough to be useful) for all teams from 2001-02 through 2005-06 along with a line showing the relationship between the two. The correlation isn't anywhere close to perfect, but it exists (r2 of .169, to be technical).

More interesting is that the slope of the line - it's not anywhere close to one (in fact, it's a little less than .25). What does that mean? Good teams win close games more frequently than bad teams, but the difference is not as dramatic as it is for games decided by at least six points. This makes sense conceptually. Think of the difference between the one-and-done NCAA Tournament and the best-of-seven NBA postseason in terms of upsets. The better team comes up empty far more frequently in the NCAA (occasionally, like Wisconsin or Texas, ruining your bracket in the process) because the series is shorter. Anything can happen in a single-elimination tournament, and the same is true of the last five minutes of a close game.

The equation for the line on the chart can be used to predict what a team's record in close games should be based on their record in all other games. The Sonics, 18-25 (.419) in games decided by six or more, should win 47.8% of close games. Instead, they've actually won 34.8% (8-15). That's a difference of about three games over the course of the season. That's a lot, but the 13% difference between expected and actual performance in close games isn't as large as the largest of the 2001-02 - 2005-06 period:

Close Games
Non-Close Games
vs. Expected
.267 (8-22)
.635 (33-19)
New Jersey
.286 (4-10)
.662 (45-23)
L.A. Clippers
.300 (9-21)
.538 (28-24)
.261 (6-17)
.373 (22-37)
.364 (8-14)
.617 (37-23)
.414 (12-17)
.792 (42-11)
Golden State
.333 (7-14)
.492 (30-31)
.286 (6-15)
.311 (19-42)
.400 (6-9)
.716 (48-19)
.357 (5-9)
.544 (37-31)
Golden State
.278 (5-13)
.250 (16-48)

As Hill suggests, poor performance in close games is traditionally associated with young teams. While most of these teams were young, the group also includes the 2002-03 New Jersey Nets, on their way to a second straight Eastern Conference Championship, and the 2003-04 NBA Champion Detroit Pistons.

The other surprise from the chart is that the late-game performance by the Sonics is not their worst in recent memory. The 2001-02 Sonics had virtually an identical record in games decided by five points or less despite winning the rest of their games at better than a 60% clip. For the most part, the Sonics have not done well in close games over the last six years:

Close Games
Non-Close Games
vs. Expected
.348 (8-15)
.419 (18-25)
.571 (12-9)
.377 (23-38)
.600 (12-8)
.645 (40-22)
.400 (8-12)
.468 (29-33)
.375 (9-15)
.534 (31-27)
.364 (8-14)
.617 (37-23)

The track record in Seattle may be indicative of the momentum Hill speaks to.

"Normally if you win one, then you win another one and you start getting confidence," he said. "Confidence has a lot to do with it."

In 2004-05, the Sonics got some breaks early in the season, winning two close games on their first homestand. They parlayed that into strong performance in close games throughout the year and a surprising Northwest Division Championship. By contrast, this year's squad saw fate go against them early in the year, including Hedo Turkoglu's game-winning three pointer hit falling out of bounds in Orlando.

David Locke blogs his take on the Sonics losses in close games as well as sharing the late-game spreadsheet he has compiled.
Hill has experience on both sides of the outcome in close games. His Indiana teams did not finish well, going 24-37 in games decided by five points or less in his two-plus seasons with the Pacers. Hill's 1994-95 Spurs, however, won 22 close games (against 11 losses) - more than any NBA team has won in the last five seasons.

What does that experience tell him to do now?

"You don't throw your hands up," Hill said after Thursday's practice. "You just keep teaching them. It's little things."

Lo and behold, Hill spent part of Thursday working on mini-scrimmages with the intent of putting his players in the kind of situations they face in close games. The learning process continues.

vs. Expected
Year One
Year Two

There is definitely hope for the future. Throwing out the 2005-06 teams, the worst 10 teams in close games from 2001-02 through 2004-05 (as compared to expected record) ended up winning almost exactly as many close games as expected the following season. They also ended up improving by an average of more than three games.