The Sonics Play Moneyball: Part Four - The Coaching Staff
Kevin Pelton, SUPERSONICS.COM | Feb. 14, 2005
In Seattle and around the NBA, statistics haven't made the same kind of impact on the sidelines as they have in front offices. That's to be expected. Coaching ultimately comes down to feel, and, unlike general managers, coaches need only evaluate players they watch on a daily basis, not hundreds throughout the NBA and countless more amateur prospects.

This is the final installment of a four-part series detailing how the Sonics have used statistical analysis to their advantage. If you missed them, check out the other three installments:
  • Part One
  • Part Two: The Outsider
  • Part Three: The Front Office

    Also see the following associated articles:
  • Statistical Analysis Primer
  • Dean Oliver's Individual Ratings
  • Dean Oliver's Four Factors
  • Still, there are two primary ways in which the Sonics use statistics: Evaluating their own players and lineup combinations with plus-minus statistics, and scouting upcoming opponents by looking at their statistics.

    The concept of plus-minus is a long-standing one in hockey (to the point where the NHL's league leaderboard in the statistic is corporately sponsored), but its popularity in basketball is a more recent thing.

    Until last season, coaches mainly obtained plus-minus data by tracking it themselves on a coach-by-coach basis. Following in the footsteps of his coach at North Carolina and mentor, Dean Smith, former Sonics Coach George Karl always tracked plus-minus data during his time in Seattle, receiving a report on how his players rated after each game.

    "It's the No. 1 stat I believe in," Karl once told the Seattle Post-Intelligencer.

    Following the 1993-94 season, statistical guru Harvey Pollack, the Philadelphia 76ers director of statistical information, began tracking plus-minus in his annual Statistical Yearbook. The NBA's leader that first season? None other than Sonics Coach Nate McMillan (+616), then coming off the bench for the 63-19 Sonics. Without McMillan on the floor, the Sonics were just 129 points better than the opposition.

    McMillan had a good plus-minus rating as a player. Now, he's got a star, Ray Allen, who does the same.
    Noah Graham/NBAE/Getty
    "I didn't get any bonuses for that," McMillan recalls now with a smile. He did get more respect from Karl, however, as the plus-minus data verified McMillan's importance to the team, something that couldn't always be demonstrated through the use of traditional statistics.

    Pollack has annually published plus-minus data for every player for over a decade now, but that information doesn't really help coaches because Pollack can't tabulate it until the following December. Still, Pollack's was the only source available to fans - except when coaches, like Karl, made a point of mentioning plus-minuses in the media - until October 2003, when the Web site made its debut.

    For the last two seasons, founder Roland Beech has tracked plus-minus for both individuals and lineups on a daily basis. While fans have enjoyed seeing the curtain pulled back on plus-minus data, NBA coaches have benefited as well, as teams throughout the league - including the Sonics - use's data.

    Beech has taken plus-minus data a step further by adjusting for team quality. While all the league leaders since McMillan have been perennial All-Stars (David Robinson in 1994-95; Michael Jordan in 1995-96 and 1996-97; Shaquille O'Neal in 1997-98, 1999-00 and 2001-02; Tim Duncan in 1998-99 and 2000-01 and Dirk Nowitzki in 2002-03 and 2003-04), the last-place finishers have included solid players like All-Stars Shareef Abdur-Rahim and Antawn Jamison. These players happened to play the most minutes on bad teams. Beech's "Roland Rating" compares how a team does with and without a player on the court, both on a per-48 minute basis. (Click here for the Sonics Roland Ratings this season.)

    "It's not something that I substitute by, or use every day, as opposed to something nice to look at, to get a feel for a guy," says McMillan. "Sometimes there's some surprises there."

    One surprise last season was swingman Richie Frahm. After making the Sonics during training camp, Frahm consistently rated as a positive presence for the team when he was in the lineup early in the season. His final Roland Rating of +10.0 was far and away the best on the Sonics.

    "Last year, one guy that was always in the good plus-minus was Richie Frahm," says Associate Head Coach Dwane Casey. "Every time you looked up, golly, Richie Frahm, Richie Frahm. So he went and he had an immediate impact when he went into the game. That was always a different variable that you really didn't notice when he was going into the game."

    At the same time, Frahm's limitation illustrates some of the weaknesses of plus-minus data. A significant part of Frahm's good rating owed to blowout games at New Jersey and Boston Frahm entered with the Sonics trailing big. They rallied in the fourth quarter, giving Frahm a combined +32 plus-minus in those two games.

    "A funny thing can be, a guy might have a good plus-minus, but it might not be because of him," adds Sonics assistant Bob Weiss. "Like if you have a player who goes in, and let's say he starts the game and things start to go badly. The coach doesn't have much confidence in that player, so he pulls him. He may have only gotten a -6 while he was out there because you pull him right away. Whereas another player might be in there and things are going bad, you're in there for the whole slide, and you might get a -20. But, on the other hand, if that same basically non-productive player is in there and you're on a run, you'll stay with him, so all of a sudden he gets a +15 where he didn't really contribute to it.

    Frahm had a fine plus-minus rating with the Sonics.
    Ray Amati/NBAE/Getty
    "You have to look at statistics and then you have to use your common sense as well to see which are actually verifiable and which aren't."

    Where plus-minus data might be more valuable is in evaluating lineup combinations to see how well they work on the court. One lineup the Sonics have been using more lately has paired Antonio Daniels and Ray Allen in the backcourt with a big frontcourt of Vladimir Radmanovic, Nick Collison and Danny Fortson. The numbers illustrate this lineup's effectiveness; it has outscored opponents by 59 points in just 119 minutes together.

    No fancy statistics are necessary for much of the advance scouting of future opponents shared by Casey and Weiss. They're looking for key indicators of player skills and how those skills will affect the Sonics strategy.

    "The biggest stat would probably be three-point shooting, outside shooting, where their points come from, how do they score?" says Weiss. "With each team it's going to be a little different, but three-point shooting is probably one of the predominant ones.

    "You have to know who you can rotate off of and who you have to stay with."

    Weiss has been known to enlist Sonics interns to conduct his own statistical studies. One of the most successful came during the 2002-03 season, when he discovered that the Philadelphia 76ers were 11-1 that season when Allen Iverson attempted at least 10 free throws, 1-4 when he failed to hit that mark.

    "We used stats in that game to show that if we keep him off the free-throw line, make him shoot the three-point shot, (we would be successful)," said Weiss. "All of a sudden, you can turn those statistics into positives for yourself."

    Free-throw percentages are another important marker, as they let coaches know which players should be fouled when they get a good look down low and which players they should look to foul when trailing down the stretch. Coaches also look at fouls and blocks on the defensive end of the court.

    As they scout, the coaching staff can make use of advanced statistics provided by Sonics consultant Dean Oliver, featured in an earlier part of the series. Oliver sends the coaching staff advanced statistics about upcoming opponents and shot charts prepared with Beech's help that answer Weiss' questions about where opposing players are scoring from.

    "I felt comfortable with his information and the way he brought it in," says Casey, referring to when Oliver first began communicating with the coaching staff during training camp. "His information just gives you a base. We're not making decisions based strictly on Dean's stuff. It's great stuff, but as a base to go by, not to make hard decisions on."

    Another piece of proprietary information the Sonics use is what combinations opponents like to use. The coaching staff charts games just before they play an opponent to see the rotation pattern the opposing team is using.

    "We don't do their plus-minus per se, but we keep a clock of the guys on the floor," says Casey. "We know who's in at the end of the first quarter, end of the second, end of the third and in crunch time at the end of the fourth."