Daryl Morey is an interesting subject to me. Much like Warren Buffet, he is generally viewed as a leader in his field. Also, much like Warren Buffet (and this is strictly my opinion, based on anecdotal evidence), he often tells the public one thing while doing another. This may seem dishonest, but when enough of the market is paying attention to your words and actions, it’s impossible to gain an advantage without withholding information. When you are constantly hounded by the media and peers for advice and thoughts, your options are either to 1) make no comment, 2) reveal your best secrets and lose any competitive advantage, or 3) tell a story that sounds plausible, but is not the full truth.
The topic of interest today is the seeming disconnect between his strong pursuit of a “star” and his disbelief in the value of “clutch players.” According to 82games.com, clutch is defined as “4th quarter or overtime, less than 5 minutes left, neither team ahead by more than 5 points.” In the 2009-2010 regular season, 172 games, or 13.98% of all games played, were decided by 3 points or less. That’s a pretty high number of games that come down to clutch situations. Last season the Rockets were 5-5 in games decided by 3 points or less, and 20-20 in games decided by 4 to 9 points. For a GM purportedly not interested in clutch players or statistics, it has to be a concern that the Rockets are already 2-4 this year in games decided by 3 points or less, and 4-5 in games decided by 4 to 9 points.
Circling back, we know that Morey has potentially made conflicting assertions. If a star’s usage increases during clutch time, it seems that the need for a star is concurrently exacerbated during clutch time. With such marginal clutch performances last year, followed by a downward trend thus far this year, I am starting to believe that Morey may have misled us in his apparent disregard for clutch-time statistics. After all, actions speak louder than words, and his intentions this past off-season were in clear pursuit of a big-time star and legitimate go-to player in the clutch. So, before we blindly go with the assumption that Morey actually does look at clutch statistics, let’s take a look at some of the evidence supporting this view.
I wanted to see if there is any difference in wins for a team leaning heavily on one player in the clutch versus a conglomerate of players. In this mini-study, I chose one player from each team in the ’09-’10 season, courtesy of 82games. The player chosen had the highest combination of games played, minutes, and FGAs per 48 minutes of their respective teams. I’ll call each of these players their teams’ “clutch players”. Last year it was Aaron Brooks for the Rockets. Here is the complete list:
The “star” versus “committee” difference takes care of itself when you take a single player from each team, because if a player is sharing possessions with someone else in the clutch, his per-48 minute numbers will be lower than a player taking the lion’s share of possessions (ie. LeBron). In other words, players with lower per-48 minute numbers are on teams with no defined go-to player or “star.”
As usual, I compared a number of different metrics with wins, and looked to see which were the most correlated.* Here are some of the more interesting results**:
1) Points scored are surprisingly the most correlated statistic, more so than any other metrics taking assists into account.
2) Having a single player use more FGAs in the clutch, and even more accurately, more possessions, is more important to winning than that player’s scoring efficiency.
3) Assist numbers are not very correlated to wins (r2 of .038). When your clutch player passes more as opposed to shooting, the correlation to winning decreases. That would help explain why LeBron got so much flack several years ago when he passed to Donyell Marshall in the corner for a missed 3-pointer.
This evidence points towards the need for a go-to scorer in the clutch. The type of player that, regardless of the results, you know decided your team’s fate in the waning minutes of the game. Kevin Martin may be developing into that kind of player, but what about when Brooks is healthy again? With Scola using possessions in the post as well, it seems like we are truly a “clutch by committee” team, which does not bode well for winning. To the right is the list of the top clutch scorers per-48 minutes last year, and their teams’ wins. Looks like good company to me.
For a more detailed, albeit ancient, examination of clutch stats, see this article.
*A couple of disclaimers here: Total team wins are used, while these stats are only from clutch games, which means that there are many wins that were not attributable to the clutch statistics (number of clutch games played for each player ranged from 20 to 47, with the average being 35.3). Also, these stats are not adjusted for pace, so a faster-paced team will have slightly inflated numbers.
** “Team points”, “created points”, and “team created points” are metrics I invented on the fly in an attempt to break down a player’s impact on the floor. “Created points” removes points that a player scored from an assist by another player. It is essentially points scored in isolation scenarios. “Team Created Points” is created points plus an estimate of points created from a player’s assists. It can be viewed as all team points that a player created (but didn’t necessarily finish).