The objective of this post is to use available information for a preliminary statistical assessment of the Timberwolves’ prospects for the upcoming 2017–18 season.
All data for this post come from basketball-reference.com. First, I collected
data on NBA averages for starters last year, by team and by the league average
for all starters, as well as data on the probable 2017–18 Timberwolves starters:
Ricky Rubio Jeff Teague, Andrew Wiggins, Jimmy Butler, Taj Gibson, and
Karl-Anthony Towns. Next I identified the starters for each team. I limited the
sample to “starters” not only to narrow this to a reasonable scope for a blog
post, but also because (1) Tom Thibodeau plays his starters so many minutes and
(2) starters are generally the best and most effectual players on a team’s
roster. Both of these reasons shine a light on how starters play an outsize role
in generating a team’s wins (and losses).
This is admittedly an imperfect sampling method. In real life, teams sometimes use many
different starting lineups throughout a season or have a 6th man who’s more
important to the team’s success by coming off the bench than the guy who’s
starting for him. Some teams simply rely more on their bench than others do. Tom
Thibodeau-coached teams tend not to be in this category. Thibs relies
heavily–perhaps to a fault–on his starters to log big minutes each night. So,
for brevity’s sake, I limit the analysis to the five players who started the
most games in 2016-17 for each team. (Editor’s Note: Sampling could be done more
flexibly–looking, for example, at each team’s “core rotation” defined by some
arbitrary number of minutes played per game or in other ways. This might be a
Having defined our sample, I calculated basic statistics for each team’s
starting unit for each season since 1980, but ultimately focused on the 2016-17
season for the immediacy of the comparisons. I calculated each team’s starters’
average basic stats and advanced stats as the average of each stat for each
team’s starting five. My aim is to develop a sense of how much better the Wolves
might be this year after adding three new starters, so the analysis in the rest
of the post focuses on a few advanced statistics that strongly correlate with
winning: Player Efficiency Rating (PER); Win Shares, Win Shares per 48
minutes; and Value Over Replacement Player (VORP).
If you aren’t familiar with these stats, click through the links for
definitions and background information.
To explore the data, I put together the stats of the Timberwolves’ probable
starting lineup for the upcoming 2017–18 season based on their 2016-17 stats. I
then analyzed how their numbers—for the sake of simplicity, I assume their
numbers will be the same this season–to try and get a sense of how they
would’ve stacked up against other teams last season. The analysis is not a
forecast of the Wolves’ 2017-18 W-L record as much as it is an initial attempt
to develop some basic quantitative expectations about the team’s potential and a
baseline for interpreting the Wolves new-look starters’ advanced statistics
relative to an “average” NBA starting unit.
The first comparison I checked out was that of last season’s Wolves starters
(Rubio, Zach LaVine, Wiggins, Gorgui Dieng, and Towns) and this season’s likely
starters. Again, for the upcoming season’s crew, I used each player’s individual
statistics from 2016–17. For the Wolves, Wiggins and KAT are the two “constants”
in the equation. The three new starters provide the variation between last
season’s stats and the probable lineup we’ll see in 2017-18. Jeff Teague
replaces Ricky Rubio, Jimmy Butler replaces Zach LaVine, and Taj Gibson replaces
Gorgui Dieng. (Editor’s Note: It is possible that Gibson will come off the bench
and Dieng will start, but for the sake of looking at the Wolves’ key offseason
personnel moves, I decided to assume Taj will be the starting power forward.)
The bar graph shows the averages of the 2016–17 Wolves starting five, the
2017–18 Composite Wolves, and the 2016-17 league average for NBA starters.
The bottom line up front: the 2017-18 Wolves should be much better not only than
last season’s team, they are likely to be an above-average NBA team compared to
teams last season. I’ll refer to the 2017-18 Wolves as The Composite
Timberwolves (TCT) because the statistical estimates are based on each player’s
numbers last season and assume that their similar statistical output will be
similar this season.
Let’s first compare the numbers of last season’s Wolves starters and this
season’s Composite Wolves starters. The Composite Timberwolves’ starting unit
had an average PER of 20.34 (up from last year’s starters’ average of 18); it
had an average of 8.6 win shares, compared to last season’s 6.3; their average
ws/48 is 0.15 (up from 0.11 last season); and, finally, The Composite
Timberwolves starting lineup had an average VORP of 2.7, which would be a
substantial increase from the middling 1.9 VORP of last year’s group.
A second question to explore is how The Composite Timberwolves’ advanced stats
compare with the overall NBA average for all teams’ starting units in 2016-17.
The Wolves haven’t had a .500 or better season since 2004-05. Statistics that
project to such a season would satisfy many Wolves fans who simply want to see
NBA-quality basketball every night. The chart tells two stories, really. One is
that the Wolves’ starters last season were alright. They were close to the
league average along each metric. The other is that The Composite Wolves’ stands
a good chance of being significantly above average if they perform at the same
level in 2017-18 as they did last year. With all due respect to Jeff Teague and
Taj Gibson, this really shows what a difference-maker Jimmy Butler could be.
Zach LaVine, whom Butler replaced, scored and shot well last season, but this
didn’t translate to high-quality team basketball, as evidenced both by the stats
and the eye test.
Another way to look at the comparison is in terms of percentage changes from
last year to The Composite Timberwolves probable starting five heading into this
season. The percentage-change improvements in 2017-18 on our key metrics could
be substantial: the projected PER for The Composite Wolves starters is 11
percent increase. The projected mean increase in win shares is 37 percent, while
the projected increase in ws/48 is 38 percent. The projected improvement in VORP
from last year’s group to this one’s is 42 percent. Even if The Composite Wolves
do not perform at quite the level they did in 2016-17, this year’s core group
shows every indication that it will be much, much better than last year’s.
(Editor’s Note: One methodological issue to keep in mind here is that the
comparisons implicitly assume that KAT and Wig will perform identically to their
2016–17 selves. Both players are 22-years-old or younger, however, and are
expected to continue to improve. The Composite Timberwolves’ projected stats
might thus have a downward bias. That said, it is also possible that some
permutation of Butler, Teague, and Gibson–all veteran players–won’t put up
numbers as good as last year’s, which would imply an upward bias in The
Composite Timberwolves’ projected numbers. While either or both scenarios are
possible, we can use last season’s stats as a starting point to think broadly
about how good this team’s core will be. Relatively speaking, Tom Thibodeau has
taken a long view toward rebuilding the roster. Yet the acquisition of Jimmy
Butler in particular (not to mention the anticipated improvement of Karl Towns
and Drew Wiggins) should make the Wolves a dangerous team now, in the
Thibs-Layden regime’s second season in ’Sota.)
Evidence shows that the Wolves are likely to be improved, perhaps dramatically,
in the upcoming season.
Let’s look next at how The Composite Wolves ranked statistically compared to the
starting units of other NBA teams last in 2016-17.
PER and VORP
Below is a plot of average PER and VORP (value over replacement player) for NBA starting lineups last season, plus the average the Wolves
probably 2017–18 starters had. The Composite Timberwolves checks out pretty well
with an average PER score of 20.34: the only team with a higher player
efficiency rating among its starting lineup was the title-winning Golden State
Warriors. Last year’s Wolves’ starters–again, that’s Rubio, LaVine, Wiggins,
Dieng, and Towns for the purposes of this post–had an above-league average PER
at 17.6, placing them in the top–10 among starting units.
Most of the bump The Composite Timberwolves get in the PER statistic comes from
swapping Jimmy Butler for Zach LaVine: Butler’s PER last season was 25; La
Vine’s was 14.6. The 2017–18 Wolves get a smaller but significant statistical
bump up from replacing Ricky Rubio (16.8 PER last season) with Jeff Teague (19
PER). The difference between Taj Gibson and Gorgui Dieng–both of whom are
expected to see plenty of time at PF this season–is basically negligible.
Looking at the other metric, VORP, we see that The Composite Timberwolves don’t
fare quite as well as they do using PER, but they would still have ranked 6th in
the league, behind teams with the league’s most stellar starting lineups–GSW,
HOU, LAC, CLE, and OKC.
WS and WS/48
The next charts show where The Composite Timberwolves would have ranked in
2016–17 based on the stats put up by their returning incumbent starters (Towns,
Wiggins) and their probable new starters’ (Butler, Teague, Gibson) stats from
Again, the 2017–18 Composite Timberwolves’ average number of win-shares ranks
second in the NBA behind only Golden State. (Editor’s Note: This is your daily
reminder that the Warriors are incredible.)
While we’ll have to wait to see how the Wolves’ lineup will gel, this is another
encouraging empirical indicator suggesting that the Wolves have a core that is
elite right now. Indeed, The Composite 2017–18 Timberwolves are ahead of teams
like the Clippers, the Rockets, the Cavs, and the Spurs, all of which won more
than 50 games in 2016–17. The league average for teams’ starting units in
2016–17 was 5.3 and the Wolves’ group of Ricky/LaVine/Wiggins/Dieng/Towns
average was 6.3. By comparison, the average number of win shares for The
Composite Timberwolves’ lineup of Teague/Wiggins/Butler/Gibson/Towns starting
lineup was 8.5, which again indicates the high potential ceiling for the 2017–18
To be sure, win shares appears to be a credible predictor of actual wins: the
sum of the last season’s 31–51Wolves starting unit’s win shares was 31.7. (Editor’s
Note: The sum for the entire team was 39, which still is a reasonable prediction
because the stat is unable for the number of wins the team didn’t get due to
second-half collapses in which they ceded a large lead and ultimately ended up
Extrapolating from the win shares statistic is interesting. The Composite
Timberwolves starting five’s win shares in 2016–17 summed up to a predicted 42.8
wins. And if you include the win shares of the rest of the rotation–Dieng, Jamal
Crawford, Shabazz Muhammad, and Tyus Jones–the estimated number of wins based on
win shares sums up to 56.1. (Editor’s Note: Vegas has the over-under for Timberwolves wins at 45.) Last year, 56 wins would’ve been good for third
place in the Western Conference, behind only Golden State (67 wins) and San
Antonio (61 wins), and first place in the East, which Boston won with 53 wins.
The upcoming season’s Composite Wolves starting unit does not grade out quite as
well in the chart in the right-hand side panel–WS/48–as it does in win shares.
Still, the team’s improvement on this metric would move it up from several spots
below the NBA average to significantly higher than the NBA average for starting
units. If the numbers hold, they would have the Wolves’ core performing better
on WS/48 than some teams that were quite good overall in 2016–17, including
Cleveland and Boston.
From Whence the (Estimated) Win Shares Came
Finally, I looked at separate statistics for win shares: offensive win shares
(OWS) and defensive win shares (DWS). If this season’s Wolves team takes a
significant step toward becoming a playoff team amid the Draconian competition
of the Western Conference, should we expect the increase in wins to come from
better offense or defense? (Editor’s Note: The obvious–and correct–answer is
“both,” but we want to dig a little deeper here to understand the estimated win
shares and to compare changes in offensive win-shares and defensive win-shares
between last season’s starters and this season’s.)
Last season, the Wolves were able to put points on the board, but were among the
worst teams in the league defensively. Should the team improve as it is expected
to, will that improvement be more likely to come from improved defense? Or is it
more likely to come from the additional offensive firepower acquired this
The figure below provides a visual projection and comparison.
The Composite Timberwolves’ core group ranks first in the league in OWS, an
estimate and ranking that probably won’t be sustainable as the Wolves settle
into their offensive roles this season. Still, to have an offense that is even
in the conversation with outstanding teams like last season’s Finals
contestants, Golden State and Cleveland, should make for some (reasonably)
heightened expectations for the Wolves heading into the season.
Moving to DWS, well, it is no secret that the Wolves haven’t had a good
defensive team in years, despite hiring the renowned defensive coaching guru Tom
Thibodeau prior to last season. Also not a secret is the fact that Zach LaVine
was really bad at defense, while his replacement in the starting lineup, Jimmy
Butler, is an outstanding defender, as is free-agent acquisition Taj Gibson.
(Editor’s Note: Ricky Rubio was strong defender for the Wolves, but seemed to be less effective in Thibodeau’s schemes.)
The 2017–18 Composite Wolves starters’ numbers indicate that they might be a few
spots better than the league average defensively as proxied by DWS. Last
season’s starters, by contrast, were well below average, barely outranking the
sluggish defensive units of the Brooklyn Nets, the New York Knicks, and the
Sacramento Kings. Given Tom Thibodeau’s reputation as a premiere defensive
coach, the additions of the stellar defenders Jimmy Butler and Taj Gibson to the
starting unit, and the additional year of experience KAT and Wiggins will have,
it will be interesting to see how much this year’s team will improve for real on
the defensive side of the ball.
Pretty much every NBA writer expects the 2017–18 Wolves team to be better than
the 2016–17 version was. But here in Minnesota, many folks tend to fall into a
category of fan often referred to on message boards as “pessimists.” With the
2017–18 just around the corner, it feels as though most of these fans are taking
a “wait and see” approach to the Timberwolves, especially since so many were
burned after having unrealistically optimistic expectations going into last season.
While no set of statistics is going to tell us exactly how the 2017–18 team will
come together, or what role each player will be assigned, or if Thibs will rely
more on the bench, or anything else, we can say something about the caliber of
players Front Office Thibs have assembled for this year’s team. Barring
unfortunate injuries, the caliber is high. The pressure is now on Coach Thibs to
get the group to play as well–or better–than the sum of its parts.