The FIBA World Cup starts this weekend and Team USA’s first opponent is Finland, and with many international stars skipping the competition — from Tony Parker and Manu Ginobili to Lithuania’s starting point guard Mantas Kalnietis, injured, along with Ilyasova and Turkogulu from the silver medal Turkey squad — it appears to be reasonable to write off most of the field, but it’s that kind of thinking that led to the embarrassments for Team USA in 2002 and 2004. And given the injuries and drop-outs for the U.S., there are fears this team isn’t powerful to make it through unscathed and there’s a general curiosity about how good they are. With a few advanced stats and a keen eye toward history, this is possible.
Minutes and metrics
The first step is partitioning out player minutes. The easier thing to do would be to simply use the minutes played in the exhibition games, but this has some glaring systematic and random flaws: the best players are used less because the games are meaningless and sometimes players skip a game for an unrelated matter, like DeMarcus Cousins with a minor injury. I fixed these issues and tailored the high-end guys toward the minutes ceiling found in recent Team USA rosters: 25.8 from Chris Paul in 2012, 28.2 from Kevin Durant in 2010 (the best parallel to this team), and 24.8 from LeBron James in 2008. Thus, the maximum a guy plays, James Harden, is roughly the average of those three values. I did this by slightly changing minutes, giving guys who played a lot more minutes and the bench warmers less. (By using a factor of 1/5*(MPG-40/12*5), Harden’s minutes increase to the benchmark of 26.7 and the other changes appeared reasonable.) I also made a manual adjustment for Derrick Rose, docking him a couple minutes in case injuries spring up again and giving those minutes to Stephen Curry.
| Player | MPG |
|---|---|
| Derrick Rose | 18.6 |
| Stephen Curry | 22.9 |
| James Harden | 26.7 |
| Kenneth Faried | 16.3 |
| Anthony Davis | 22.9 |
| Kyrie Irving | 21.1 |
| Klay Thompson | 24.6 |
| DeMarcus Cousins | 14.4 |
| Mason Plumlee | 6.8 |
| Rudy Gay | 15.6 |
| Andre Drummond | 3.5 |
| DeMar DeRozan | 6.6 |
Why did I detail my steps for something as simple as minutes? Because one influential sports site, FiveThirtyEight, erroneously only used exhibition minutes. The error here is smaller than it would be with an NBA system since the starters typically play light minutes in international competitions, but it is still an oversight.
Then with ESPN’s Real plus/minus, which is a blend of box score stats and an advanced plus/minus model, I can make an approximation of their strength. To confirm its veracity, you can check out the last article using this method to see how it rated past teams. (Note: I average the two seasons adjacent to the summer, so I had to use an aging curve to estimate the 2015 values.)
Translating NBA success to FIBA
Now here’s the tricky part: what good is an NBA metric in a totally different league?
I need a translation for the FIBA tournament itself. You can compare historically the team’s real plus/minus to their tournament strength rating, but there are a couple issues here people largely ignore: the overall strength of international competition or the baseline, and how many teams are playing. If there are a large number of teams, then there are more weak teams and thus a team’s rating will get inflated. To get around this problem, I looked at the average team rating (it’s simply adjusted point differential) of the top five teams, excluding the U.S. every competition, and compared that to the number of teams in the tournament. I came up with a simple adjustment: add 4.2 to the average RPM for the typical 12 team tournament in the Olympics and then add an additional 6.7 for a 24-team tournament (like this year.)
But I have to point out this is a messy process with a tiny sample size, limited information, and the precision is troubling. Yet this is the best I could do — and without further ado, here’s the first set of estimates.
Average Real plus/minus: 8.6
Estimated strength rating (point differential): 19.5
To put those numbers into perspective, real plus/minus sees this team as a +8.6 one in the NBA, which is a little better than the San Antonio Spurs (+8 adjusted point differential). A strength rating of 19.5 in the World Cup is, believe it or not, a worse rating than the U.S. had in 2002 and 2006 when they placed sixth and third, respectively. Also, three other teams have had a better rating since the turn of the millennium: Turkey in 2010, Spain in 2006, and Yugoslavia in 2002, and it was within striking distance of many other teams including Argentina twice. A rating like this is, simply put, mortal.
Age: the undiscovered country
Yet the analysis doesn’t stop here. There’s something fishy about these ratings, and you can see this by looking closely at the 2010 team. The average 2014 RPM score is 8.2 and with the aging curve it’s 8.9 in 2015. Why is there such a small change? In 2010, the average for the Team USA players (weighted by minutes in the tournament) was 12.2, and in 2012 this ballooned to 15.6. Summer competition has long been a precursor to a break-out performance, and that’s what happened last time the US sent a group with few stars and underwhelming young players.
Aging curves are usually built to handle an entire league, not super athletic and skilled young players bursting with talent. Stars naturally improve faster than the average NBA player. If we use an optimistic projection based on the 2010 team, when the aging curve is boosted by a factor of 4.15, we get a rosier estimate.
Average Real plus/minus: 9.6
Estimated strength rating (point differential): 20.5
But even with a strong aging curve for the youngest U.S. team ever since professionals were used, the results are not impressive. It projects as a team that would have problems with a strong international squad, which appears to be likely with Spain and its homecourt advantage. The underlying problem here is plus/minus: it does not like a good portion of the roster.
– Harden’s abysmal defense is picked up here, while box score metrics cannot identify just how bad he is. Thus, he falls out of the superstar range with a heavy penalty on defense, but with a smaller burden and a commitment on “locking guys down” he may be more effective for the US.
– Kenneth Faried is the prototypical energy guy who teleports from end part of the court to another, but largely due to defense and a lack of playmaking plus/minus metrics are not impressed. He has, however, carved out a great role on the team and earned a starting spot.
– Anthony Davis is on track to be a superstar, but his eye-popping defensive stats have yet to translate to the team level. We’ve seen this before with Kevin Durant, who was a net negative overall his first two years and then skyrocketed to the top by his third season.
– Kyrie Irving has vexed plus/minus proponents because via the “eye test” he’s one of the most skilled players in the league and a high scoring guard with good efficiency shouldn’t be a negative. But RPM has him at -1.4. And remember, this is a blended metric that includes box score stats. Neil Paine’s numbers (box score only) have him at +2.8. This divergence is troubling, compounded by his sterling play in exhibition where he led the team in assists and, oddly enough, raw plus/minus. This is where I think RPM will be proven wrong, and for the US’s sake I hope so.
– Cousins just finished a 20/10 season and shows a lot of skill with ballhandling for a player his size and a Moses-esque tenacity on the boards. But RPM is not impressed, and there are legitimate concerns about his defense, which is where most big man value is derived from. So if there’s a time to fix some of his defensive issues and consistency, it’s under Coach K and Thibodeau on a team that has enough scoring.
– Mason Plumlee had a decent rookie season for one of the worst rookie classes ever, and here he can concentrate on what he does best: blocking shots and finishing at the rim. But RPM saw him as one of the worst players in the league at -3. With all these negative players, it seems like an indictment on RPM, but it’s a 2014 phenomenon. Plumlee is the lowest rated player since 2002, virtually everyone else negative was either very young or a veteran who had a down season, and only one player has been negative in both seasons adjacent to the summer tournament: Carmelo in 2004, but even he was much better. However, given the bizarre Brooklyn season, I’m willing to give him a small pass here.
– Rudy Gay, Andre Drummond, and even DeMar DeRozan get honorable mentions here. Most metrics do not like Gay, and a rating of +1.53 is pretty generous but also mediocre for a Dream Team. Drummond’s box score stats are alarmingly good, but plus/minus is, predictably, bullish on a young player on a bad team. DeRozan rates as a slight negative, and I’m also not high on volume scorers without great efficiency or playmaking.
There’s a factor related here between age and real/plus minus: defense. Box score metrics have problems picking up on the subtle aspects of defense and will see high energy players without discipline as significant forces when they actually lead below average defenses. This issue is not to be dismissed, as the U.S. has been routinely embarrassed by smart players and teams comfortable with each other over the past few years.
How good could the team have been at full strength?
Paul George suffered a horrific injury, Blake Griffin had a minor one of his own, Kevin Love was dealing with free agency, and Durant was apparently exhausted. The U.S. lost four of its five best players and was reeling to find replacements. With those guys aboard, we wouldn’t have the same criticisms.
In bringing in four superstars, I gotta jettison four other players. DeRozan, Gay, and Faried are the obvious picks — Gay was brought in at the last minute as an injury replacement, DeRozan feels out of place, and Faried won’t be needed with Griffin/Love/Davis on the team. The last guy to get the boot is Plumlee, who is far from a star player — and since I’m trying to build the best possible team, his inclusion isn’t smart.
I distributed the minutes as best I could, giving the most to Harden and Durant followed by Curry, Griffin, Davis, Love, and George. The projection rises to an average RPM of 18.2, which crushes any NBA regular season team, and with a more aggressive aging curve it tops out at 20.1. The FIBA tournament rating would then be 26.8 or 27.7, based on which aging curve you trust. So they still wouldn’t be the best Dream Team in the redemption era; they’re strongly affected by Irving’s negative rating and a few guys with mediocre ratings like Rose and, surprisingly, Davis.
And with a couple small realistic changes instead, you can push the ratings to 21 to 22 with Kyle Korver and Paul Millsap on the team, instead of DeRozan and Plumlee. In fact, it’s surprising Millsap wasn’t on board because he’s a rare NBA all-star who’s a legitimate power forward with three-point range, and he adds solid defense. Korver’s record-breaking outside shooting also made him seem like an obvious pick for whenever the US was cold from outside and/or Curry and Klay Thompson were in foul trouble, which happened in an exhibition game. One issue the US didn’t fix is that they can no longer play space/pace smallball: they have no stretch 4’s, Rudy Gay is a weak three-point shooter, and their starting PF has no range.
The glass is half full: exhibition dominance
Many people are shrugging off the disappointing roster because we blew out teams in our four game exhibition tour, but I wouldn’t place a high amount of confidence in a small number of tune-up games that don’t count. Kevin Pelton had every single exhibition game and computed adjusted point differentials like I did, but the difference here is that it includes games against teams who won’t be in the tournament, and we can assume these teams are generally weaker. I tweaked his ratings by forcing the average of the tournament teams to zero, and the result is a more optimistic rating: 26.4. That’s on par with the 2010 team, the natural comparison.
But, again, these games are meaningless, and the best teams are smart enough to save their best stuff for later. The better teams have coaches who are fine-tuning rotations and will play their benches more often. This is a problem for international teams because they lack the depth of the U.S. So the exhibition schedule may in fact inflate USA’s rating.
Yet checking back on the 2010 team, their adjusted rated from my calculations was 27.1, which is quite close to their real score of 26.2 in the tournament. For strength of schedule I used tournament ratings, but this is still an indication that exhibition games may in fact have some utility. Of course, I have to offer another counter: Turkey played poorly in its 2010 exhibition games but ended the tournament with a 20 rating and a silver medal.
Then there’s the issue of Spain. NBA studies have found homecourt advantage is worth around three points, and no one would be shocked if Spain effectively had a highest advantage playing in their home country. This would push their rating past 16 or so, which should make any U.S. fan nervous.
Conclusion
After a few years of gold medals, complacency is a risk. We cannot assume we’ll blow out any team just because we have a bunch of NBA players. There have been a few close calls since 2006 with one loss versus a team that appeared to be much weaker on paper. And my ultimate example for the volatility of single game results is that the expansion Toronto Raptors beat the 72-win Chicago Bulls in 1996. Perfection is temporary.
My most optimistic estimate is an average of Pelton’s exhibition rating and my RPM projections with a strong age curve: 23.4. Even though the U.S. has a few players plus/minus hates, the system still sees them as the favorite for a gold medal, but depending on which number you use it’s not set in stone.
The comparison I keep bringing up is the 2010 team. Good news: they vastly outplayed their RPM projections, especially if you only look at their 2010 values and not the next season. On the other end of the spectrum is is the 2004 team. Kevin Pelton wrongly stated that the mid-2000’s downturn was about talent. Their RPM average was 15.3, which is roughly the same as we’ve had in 2002, 2006, and 2010. Yet they ended the Olympics with a paltry rating of 5.9. These tournaments are unpredictable and there is no certainty my numbers here, or anyone else’s, will be correct.
But nitpicking a star-filled team misses a fairly important point: the impending and seemingly inevitable gold medal game between Spain and the U.S. should be epic.
Let’s hope we get it.
