The first indicator last fall that the Golden State Warriors were going to emerge as a championship team came in the preseason.
During Steve Kerr's debut as coach, the Warriors went 6-2 in exhibition play with a league-best plus-11.0 point differential. That dominance was one reason ESPN's Ethan Strauss picked Golden State to win the title before the season.
Fast forward a year and while the players are still largely the same, we saw a different Warriors team this preseason.
Before blasting the Los Angeles Lakers by 39 points in its final tuneup, Golden State had lost four of its first six exhibition games, and despite the lopsided margin Thursday, the Warriors were still outscored overall.
If last year's October results were reason to believe in Golden State, should this year's be reason for skepticism?
History tells us that for some teams, preseason can be very meaningful. But that's less true for teams like the Warriors.
The meaning of preseason
ESPN.com has tracked preseason records since 2002-03. Over that span -- discounting 2011-12, when teams played just two games each -- preseason results are somewhat meaningful. Overall, teams perform better in the regular season as their preseason record improves. At the extreme ends, the average gap between a team going undefeated in the exhibition season and winless translates into the difference between a 32-win season and a 49-win one.
That's consistent with the results found by past studies of the preseason, including one by fellow Insider Tom Haberstroh five years ago.
The problem with this approach is not all teams start in the same place. We entered the season with far higher expectations for the Warriors than the Charlotte Hornets, who went 7-1 in preseason play. And the Hornets' superior performance in games that ultimately don't count can't make up that gap.
To quantify how much exhibition results should affect our outlook for teams, I turned to a database of preseason Vegas over/under win totals I've compiled dating back to the 2005-06 season. Throwing both over/unders and exhibition win percentage in a regression suggests that the over/unders predict the final record 10 times better than performance in the preseason.
At the extremes, there certainly is some meaning to the preseason, but after factoring in expectations, it's muted compared to that 17-game swing we saw earlier. The 10 teams that most outperformed their over/under win percentages during the preseason ended up winning nearly three more games than we'd have expected based on their lines, whereas the 10 teams that underperformed them the most won four games fewer -- a seven-game swing from top to bottom.
While that isn't as encouraging as the unadjusted preseason results, it's still good news for teams like the Hornets and the 5-1 Sacramento Kings, both of whom can be expected to win about three more games than their over/under would suggest because of their strong preseasons.
Where preseason isn't predictive
As important as what preseason tells us is what it doesn't tell us: the places where exhibition results are most misleading. Sorting by over/under totals reveals the most important of these: Preseason records don't have any correlation whatsoever with regular-season results for teams projected to win at least 50.5 games. That makes sense given that championship contenders are more likely to be sorting out the last spots on their rosters and giving minutes to young players than worrying about peaking in October.
The San Antonio Spurs are probably the best example of this. Dating back to 2002-03, the Spurs have gone 44-49 in preseason play, and they went 3-4 in 2013-14 before hoisting the championship trophy eight months later.
With an over/under of 50.5 wins, last year's Golden State team was right on the borderline of where preseason results matter. Coming off a title, this season's Warriors -- pegged for 60.5 wins by the Westgate Las Vegas SuperBook -- had far less urgency.
Playing without Kerr, who has taken a leave of absence following offseason back surgery, Golden State struggled at the defensive end. The Warriors allowed 101.7 points per 100 possessions in the preseason, good for 25th in the league per NBA.com/Stats and a far cry from last year's league-leading defensive rating.
Here's the good news: Preseason defense doesn't seem to have any predictive power, either. Looking at the 2010-11 and 2012-13 through 2014-15 preseasons, for which RealGM.com has complete team stats, there's a decent relationship between offensive rating in the exhibition season and the regular season. However, the variation in preseason defensive rating explains just 3 percent in the variation during the regular season.
Since defensive success relies so heavily on effort and focus, it's reasonable that teams coasting through the preseason wouldn't get the same results as when they turn things up during the regular season. We saw a taste of what the Warriors are still capable of defensively against the Lakers, who posted a 46.3 effective field goal percentage and turned the ball over 23 times Thursday.
With so many of the West's other top teams improving and the Cleveland Cavaliers seeking a return trip to the NBA Finals, there are certainly reasons to believe Golden State might not be able to repeat. But the way the Warriors played in the preseason shouldn't be one of them.