It's never easy to rank NBA teams, a task that our Marc Stein undertakes each week, but the most difficult time for evaluations might be this middle point of November. We're beyond the point where truly fluky outcomes like the Philadelphia 76ers' starting 3-0 are possible, but we're not yet far enough into the season that the standings have truly settled down. The biggest challenge for power rankings this week? The Portland Trail Blazers, who, after holding off the Toronto Raptors in a 118-110 overtime win, improved their record to 8-2 and own the third-best record in the league.
But is Portland really the third-best team in the league? Probably not. But figuring out exactly where to put the Trail Blazers (Stein ranked them seventh in his power rankings; The Hollinger Rankings have them eighth) is tricky. Here's how statistics can help.
Step 1: Throw out W-L records
As difficult as this can be to accept, NBA team rating systems generally don't include wins and losses whatsoever. When it comes to predicting the future, a team's point differential performs better because it's a truer indication of how it has performed. Last season's Blazers were an excellent example. Through early January, Portland ranked among the West's playoff teams at 20-15. However, the Blazers' minus-1.8 differential per game suggested they had actually been outplayed by opponents. Lo and behold, Portland was unable to sustain an above-.500 record and fell out of the playoff race.
Point differential works better for a couple of reasons. First, it mitigates the role of good fortune in the outcome of close games. While good teams win close games somewhat more often than their weaker counterparts, over a handful of outcomes a few bounces either way can have a significant impact on a team's record in close games. Last season's Blazers started 9-2 in games decided by five points or fewer but went 3-11 in such games the rest of the season.
Additionally, point differential doesn't throw away useful information about margin of victory. Research has also shown that blowout wins tend to do a better job of separating elite teams than close wins, or even moderate ones.
Even with Toronto's comeback Sunday, this season's Blazers haven't needed as many close wins as last season's team. Five of their eight wins have come by double figures, more than they had last season before Christmas. Still, they haven't been quite as dominant as their record would indicate, and six teams have a better point differential than Portland's plus-4.9 mark.
Step 2: Adjust for schedule
Over the course of the NBA regular season, schedules mostly even out. Because teams play conference rivals at least three times each, 72 of 82 games are against common opponents for teams in the same conference. So by season's end, the difference in schedule strength can be attributed to conference (the West historically being much tougher than the East, as is the case this season) and the team's own strength. The best teams have easier schedules simply because they cannot play themselves.
Through 10 games, however, schedules can vary wildly. As Per Diem colleague Tom Haberstroh noted last week, a light schedule was the major reason the undefeated Indiana Pacers rated behind the Golden State Warriors.
My team ratings account for this by comparing differential in each game to what would be expected based on the location of the game (home-court advantage has been worth 3.4 points per game this season) and the opponent's rating. The Miami Heat have faced the league's easiest slate of opponents, on average 2.2 points per game easier than average. (Indiana, which was last before losing to the Chicago Bulls on Saturday, now ranks 28th.) The Utah Jazz have had the hardest schedule, 2.1 points harder than average.
Add point differential and schedule strength and the result is a measure of how many points per game a team has been better or worse than league average this season. Schedule looks like a factor in the Blazers' start; so far they've played just three teams projected to make the playoffs by the ESPN Fall Forecast. But largely because the Phoenix Suns have exceeded expectations so dramatically this season, it actually rates as just 0.1 points per game harder than average.
The Los Angeles Clippers, who have a slightly weaker point differential, move ahead of Portland when schedule is factored in, dropping the Blazers to eighth in the league in my ratings. They're also eighth in the Hollinger Power Rankings, which launched for 2013-14 on Monday.
Step 3: Add in preseason expectations to define true talent
We could stop here and have a good measure of how teams have performed this season, which is what the Hollinger Power Rankings and Basketball-Reference.com's Simple Rating System are designed to evaluate. If the goal is evaluating the true talent level of teams going forward, another step is needed -- incorporating preseason expectations. And that's where things get interesting for Portland, which was widely pegged as a .500 team before the season. (The ESPN Forecast said 39 wins; my SCHOENE projection system had 41.)
Research by former Insider colleague Neil Paine, now working as a consultant for the Atlanta Hawks, sheds some light on how to balance performance to date with preseason expectations. Ten games into the season, preseason expectations still should make up more than 70 percent of the projection. By 25 games, the two measures will be equally weighted. And even at the end of the regular season, there's still predictive value to preseason expectations.
Using the average of ESPN Forecast and SCHOENE projections to set preseason expectations, the best guess is that the Blazers have the true talent of a 44-win team. Their increase of four wins in "true talent" is one of the largest in the league, though not as big as that of teams like the Phoenix Suns (plus-nine wins) and Orlando Magic (plus-five) who have dramatically exceeded low expectations.
Portland's rating going forward still puts the team just 14th in the league and eighth in the Western Conference. The Blazers have jumped the Denver Nuggets (42.4), but need to play at this level longer to confirm they have surpassed the Dallas Mavericks (45.5) and slumping Memphis Grizzlies (46.2). Wait and see isn't a fun approach to power ratings, but it's appropriate with early surprises.