Why does BPI rate Marquette higher than Duke or Kentucky?

Steve Wojciechowski will have a high-octane offense at Marquette. Zach Bolinger/Icon Sportswire

The dog days of summer are almost over -- we're now 100 days until the Champions Classic, the official start of the 2018-19 college basketball season.

With that in mind, it's time for a summer update to ESPN's College Basketball Power Index. The BPI is a predictive metric that measures how many points above or below average a team is.

The preseason rankings are based on the following categories: the quality and quantity of the returning players on the team, including transfers and players who missed last season because of injury; recruiting rankings, both overall and the number of five-star prospects; and coaches' past performances on offense and defense.

Many of the rankings look similar to qualitative preseason top 25s, including Kansas atop the rankings, but there's also some that could cause some debate. We took a deeper look at a handful of the more notable rankings.

It's only the second year the BPI has produced a summer update, but the preseason BPI ranking in October has been fairly predictive over the past decade. Since 2007, only two national champions -- both UConn teams -- have been ranked lower than fourth in the preseason BPI ranking.

That could be good news for Kansas, Gonzaga, Tennessee or Marquette.

1. Kansas
2. Gonzaga
3. Tennessee
4. Marquette

This one will immediately raise some eyebrows. The Golden Eagles finished .500 in the Big East last season, got pummeled in the Big East tournament by Villanova, then lost in the NIT to Penn State -- and then lost leading scorer Andrew Rowsey and his 20-plus points per game. However, the model projects Marquette to have the best offense in the country ... and the 90th-ranked defense. Markus Howard returns in the backcourt, and Sam Hauser could be poised for an all-Big East campaign. Steve Wojciechowski is expecting a big boost from his newcomers: Nebraska transfer Ed Morrow, Fordham graduate transfer Joseph Chartouny and freshman Brendan Bailey.

5. West Virginia

The Mountaineers lose their starting backcourt of Jevon Carter and Daxter Miles from last season, but the model still likes Bob Huggins' program. That shouldn't come as a surprise, as efficiency-based models have tended to rank West Virginia favorably since it switched to the Press Virginia style in the 2014-15 season. The biggest key in the Mountaineers' efficiency rankings is the fact they beat lesser opponents by more points than some other teams near the top of the rankings, with the nonstop pressure a main cause. Huggins will lean on James "Beetle" Bolden and a trio of freshmen to replace Carter and Miles.

6. North Carolina

7. Syracuse
8. Wisconsin

Until the NCAA tournament, neither team was all that impressive last season. Syracuse finished below .500 in the ACC before making a run to the Sweet 16. Wisconsin didn't even reach the postseason but finished by winning three of its final four games in the regular season -- including a victory over Purdue. In both cases, though, the production returning is the key factor in their top-10 ranking. Syracuse received good news when Tyus Battle decided to go back for his junior season, and Ethan Happ returned to Wisconsin. Both teams return almost all of their key players from last season. Another factor is the consistent efficiency of both programs. Syracuse and its 2-3 zone rank near the top of defensive efficiency models under Jim Boeheim, and Wisconsin and its ultra-slow system are often very efficient.

9. Duke
10. Virginia
11. Nevada
12. Florida
13. Michigan State
14. Kansas State
15. Virginia Tech
16. Villanova
17. Mississippi State
18. Auburn
19. Iowa
20. Alabama
21. Illinois

We'll group the Big Ten tandem of Iowa and Illinois together. Neither team was good last season, both winning only four games in conference play and 14 games overall. Yet the model projects both teams to take a significant jump. Iowa has the highest quantity of returning talent the model allows, which therefore gives the Hawkeyes the seventh-best offense in the country (and the No. 101-ranked defense, the lowest among any top-25 team). Meanwhile, Illinois also brings back a large amount of talent and welcomes in a solid recruiting class led by ESPN 100 point guard Ayo Dosunmu. The Fighting Illini project to have a top-35 unit at both ends of the floor.

22. Kentucky

This is the biggest eye-opener. Kentucky is widely considered to be the primary competition to Kansas for preseason No. 1 honors, yet the Wildcats are barely in the top 25 of the BPI. In general, the BPI is lower on Kentucky -- and Duke -- than consensus qualitative rankings. In fact, Duke at No. 9 is actually higher than it was in the 2017-18 preseason rankings, when most people had the Blue Devils at No. 1. The reason behind both Duke and Kentucky's lower-than-expected rankings is the fact the model heavily weighs returning production, as history shows returning starters do better than five-star freshmen in general. Of course, Duke lost its entire starting five and Kentucky saw seven players depart from last season.

23. TCU
24. Miami

We point out this one because the Hurricanes went 22-10 last season, lost in the first round of the NCAA tournament before seeing Lonnie Walker and Bruce Brown leave early for the NBA draft. But the model doesn't have Miami dropping too far, with the addition of Florida Gulf Coast transfer Zach Johnson providing a boost. While individual numbers are opponent-adjusted, Johnson's production at FGCU -- like every other transfer -- counts the same as a returning starter.

25. Texas