CBB: Duke 2010 — “Easy” Road?
Posted by Neil Paine on April 6, 2010
Note: This post was originally published at College Basketball Reference, S-R's brand-new College Hoops site, so when you're done reading, go over and check it out!
Prior to the 2010 tournament, many media pundits felt that Duke had the easiest bracket of any #1 seed, despite Kansas actually being the top overall seed in the field. If no upsets happened, Duke would have to go through AP #9 Villanova to reach the Final Four; by comparison, Kansas would have to go through #5 Ohio State, Syracuse would have to go through #7 K-State, and Kentucky would have to go through #6 West Virginia.
As the tournament progressed, the only upset that happened along Duke's path was #3 Baylor reaching the Regional Final instead of Villanova, who had been picked off by Saint Mary's (CA). This meant that instead of #9 'Nova, Duke actually only had to go through the 19th-ranked Bears to reach Indy. Once they reached the Final Four, they found #6 West Virginia waiting for them, and in the Championship Game the Blue Devils had to beat #11 Butler, whom they only topped by 2 when a pair of shots by Gordon Hayward each missed by mere inches. So you can see why some are reacting to Duke's crown today with criticism that they faced one of the easiest roads to a championship in NCAA history. But is this true? Was Duke's path to glory really devoid of potholes along the way? And if so, how does 2010 Duke compare to other past champions who had more grueling roads?
Well, let's answer the last question first: Duke 2010 was legitimately a great team, even if their coach didn't want to say it. Using SRS, look at how they stack up to past NCAA champs since 1980:
Year | School | Conf | SRS |
---|---|---|---|
2010 | DUKE | ACC | 26.099 |
2009 | UNC | ACC | 25.478 |
2008 | KAN | BG12 | 26.891 |
2007 | FLA | SEC | 23.743 |
2006 | FLA | SEC | 20.209 |
2005 | UNC | ACC | 28.425 |
2004 | CONN | BIGE | 22.620 |
2003 | SYRA | BIGE | 19.016 |
2002 | MARY | ACC | 23.554 |
2001 | DUKE | ACC | 32.273 |
2000 | MICS | BG10 | 25.036 |
1999 | CONN | BIGE | 24.743 |
1998 | KEN | SEC | 22.948 |
1997 | ARIZ | PC10 | 21.602 |
1996 | KEN | SEC | 32.142 |
1995 | UCLA | PC10 | 23.742 |
1994 | ARKA | SEC | 23.884 |
1993 | UNC | ACC | 29.037 |
1992 | DUKE | ACC | 24.798 |
1991 | DUKE | ACC | 24.901 |
1990 | UNLV | BIGW | 24.454 |
1989 | MICH | BG10 | 27.635 |
1988 | KAN | BIG8 | 15.711 |
1987 | IND | BG10 | 20.690 |
1986 | LOU | MECA | 20.659 |
1985 | VILL | BIGE | 12.029 |
1984 | GTWN | BIGE | 18.752 |
1983 | NCST | ACC | 15.217 |
1982 | UNC | ACC | 20.171 |
1981 | IND | BG10 | 21.178 |
1980 | LOU | MECA | 15.576 |
As you can see, 2010 Duke actually had one of the best ratings of any champion over the past 30 years -- even better than (gasp!) the vaunted 2009 North Carolina title team. (Don't worry Tar Heels, two UNC squads were better than the '10 Devils -- 1993 and 2005.) So you can dispel any notion that this year's Duke team was a mediocre group that got lucky when fellow top seeds Kansas, Syracuse, & Kentucky were upended, because Duke probably should have been favored over that trio even if they had held up their end of the bargain by advancing to Indianapolis:
Year | School | Conf | SRS |
---|---|---|---|
2010 | DUKE | ACC | 26.099 |
2010 | KAN | BG12 | 25.842 |
2010 | SYRA | BIGE | 22.481 |
2010 | KEN | SEC | 21.922 |
2010 | KSU | BG12 | 20.623 |
2010 | OSU | BG10 | 20.024 |
2010 | WVIR | BIGE | 19.875 |
2010 | BYU | MOUW | 19.455 |
2010 | TEX | BG12 | 19.448 |
2010 | MARY | ACC | 18.850 |
2010 | BAYL | BG12 | 18.801 |
2010 | WISC | BG10 | 18.667 |
2010 | MIZZ | BG12 | 18.333 |
2010 | CAL | PC10 | 18.048 |
2010 | PUR | BG10 | 18.009 |
2010 | GTWN | BIGE | 17.770 |
2010 | VILL | BIGE | 17.412 |
2010 | WASH | PC10 | 17.004 |
2010 | XAVR | A10 | 16.933 |
2010 | TENN | SEC | 16.874 |
2010 | CLEM | ACC | 16.841 |
2010 | MICS | BG10 | 16.445 |
2010 | MINN | BG10 | 16.401 |
2010 | TXAM | BG12 | 16.399 |
2010 | MARQ | BIGE | 16.116 |
But I still need to address the issue of how difficult Duke's road to a championship was. I can do this in two ways; first, by looking at the average SRS of the opponents they faced en route to the title:
Year | School | Conf | Region | Seed | TournW | TournL | SRS | AvgOpp |
---|---|---|---|---|---|---|---|---|
2007 | FLA | SEC | Midwest | 1 | 6 | 0 | 23.743 | 12.528 |
2006 | FLA | SEC | Minneapolis | 3 | 6 | 0 | 20.209 | 12.581 |
1995 | UCLA | PC10 | West | 1 | 6 | 0 | 23.742 | 12.724 |
1980 | LOU | MECA | Midwest | 2 | 5 | 0 | 15.576 | 12.790 |
1982 | UNC | ACC | East | 1 | 5 | 0 | 20.171 | 12.806 |
1981 | IND | BG10 | Mideast | 3 | 5 | 0 | 21.178 | 13.097 |
1994 | ARKA | SEC | Midwest | 1 | 6 | 0 | 23.884 | 13.168 |
2009 | UNC | ACC | South | 1 | 6 | 0 | 25.478 | 13.168 |
1987 | IND | BG10 | Midwest | 1 | 6 | 0 | 20.690 | 13.303 |
1986 | LOU | MECA | West | 2 | 6 | 0 | 20.659 | 13.365 |
1984 | GTWN | BIGE | West | 1 | 5 | 0 | 18.752 | 13.523 |
1990 | UNLV | BIGW | West | 1 | 6 | 0 | 24.454 | 13.660 |
2010 | DUKE | ACC | East | 1 | 6 | 0 | 26.099 | 13.738 |
1998 | KEN | SEC | South | 2 | 6 | 0 | 22.948 | 13.745 |
2004 | CONN | BIGE | Phoenix | 2 | 6 | 0 | 22.620 | 13.770 |
1983 | NCST | ACC | West | 6 | 6 | 0 | 15.217 | 13.828 |
2008 | KAN | BG12 | Midwest | 1 | 6 | 0 | 26.891 | 14.128 |
1996 | KEN | SEC | Midwest | 1 | 6 | 0 | 32.142 | 14.251 |
1999 | CONN | BIGE | West | 1 | 6 | 0 | 24.743 | 14.283 |
2002 | MARY | ACC | East | 1 | 6 | 0 | 23.554 | 14.364 |
2000 | MICS | BG10 | Midwest | 1 | 6 | 0 | 25.036 | 14.370 |
2005 | UNC | ACC | Syracuse | 1 | 6 | 0 | 28.425 | 14.854 |
1992 | DUKE | ACC | East | 1 | 6 | 0 | 24.798 | 15.254 |
2001 | DUKE | ACC | East | 1 | 6 | 0 | 32.273 | 15.278 |
1993 | UNC | ACC | East | 1 | 6 | 0 | 29.037 | 15.359 |
1985 | VILL | BIGE | Southeast | 8 | 6 | 0 | 12.029 | 15.480 |
2003 | SYRA | BIGE | East | 3 | 6 | 0 | 19.016 | 15.970 |
1988 | KAN | BIG8 | Midwest | 6 | 6 | 0 | 15.711 | 16.155 |
1991 | DUKE | ACC | Midwest | 2 | 6 | 0 | 24.901 | 16.362 |
1997 | ARIZ | PC10 | Southeast | 4 | 6 | 0 | 21.602 | 17.115 |
1989 | MICH | BG10 | Southeast | 3 | 6 | 0 | 27.635 | 17.168 |
By this measure, Duke ranks in the middle of the pack, having faced essentially the same path difficulty as 1990 UNLV, 2004 UConn, or 1998 Kentucky. The easiest paths by this metric were the Florida teams in 2006-07, the first of which was fortunate to face George Mason in the National Semifinals, the second of which drew a significantly below-average Jacksonville St. team in the opening-round. Of course, Jacksonville St. was never going to beat UF anyway, and their horribly low SRS (-12.3) dragged down what would have been a difficult path for the Gators, exposing one of the biggest flaws of taking the average opponent SRS in the tourney. Another flaw is that it doesn't recognize the extra game that teams in the early 1980s didn't have to play, since they received a 1st-round bye until the tournament expanded to 64 teams in 1985 (well, '83 NC State didn't, but that's just because they were an epic Cinderella).
How do you get around these problems? Well, instead of simply averaging the SRS of a champ's opponents, why not find the likelihood of their accomplishment using the ratings and a win probability formula? Taking the formula I used in this post, I can find the likelihood of the actual winner winning each game of the tournament; multiply the single-game win probabilities together for every champion, and you have the likelihood of them winning the whole enchilada:
Year | School | Conf | Region | Seed | TournW | TournL | SRS | p(Win) |
---|---|---|---|---|---|---|---|---|
1996 | KEN | SEC | Midwest | 1 | 6 | 0 | 32.142 | 0.605 |
2001 | DUKE | ACC | East | 1 | 6 | 0 | 32.273 | 0.464 |
2005 | UNC | ACC | Syracuse | 1 | 6 | 0 | 28.425 | 0.348 |
1993 | UNC | ACC | East | 1 | 6 | 0 | 29.037 | 0.342 |
2009 | UNC | ACC | South | 1 | 6 | 0 | 25.478 | 0.316 |
2010 | DUKE | ACC | East | 1 | 6 | 0 | 26.099 | 0.293 |
2008 | KAN | BG12 | Midwest | 1 | 6 | 0 | 26.891 | 0.291 |
1990 | UNLV | BIGW | West | 1 | 6 | 0 | 24.454 | 0.264 |
1989 | MICH | BG10 | Southeast | 3 | 6 | 0 | 27.635 | 0.245 |
1981 | IND | BG10 | Mideast | 3 | 5 | 0 | 21.178 | 0.229 |
1982 | UNC | ACC | East | 1 | 5 | 0 | 20.171 | 0.227 |
2000 | MICS | BG10 | Midwest | 1 | 6 | 0 | 25.036 | 0.217 |
1995 | UCLA | PC10 | West | 1 | 6 | 0 | 23.742 | 0.189 |
2007 | FLA | SEC | Midwest | 1 | 6 | 0 | 23.743 | 0.184 |
1994 | ARKA | SEC | Midwest | 1 | 6 | 0 | 23.884 | 0.162 |
2006 | FLA | SEC | Minneapolis | 3 | 6 | 0 | 20.209 | 0.152 |
2002 | MARY | ACC | East | 1 | 6 | 0 | 23.554 | 0.142 |
1984 | GTWN | BIGE | West | 1 | 5 | 0 | 18.752 | 0.126 |
2004 | CONN | BIGE | Phoenix | 2 | 6 | 0 | 22.620 | 0.117 |
1987 | IND | BG10 | Midwest | 1 | 6 | 0 | 20.690 | 0.094 |
1992 | DUKE | ACC | East | 1 | 6 | 0 | 24.798 | 0.090 |
1998 | KEN | SEC | South | 2 | 6 | 0 | 22.948 | 0.090 |
1991 | DUKE | ACC | Midwest | 2 | 6 | 0 | 24.901 | 0.083 |
1999 | CONN | BIGE | West | 1 | 6 | 0 | 24.743 | 0.074 |
1980 | LOU | MECA | Midwest | 2 | 5 | 0 | 15.576 | 0.070 |
1986 | LOU | MECA | West | 2 | 6 | 0 | 20.659 | 0.057 |
2003 | SYRA | BIGE | East | 3 | 6 | 0 | 19.016 | 0.031 |
1997 | ARIZ | PC10 | Southeast | 4 | 6 | 0 | 21.602 | 0.022 |
1983 | NCST | ACC | West | 6 | 6 | 0 | 15.217 | 0.012 |
1988 | KAN | BIG8 | Midwest | 6 | 6 | 0 | 15.711 | 0.004 |
1985 | VILL | BIGE | Southeast | 8 | 6 | 0 | 12.029 | 0.002 |
By this measure, Duke did have things pretty easy along the way to the title, though it's worth noting that last year's UNC team actually had it even easier. And 1996 Kentucky was ridiculously dominant, with no team they faced ever having more than a 15% chance of toppling them. However, this fact betrays a potential problem with this metric as well: namely, that it takes into account the dominance of the champion in addition to the relative ease of their path. When we say "Duke had an easy road to the 2010 championship," we don't mean "Duke was really good and they were favored in most of their games" as much as we mean "Duke's opponents were not very good and therefore any contending team would have been able to win with the same schedule." Putting aside the fact that some of these "contenders" from 2010 blew their chance to even test this theory by losing in a round where Duke was winning by 15 (I'm looking at you, Kansas), we can adjust the concept above to find the likelihood that a generic "contending team" (SRS = 23, or the average of all champs' SRS scores since 1980) wins it all with the path presented to a given school:
Year | School | Conf | Region | Seed | TournW | TournL | p(W_generic) |
---|---|---|---|---|---|---|---|
1980 | LOU | MECA | Midwest | 2 | 5 | 0 | 0.372 |
1982 | UNC | ACC | East | 1 | 5 | 0 | 0.370 |
1981 | IND | BG10 | Mideast | 3 | 5 | 0 | 0.317 |
1984 | GTWN | BIGE | West | 1 | 5 | 0 | 0.306 |
2006 | FLA | SEC | Minneapolis | 3 | 6 | 0 | 0.276 |
1990 | UNLV | BIGW | West | 1 | 6 | 0 | 0.196 |
2009 | UNC | ACC | South | 1 | 6 | 0 | 0.194 |
1985 | VILL | BIGE | Southeast | 8 | 6 | 0 | 0.174 |
1987 | IND | BG10 | Midwest | 1 | 6 | 0 | 0.171 |
1983 | NCST | ACC | West | 6 | 6 | 0 | 0.161 |
1995 | UCLA | PC10 | West | 1 | 6 | 0 | 0.159 |
2007 | FLA | SEC | Midwest | 1 | 6 | 0 | 0.155 |
2010 | DUKE | ACC | East | 1 | 6 | 0 | 0.150 |
1996 | KEN | SEC | Midwest | 1 | 6 | 0 | 0.144 |
2000 | MICS | BG10 | Midwest | 1 | 6 | 0 | 0.135 |
2008 | KAN | BG12 | Midwest | 1 | 6 | 0 | 0.132 |
1994 | ARKA | SEC | Midwest | 1 | 6 | 0 | 0.131 |
2004 | CONN | BIGE | Phoenix | 2 | 6 | 0 | 0.129 |
2002 | MARY | ACC | East | 1 | 6 | 0 | 0.123 |
2003 | SYRA | BIGE | East | 3 | 6 | 0 | 0.115 |
1986 | LOU | MECA | West | 2 | 6 | 0 | 0.112 |
2005 | UNC | ACC | Syracuse | 1 | 6 | 0 | 0.111 |
1998 | KEN | SEC | South | 2 | 6 | 0 | 0.091 |
1993 | UNC | ACC | East | 1 | 6 | 0 | 0.091 |
1989 | MICH | BG10 | Southeast | 3 | 6 | 0 | 0.077 |
2001 | DUKE | ACC | East | 1 | 6 | 0 | 0.075 |
1988 | KAN | BIG8 | Midwest | 6 | 6 | 0 | 0.068 |
1992 | DUKE | ACC | East | 1 | 6 | 0 | 0.052 |
1991 | DUKE | ACC | Midwest | 2 | 6 | 0 | 0.047 |
1999 | CONN | BIGE | West | 1 | 6 | 0 | 0.046 |
1997 | ARIZ | PC10 | Southeast | 4 | 6 | 0 | 0.036 |
Now, we can see the effect of having to play 1 fewer game coming to the fore, as the 4 easiest runs of the past 30 years were the only 4 champs who had a 1st-round bye. In the 64-team era, Florida '06 comes out with the easiest schedule, because George Mason (a great story, but one of the worst Final Four teams ever with an SRS of 12.1) was their obstacle in the National Semi. As for 2010 Duke, once more they rank near the middle of the pack, neither with a particularly tough path nor a cakewalk. UNC fans will loathe this article (and believe me, I have no rooting interest for Duke -- as a Georgia Tech grad I hate both teams), but their 2009 National Championship team had an easier road to the title than this year's Duke squad.
So while it's easy to hate Duke (just about everyone I talked to yesterday had a similar thought for the game last night: "Dear God, please let Duke lose, preferably in a blowout"), you should probably stick to mocking their graduates' well-publicized failures at the NBA level, because complaining that "Duke only won this year because they lucked into an easy path" doesn't really hold up if you look at the numbers.
April 6th, 2010 at 7:03 pm
Nice. I'm not convinced that SRS is the best way to measure the champions but the work was consistent and look for the best answer with the least bias.
March 21st, 2011 at 1:18 pm
Well publicized failures at the NBA level eh??? I'd like to see a statistical analysis of that claim honestly. Duke grads have made more money in the NBA in the last 10 years and I bet they have the best cumulative stats as well.