If you watched Duke's 79-73 loss at home to North Carolina Wednesday night, or Duke's 68-67 loss at home against Florida State on Sunday, or Duke's 68-66 loss at Virginia the game before that, you know that this isn't a great team.
What's astounding is that you'd have a very hard time convincing a computer of that. As I write this two days after the game, the USA Today computer ratings compiled by Jeff Sagarin still have Duke as the #12 team in the country even after the three-game losing streak. The more respected Ken Pomeroy rankings (kenpom.com) stunningly have Duke as the #7 team in the country.
We're taking about a team that's 5-5 in conference play. We're talking about a team that's at .500 in the league despite getting to play six of its first 10 ACC games at home. We're talking about a team whose big challenge right now according to Dick Vitale, easily the biggest Duke hyper in the history of the planet, is to finish .500 in their conference.
How could THAT team be #7 in the country, #12 in the country, or even top 20 in the country?
They can't. They're not. Top caliber teams don't go 5-5 in their first 10 conference games. Step back and look at the full season, and Duke is just 6-6 vs. teams in the top 50 according to USA Today. What voodoo do they do that tricks computers into thinking they're so good?
I think one of the key factors here involves a dichotomy we discussed during the bowls. If you weren't with us for the Major Wager Bowl Preview series, one of the main themes was that some teams look worse than they really are in the regular season because they don't run up the score on patsies, while others create illusions of greatness in the numbers by piling on the points after a game has been decided.
Assume for a moment that Team A and Team B are even in talent and ability:
*Team A may win their easy games by 14-20 points *Team B may win their easy games by 24-40 points
Any sort of computer or statistical analysis is going to tell you that Team B is much better than Team A. If they've only played soft schedules so far, it's going to look A LOT better. But, we know going in that the teams are equal. When they play each other, it's going to be a close game.
Duke is playing a lot of close games lately, aren't they?
Check out these numbers: *Duke as a double-digit favorite: 8-1 ATS this year *Duke in all other games: 4-10 ATS this year
Duke is a bully. Duke runs up the score on bad teams and creates illusions about how great they are. When matched up against quality opposition, they're just another team. Yes, they can play with anybody, and they can beat anybody if the shots are falling. But, they're not appreciably BETTER than other highly regarded teams.
Let's run through a sampling of recent ACC results and see if they strike you as something characteristic of a top 10-15 caliber team. I'm going to factor home court advantage into the results (4 points) as an equalizer, and I'll use the rankings from USA Today because I wrote those down first when taking notes:
*Duke was 10 points worse than 1st rated North Carolina
*Duke was 2 points worse than 15th rated Clemson
*Duke was 5 points worse than 24th rated Florida State
*Duke was 7 points worse than 25th rated Georgia Tech
*Duke was 6 points worse than 33rd rated Virginia Tech
*Duke was 4 points better than 27th rated Virginia
That's the five conference losses and the buzzer-beater victory over Clemson. In head-to-head action, the only "better than" in the top 35 was that tough OT road loss at Virginia.
How can that be the #7 team in the country? Or even a top 20 team? It's not. But, it's a team that can create that illusion by posting big victory margins against cupcakes. They really DO look like a national power when playing those easy opponents, the computers aren't imagining it. To get an accurate read though, the computers should be dismissing, or at least condensing, the cupcake results.
How good you are should be determined by what you do against comparable and/or quality opposition. It shouldn't be overly influenced by what happens in garbage time in blowouts.
Remember the edges Duke has against patsies:
*Duke has a solid head coach who knows how to exploit weaknesses in inferior teams.
*Duke has talent depth because they recruit well. This year's squad isn't as deep as some past entries...but the Blue Devils bench is still going to be better than the cupcakes they play early in the season, or the bottom of the barrel teams in the ACC.
*Duke typically benefits from friendly officiating in non-conference games at home. Whether that's because of something fishy or not can be debated another time. Just be aware that the team enjoyed edges in FT attempts by margins of 36-14 (Columbia), 32-14 (Georgia Southern), 34-10 (Davidson) in some of the tune-ups.
I don't want to imply that Duke does nothing but play patsies in November and December. They schedule as tough as anybody. That's actually not a very high threshold though. EVERYBODY gives themselves several easy tune-up games. In this sport where the projected blowouts can amount to a third of your schedule or more, it's easy for short term illusions to be created for the kinds of teams best positioned to run up the score on patsies.
If Duke is laying double digits...you might as well take them. It's an 89% cover scenario this year. On a level playing field though, against other good teams who match up well in talent, depth, coaching, and free throw respect, Duke suddenly looks like any other team. The hype and the blowouts lead to Las Vegas pointspreads that are higher than they should be, and Duke falls into what is currently a 29% cover scenario.
If you're an avid reader of Mess Hall posts here at MajorWager.com, you know that gamblers are pretty much aware that Duke is overrated in big games. It's hard to miss something like that when your money is at stake. Listening to Vitale during Wednesday night's telecast, it's clear that he gets it too, at least for this year's team. The question is...how are we going to make the computers get it?!
*Use a victory margin cut off of 18 points. Anything greater than that just gets entered as an 18-point win. This will reduce the irrelevant differences in how teams handle garbage time.
*Triple count games against teams in the top 25 of the rankings, and double count games against teams in the 25-50 range in the rankings. This will place a further emphasis on performance against quality opposition.
*Provide extra penalties for home losses, and extra reward for road victories. It's just nuts that a team which lost home games to both Virginia Tech and Florida State could rank in the top ten of a computer evaluation.
*Stop using a numerical system, and go to a more general letter grade system. Because of competitive balance, there's not much difference right now between a team ranked, say 15-20th in the country and one ranked 30-35th in the country. When you run your finger down the numbers, it seems like a big distance. If you just list the teams who are A, and the teams who are B, etc...then you start to look at "hunks" of evenly matched teams. That's really the more accurate prediction. Duke belongs in a hunk of teams that includes Florida State, Virginia Tech, Virginia, and Georgia Tech. Call them all A-minuses or something. The numerical separations are misleading, arbitrary, and easily polluted. Dump them!
It would be interesting to see what ratings compiled that way would look like. Hopefully, someday we'll know.
If you're a sports handicapper, be sure your power ratings or stat analysis aren't polluted by what happened in garbage games. You don't want to be laying points when two teams who are really dead even are playing each other. Value has historically been on underdogs in all sports for this very reason. The public sees differences that aren't there. For now, the computers do too.