By the (Advanced) Numbers: Duke

Having a sufficient sample size, I'm now able to dive a little deeper into the personalities of each team as I preview them from a statistical perspective...but first a tribute to our defense. Statistically of course.

Football Outsiders uses two rating systems—S&P+ (which I use for overall team ratings in previews) and FEI (an efficiency-based method). This week, these are the FEI top 5 defenses:

  1. Virginia Tech (-0.926)
  2. Stanford (-0.779)
  3. Alabama (-0.739)
  4. Michigan State (-0.672)
  5. Missouri (-0.661)

Just look at that gap. By this measure, Tech boast not only the top defense in the country, but the gap between VT and No. 2 Stanford is about the same as the gap between Stanford and No. 8 UCLA. Since the rating -0.926 is difficult to understand without context, Florida finished 2012 with the top defense at -0.795 (the more negative the better). Alabama finished 2011 as the top defense with a -0.780. And those were the highest two ratings since the metric was posted since 2007! I'm not saying we will finish with such a high rating, but by FEI standards we are watching an historic defense. Oh, what's that you say? You think I just cherry-picked a system that makes us look good? Fine, here are the S&P+ defensive ratings:

  1. Virginia Tech (169.5)
  2. Stanford (156.6)
  3. Alabama (151.9)
  4. Michigan State (147.3)
  5. Missouri (147.1)

Boom. Can we move on now?

One other note: congratulate a Hoo today, because they are now the worst team in the ACC! I'm sure their lacrosse team is doing great though.

And now the preview, starting with a comparison of our rankings in the various computer systems:

It has been noted in multiple major media outlets that Virginia Tech was a bit of a surprise in the computer ratings used for the BCS. Any positive attention is great since we started the season off radar. I do want to make the point, however, that BCS users are prohibited from using margin-of-victory, which helps us in those computers. More accurate rating systems that use MOV aren't quite as generous - struggling to beat Marshall will do that. But hey, if I told you two months ago that at this point we'd be ranked around 10th in BCS-compliant ratings and around 15th across all ratings, you wouldn't have complained, right?

Moving on to how these ratings predict the score of the game versus Duke, a 13.5-point underdog:

Seems like the computers are in agreement with Las Vegas this week despite a couple of errant predictions of a close win and one of a 30-point blowout (I like that one).

On to the overall, offense, and defense ratings and the standard line about our offense to be average while theirs struggles against our D:

At least this week we have more of an overall advantage. How bad is UVA? They just lost to this Duke team, and Duke is still the most favorable ACC matchup yet!


I'm sorry for the poor seems that when I made the macro to generate these graphs I did not anticipate ever needing a scale to go beyond 150. Looks like the Hokie defense is literally "off the charts".

Now, to a few more advanced stats used by FEI to reflect the personality of each offense and defense...for offense the numbers reflect the offense, while on defense it refers to your opponents offense. Make sense? First, some definitions:

First Down Rate - The percentage of drives in which the team either gains a first down or a touchdown

Available Yards - The percentage of possible yards gained (so if you start on your own 20 you have 80 yards to go...make it to the opponent's 40 and you've gained 40 out of 80 possible yards)

Explosive Drives - The percentage of drives that average at least 10 yards per play

Methodical Drives - The percentage of drives that run 10 or more plays

Value Drives - The percentage of drives that begin on the offense's side of the field and reach at least the opponent's 30-yard line

Rather than looking at raw numbers for each team, I will compare the stats for each field situation, starting with Virginia Tech on offence and Duke on defense:

With the exception of methodical drives, Duke's defense has a small advantage across the board against our offense. Although the differences are not large, this points to it being most likely that offensive success for Virginia Tech to come in long, methodical drives rather than a few big plays. Being small differences, however, almost anything could happen.

Now for the situation with Duke on offense and Virginia Tech on defense:

Again it appears that success for the offense will come from methodical drives, but here the other ratings point towards that being almost the only way they may have success. Virginia Tech's defense is quite a bit more adept at forcing three-and-outs (or turnovers on the first set of downs) than Duke's offense is at avoiding it, and Tech restricts opposing offenses to a small percentage of available yards while Duke's offense isn't great at gaining them. There is some hope for Duke to score on explosive drives, but the numbers aren't in their favor there either - just less in VT's. And Duke has little hope for frequent value drives that start on their end of the field and make it inside Tech's 30. The Hokies D is No. 2 in the country as preventing that.

It is often helpful to make comparisons to other teams in order to know what to do so I squared the differences between each team's "personality stats" and added them up, and then looked to see what team they most compared to.

Duke's offense is most similar to: Indiana, Wyoming, Oregon State
Duke's defense is most similar to: Tennessee (LOL Tennessee), Rutgers, Colorado State
Virginia Tech's offense is most similar to (ouch): Louisiana Tech, Penn State, Florida Atlantic
Virginia Tech's defense is most similar to: No one (didn't you read above?), Texas Tech, Michigan State

Finally, the break down of the offenses and defenses differently using the S&P+ data. A couple of definitions:

Passing Downs - 2nd down with 8 or more yards to go, or 3rd/4th down with 5 or more to go (note: this does not mean the team did pass, just that it was a situation that favored passing)

Standard Downs - Anything not a passing down

Passing - Team rating on the ability to pass (regardless of whether it is a passing or standard down)

Rushing - Team rating on the ability to rush (regardless of whether it is a passing or standard down)

Here's how things look when the Hokies are on offense:

Pretty balanced there, with VT an underdog when passing and a favorite when running...maybe we will finally see a breakout rushing attack this weekend?

And with VT on defense:

Have fun, Duke. This actually is a picture of a frustrating game potentially...Tech's worst matchup is when Duke has a long way to go, so the Blue Devils may gain good yardage when they need it, but struggle when they don't.



Love this!!
Bring on the stats!
It's really interesting that our rushing attach is favored


yeah I thought the same thing...but that might have something to do with Duke's style of defense and keeping everything in front of them

If a tree falls in Scott Stadium does it make a sound?

I was a little surprised by that too...when you look at the raw numbers though it makes sense. Our rushing rating is really poor, but their rush defense rating is even worse! In other words, a successful day rushing against Duke shouldn't be taken as a sign that we have turned the corner in that regard.

Unless it's SUPER 350 yards on the ground successful...then it could mean at least they're learning.

We may not turn a corner, but it would be nice if the run game had some success to give our guys some rhythm. I feel like one good game could be a confidence booster that may lead to more consistency in following blocks, hitting holes, etc. Right now we have problems with the O-line and the backs, but if one can start playing a bit more confidently, then the run game as a whole will improve. Not drastically, but anything would be good.

Case in point- last year at Duke was the JC Coleman 180 yard breakout game that wasn't.

These are great. My analytical mind loves it, and the fact that our defense is literally off the charts. Almost brings a tear to your eye :')

That team sure did suck last night. They just plain sucked! I've seen teams suck before, but they were the suckiest bunch of sucks that ever sucked.

Any way you look at it, that defense is ridiculously good. In person, or in the statistics.


Thanks, joelestra. This is great stuff.


Seeing the numerical confirmation of the eye test that we apply to our defense is very satisfying indeed! The thought comes to mind of the NFL teams with statistically dominant "best defense ever" type teams....85 Bears...2000 Ravens...73 Dolphins...76 Steelers...I could go on. The big pattern there is that every one of these teams won a Super Bowl. Championship type defenses, all. Lets hope this portends greatness for the Hokies.

Take the shortest route to the ball and arrive in bad humor.

I look forward to these stats each and every week. Awesome job!

Touchdown Tech - Bill Roth

Thanks for the effort on this, great stuff! Let's go!

"They've done studies you know, 60% of the time it works every time!"

I have to say that out of all the top-of-the-class stuff that is posted on this website this may be the best. This is the type of work you simply cannot find on any other team dedicated website of the Internets. Great work!

Exit Light, Enter Night....Go Hokies