By the (Advanced) Numbers: Rating the U

It's hard to move forward to the Miami game without first cleansing the wound suffered against Liberty.

Obviously the focus has been on the end of the game, but before diving into that let's cover the game up until that point. It certainly felt like the Hokies were losing that game, but the statistics don't really back that up. While Liberty outgained Virginia Tech by 48 total yards (5 of which are negated by penalty yard differential), the Hokies had a clear efficient advantage in both passing (+0.8 YPP) and rushing (+1.5 YPP). Tech also gained 69% of available yards to Liberty's 66%. Given the stats, according to Bill Conneley's post-game win probability the Hokies were 68% likely to have won the game.

Want to talk about Malik Willis?

Hooker matched his passing yardage and touchdowns on three fewer attempts, and out rushed him by 48 yards on just one additional attempt. But on social media I read gripe after gripe about how he could have been playing for Tech, from a fan base apparently too distraught by another letdown performance to appreciate how good the quarterback at the helm of the Hokies is now. Put another way, Willis would be Hooker's backup.

On to the end of the game. There's an obvious information bias in knowing that the Hokies blocked the first field goal attempt and returned it for a touchdown, but let's back up to the game management in general. The Flames lined up for what would be a game-winning field goal from 59 yards and at that point, Justin Fuente must quickly decide the strategy for the final seconds of the game. There are four paths, and each has its own probability of a Hokie win.

1. Block the field goal attempt and return for a touchdown
2. Field goal misses and the Hokies score a touchdown on the single play remaining time allows
3. Field goal misses, the Hokies don't score, and the game goes to overtime
4. Field goal is made and the Flames win the game

I realize there are some other remote possibilities (such as field goal is made but Hokies return the ensuing kickoff for a touchdown) but none that would seem to have any reasonable likelihood. So let's break down the probability of each, assuming no timeout is called:

1. Nationally, about 2.5% of kicks have been blocked this season. Now longer kicks are a little easier to block because of lower ball trajectory, but it's hard to imagine the odds of this scenario happening any higher than the 5-10% range.
2. Here you are talking about Virginia Tech needing to score a 51-yard touchdown given a single play against a team who only needs to stop a touchdown and is in full prevent defense. We don't have enough data on this exact scenario, but the Hokies have scored touchdowns on offense 4.4 times per game on 67.9 plays, so that's a 6.4% chance if you ignore yardage and defensive scheme, both of which substantially drop this number. At best this option is in the 1-2% chance range.
3. A higher-rated team that has slightly outplayed the other opponent during the game has a higher than 50% chance to win in overtime, but let's be conservative and say overtime is a 50% chance of a win.
4. Liberty ranks 73rd nationally in field goal percentage at 66.7%, but given small sample size perhaps they get the benefit of the doubt and we use national averages. Watchstadium.com provides a dataset of kicks made and attempted by distance, and from that a model can estimate the chances of an average kicker making a 59-yard field goal: 25.6%.

Ignoring the timeout call, the best path to a win is clearly to take the game to overtime and settle it there. So playing the odds, you simply let the field goal play happen, obviously trying to block it, but knowing it is a fairly low probability kick anyway.

But Fuente did call a timeout, and it's worth questioning what purpose that served no matter what excuses were provided to the media after the game. The only justification for calling a timeout there is that you believe that timeout will increase the odds of #3, or will increase the odds of another outcome beyond the odds of #3. The former is unlikely to matter significantly — studies on whether icing a kicker is effective have provided mixed results, and otherwise personnel is pretty irrelevant since the chances of making the kick are on the kicker and holder and the chances of winning in overtime have nothing to do with that play.

As for the latter, Fuente cited wanting to have the right personnel in for a blocked kick. Even if we didn't know the first kick was blocked, that's trying for an incremental increase in the odds of a low probability play and doesn't seem worth it. By far your best odds of a win are not chaos — they are to simply let Liberty take the kick, try your best to block it, and then play it out in overtime.

It is especially painful for fans to know that by accident that scenario did play out, and the Hokies' best turned out to deliver the low probability outcome only to be undone. I won't deep dive into the subsequent play call that allowed the Flames to instead win on a 51-yard-field goal, but those eight yards upped the field goal chances to 43.4%. The Hokies were then in a position where allowing the kick to happen was a high risk proposition, and instead needed chaos for a chance at victory. It didn't happen.

Nothing will make us forget that outcome, but beating an (overrated?) 9th-ranked Miami team sure would help with the healing...

Where the Season Stands

Virginia Tech is currently 4-3. The actual and predicted score difference and odds of winning each game are:

The odds of each possible regular season win total are now:

So a 2% chance fans have seen their last win of the regular season, but almost an 8% chance they've seen their last loss. That's the rosiest I can paint this picture.

Based on FPI, the odds of each ACC team beating the ACC teams on their schedule is:

Clemson is still in great position for a rematch against Notre Dame in the ACCCG, but it feels odd that the team with the greatest chance of putting that in peril is...Virginia Tech. Both Miami and UNC are sitting on one conference loss, so a Hokie win against Clemson (currently about a 20% chance) makes this an interesting race if either were able to win out.

But let's get more unrealistic and interesting. Suppose Virginia Tech wins out (7.6% chance). That leaves Clemson, Miami, UNC, and Wake Forest with at least two losses. But Miami plays UNC, so 2-loss Miami makes UNC also a 2-loss team. The Hokies would have the tiebreaker among the 2-loss teams in every scenario except for Wake Forest running the table (which would leave UNC, Clemson, VT, Wake Forest; WF would win the tiebreaker at a 2-1 records against the others) which is perhaps the least likely outcome among those.

To summarize, VT running the table combined with Miami beating UNC is a likely ACCCG game appearance for Virginia Tech.

Virginia Tech Leads the Nation In...

I can't believe I'm about to write this after last weekend's result but it's just so timely.

Virginia Tech is tied for last in the nation in opponent field goal percentage, having allowed opponents to make every field goal attempted this season.

The qualifiers and commentary on that fact are so obvious and painful I won't even bother to write them.

I will, however, provide some additional statistical quirks to the fact. Despite a 100% opponent field goal rate, Virginia Tech has given up fewer field goals per game than all but six Power 5 teams that have played at least five games. In fact, with no qualifiers, the Hokies are 9th nationally in opponent field goal attempts per game at just 0.9 — opposing offenses are feast or famine against the team. In the red zone, teams are scoring touchdowns 78.6% of the time, 11th nationally and the worst mark of any Power 5 team with at least five games.

Of the six trips that did not result in a touchdown, only one was an attempted field goal; otherwise teams are going for it on 4th down (as they statistically probably should), a poisonous combination against a Hokie defense that is 106th nationally in opponent 4th down conversions (76.2%).

Teams aren't trying to kick field goals against Virginia Tech on 4th down because they believe they can convert. And they are usually right.

Rankings and Computer Predictions

The computer rankings and predictions for each team:

Home field advantage is definitely playing a part and has most computers at roughly a toss-up in this game. Only one computer rated the Hurricanes higher than the polls. 23 computers rated them worse and 12 rated them more than 10 spots worse. The odds of a 2.5-point favorite winning is 57.5%.

Next is a look at any overall offensive or defensive advantages:

Despite the Miami offense receiving plenty of attention, the Hurricanes are actually much more balanced than that and feature above average, but not elite, offense and defense. Tech features a nearly elite offense, and from what I understand also features a defense.

Who To Watch Out For

It's never a question whether the U has talent on their roster but whether they convert that talent into wins:

1. QB D'Eriq King has received national attention for his performance thus far since transferring from Houston. As a passer, King's numbers are good but not elite — a 152.5 passer rating (31st nationally) with a 63.8% completion rate (41st) and 8.2 YPA (36th) all trail Hendon Hooker. But King adds another 58 yards on the ground at about 5 YPC, and, well, running quarterback. The good news is that he is averaging about 5 passes defended per game (INT+PBU), and Miami gives up 2.71 sacks per game (86th nationally).
2. LB Quincy Roche has 10 TFL on the season (10th nationally).

Statistical Key to the Game

Number of healthy Khalil Herbert's playing? Because I feel good about anything over 0.5.

One of the most ignored statistical aspects in football is penalty yards. Yards are yards, and a team that is penalized 30 more yards than their opponent has a 30-yard differential; they now have to outplay the opponent to make up those yards. Miami ranks 108th national in penalty yards per game at 75.4, and 110th in penalties per game at 8.6. VT commits 5.9 for 53.9. I'm not sure how predictive it is (or more a function of who you play) but the two teams rank 12th and 16th nationally in opponents penalty yards per game.

Situationally, the value of those yards might be more or less. For example, a penalty that gives your opponent an automatic first down on a 3rd-and-23 they weren't otherwise going to make is much more costly than the direct penalty yards. Similarly, a false start on a short field goal wasn't particularly damaging.

Let's be honest, the Hokies are unlikely to bottle up the Miami offense completely, or steamroll their defense. In a game this close, just a few critical mistakes can sink someone's ship. For me, penalties may very well do that in this game — the team that has more opponent drives sustained by penalties and more offensive drives stalled by them may put themself at too much of a disadvantage to overcome. And the Hurricanes are more likely to be that team.

Statistical Prediction

There was something about the Liberty game that just felt like it was going to be a loss.

And I don't know why, but there is something about the Miami game that just feels like a win.

Virginia Tech 37, Miami 34

As always a thanks to ESPN, Football Outsiders, cfbstats.com, and Minitab Statistical Software.