Power Ranking the Virginia Tech Program

Frankly, it's tough to project any team in a sport where over half the roster changes every other year. Still, that's what we aim to do every offseason as we prep for another year of unpredictable craziness. That's what makes college football so fun, but what also makes it such a challenging sport to analyze. Unlike professional sports where rosters are mostly stable, college football teams run out 20-25 new players every year. That makes projecting future performance a really tough task.

Take Virginia Tech for example, a team where 41% of the scholarship roster is new. That's a daunting level of turnover for anyone trying to analyze the Hokies, including the coaching staff. And it's not a unique circumstance. Across the country teams are increasingly turning over rosters via the Transfer Portal and larger recruiting classes. Combine that with more underclassmen players entering the NFL than ever before, and you see the problem that analyst face. (NIL is making waves in deterring this, with only 69 underclassmen declaring for the draft in 2023. However, that number is still higher than any in the BCS era (1998-2013).)

Looking back at the Hokies, it doesn't take an insider to imagine how different the team will look this fall than last. Offensively, the Hokies will most likely have 4 players in their top 7 of touches be new to the roster in WR's Jennings, Lane, and Felton and RB Tuten. That's an exceptional overhaul, and one that is most likely not being captured correctly by a national media. That was the first step in my inspiration to create my own power ranking system this offseason that not only ranked for 2023, but captured the direction the program was heading in the future. And I'm sure glad I did.

As stated earlier, college football is too wacky and wild to simply stack up wins and loss records of two teams. Strength of opponent also varies at such a level that comparing "apples" and "oranges" is a waste of time (7-5 Mississippi State would be favored by multiple scores vs 9-3 Toledo). That means I needed a way to measure performance for 2022 without thinking about wins and losses. Enter 'Expected Points Added Margin'.

Expected Points Added (EPA) is simply a way to contextualize yardage in a football game. the easiest way to describe it is by example: A 10 yard gain from your own 20 yard line to your own 30 yard line is hardly as impactful to the game as a 10 yard gain from your opponent's 11 to your opponent's 1. That makes sense, right? And since we have drive data from past years of college football, we can quantify that impact. There are some really in depth sources online to learn about this, but let's simplify:

The Hokies have a 1st and 10 on the 50 yard line. Let's say teams who have that same down distance and yard line score an average of 3.5 points (50% of the time they get a touchdown + PAT, 50% they don't score at all). On the next play Malachi Thomas runs for 20 yards to the 30. Now at 1st and 10 on the 30, teams score an average of 5.25 points (75% of the time they get a touchdown + PAT, 25% they don't score at all). That means that on that play, the Hokies added 1.75 expected points. Go Team!

Those aren't the real numbers, but I trust you get the basic idea. EPA can also just be flipped to measure defenses, which is also very helpful when calculating an EPA margin. Why would we want to do that? Well, EPA margin tells us how well the two teams played in the game by contextualizing yards gained and/or allowed and quieting the noisy volatility of scoring plays. For the purposes of calculating, we'll take the ending EPA margin and multiply it by 10, which will give us a number between +9.4 (Georgia) and -3.52 (Colorado) to start out with.

To get all this data I scraped data from an awesome site for us college football nerds, gameonpaper.com. Over 98% of games from 2022 have advanced data on this site, so there are a few teams that have some missing info (to save a long diatribe about monopolization of sports data rights, I'll just say shout out to the MAC).

EPA doesn't on its own adjust for the quality of the opponent or the location of the game. Both of these are massive factors in college football, so they need to be taken into account with every statistical analyses. In order to do that I needed to create my own adjustments for strength of opponent and location of the game. Let's start with strength of opponent:

Strength of Opponent:

It's tough to measure strength of opponent while sticking to your guns and avoiding using wins and losses. The best ways to do this in my mind was by combining how well a team performed against the consensus Vegas spread and overall margin of victory. Thankfully, this turned out reasonable rankings. Georgia, Tennessee, Utah, Penn State, and Michigan had the 5 highest values for our strength of opponent variable. (We'll focus on the Hokies later, but for now know that the Hokies had the 96th highest value in this variable.)

Location:

Home field advantage is one of the most compelling phenomenon in college football, so measuring this accurately was important to me. Most analysts just assign a boolean value to this (home game = true, away game = false). But that implies that all away games are created equal, which couldn't be further from the truth. This started a whole new data driven journey for me where I set out to rank home field advantages. I used a combination of:

• Record against the spread at home (Do you outperform expectations in home games?)
• Home win percentage (Do you win games at home?)
• Performance at home vs performance on the road against similar teams (Do you perform better at home than on the road?)
• Home game attendance (Do fans actually show up to your games?)

I found this very interesting, and I'm sure you will too, So here's the full list:

I would like to point out that Utah absolutely dominated the home field advantage metric. Just shattered everyone. Big props to the Ute fans and Salt Lake City. Also keep in mind this isn't measuring which stadium is the loudest, but which teams get the most on field advantage from playing at home (the altitude in Salt Lake City probably plays a huge factor in this.) Both of these values, as with all of the values I input other than EPA, are multiples on a 1 to 2 scale (1 being the worst, 2 being the best).

Using both of these metrics we can adjust the EPA margins for each game based on the quality of opponent faced and the location of the game. By using a 1-2 multiple scale, teams were not penalized for playing home games, only rewarded for the difficulty of playing on the road. Similarly, you wouldn't be penalized for playing a poor opponent, only rewarded for playing a difficult opponent.

This all started with a question about how to possibly understand roster overhauls. I'll admit, I haven't cracked the code here. But I think we can use big brush strokes to paint a pretty good picture of the level of talent teams have in their program. I will say that using returning production or recruiting rankings comes with a major underlying factor: coaches have to actually develop that talent. That will be one of the fun insights of keeping up with this over the next few years. The teams that bring in / retain talent and develop it will most likely be the ones rising in the rankings. So how can we measure returning production and incoming talent? Combination of two things: Average recruit rating and 'Transferring Assets and Returning Production' (TARP).

Average Recruit Rating:

Pretty easy one here. Took the average recruit player rating over the past two classes via 247. Why the last two classes? This accounts for players new to the program and players that either redshirted or sat back and learned in their freshman seasons. I gave a 60% weight to players from the earliest class and a 40% weight to the newest class. The thought process was players in their second season are much more likely to make an impact than players in their first season. No shockers here, as the usual suspects of Georgia, Alabama, Texas, and Ohio State led the way. The Hokies came in at #49 with these weights.

This is where the gap between the Power 5 and the Group of 5 is the widest. The top 3 ranked G5 programs for this metric were Southern Methodist (#65), Texas San Antonio (#69), and South Florida (#70). You can pretty much draw a line at the top 60 and say "above this line are the big time football programs, and below this line is everyone else". Even historically bad P5 programs like Vanderbilt, Northwestern, Duke, Virginia, and California are ahead of your G5 studs like Boise State, James Madison, App State, etc. (Slightly off topic, but it seems as if NIL is yet to bridge the recruiting gap like was previously hoped.)

TARP:

TARP is a metric devised by some really smart college football minds at the Action Network as a means of measuring returning production to a team as well as production gained from the transfer portal. It weighs certain facets of production more than others (Pass Yards > Rush Yards) and assigns scores from +12 to -12, with +/- 7 points for offense and +/- 5 points for defense. There are a few pitfalls to this including FCS to FBS transfers, injuries, and contextualization of production (think EPA vs traditional yards gained). However, on the whole this metric gives a good picture of roster turnover and does a good job pointing out potential surprising teams (both TCU and Kansas State were very high in TARP to begin the 2022 season).

Just like the home field advantage and strength of opponent measures, I rated these on a 1-2 multiple scale.

It's impossible to quantify offseason news and momentum for teams. The only way this is captured is by Vegas win totals, which move according to how the market (people who pay attention to college football news) bets on these teams. I took win totals from FanDuel, Draftkings, and Caesar's as well as their prices to gauge how the market was predicting these teams. Oddly enough, these match up very well with preseason magazines like Athlon, Lindy's, and Pick Six Previews. Makes you wonder who the money movers are in this corner of the sports betting world.

From all of this we have can assign weights to values that we think are the most important and calculate a Raw_Projection_Score for 2023:

Raw_Projection_Score = (2022 Adjusted EPA * 10) + (1.5 * (TARP + Recruiting)) + (5 * Market Expectations)

or put more simply:

Raw_Projection_Score = "How you did in 2022" + "How much better/worse your roster is" + "How does the market value you"

Now all we have to do is adjust for the teams 2023 schedule and we're done! To do this I used ESPN's FPI. FPI can put out some weird numbers here and there, but on the whole does a good job of gauging the sport as a whole, which makes it ideal for strength of schedule metrics. The toughest strengths of schedule according the ESPN's FPI belong to Ole Miss, Minnesota, and Florida while the easiest schedules belong to Sam Houston, UTEP, and Liberty. The Hokies come in with the 73rd toughest schedule in the country. That all checks out.

Now after we adjust for strength of schedule, we can create a power ranking for the 2023 season that, by including recruiting numbers and market expectations, can also help us project the direction of the program. Here are my final rankings:

Real quick, here are some teams that surprised me in a good way:

• Illinois (#17)
• Boise State (#18)
• Duke (#19)
• South Alabama (#23)
• Toledo (#32)

And some teams that surprised me in a bad way:

• Rutgers (#117)
• Georgia Tech (#99)
• West Virginia (#89)
• Auburn (#57)
• Miami (#51)

If we look at this graphically, we can see the direction of programs. For that, I'll defer to a Twitter thread I put out on August 18th that's complete with visuals. If you'd like more discussion on these teams, please hit me up on Twitter (X?) at @SamOfSaturday!

Now to the fun stuff. I'm sure you noticed that the Hokies came in at #78 in my power rankings. Comparing to other numbers in the industry (#63 in ESPN FPI, #69 in SP+, #89 in beta_rank) this is on the lower side but well within the range. What can we infer from this ranking? Let's dive in!

Virginia Tech has a better roster than in 2022:

There are a few factors that play into this. Firstly, the Hokies brought in a good amount of talent on offense, and defensively only lost a linebacker and backup corner. But when you combine our recruiting and TARP variables, the Hokies come out to #71. This is because the Hokies lost their leading rusher (King) and tackler (Hollifield), and two incoming transfers that look to play key roles (Tuten and Felton) are coming from the FCS. TARP also can't see that Malachi Thomas, the Hokies's RB1, was injured most of last season. If you were to just rattle off a ranking of rosters in college football, the Hokies would most likely fall somewhere between #50-60.

Second, we would expect much better play from the QB this season, regardless of who wins the starting job. Wells will have his second year in the system and be surrounded with much better skill talent. And if Drones wins the job? His athleticism will add a dimension to the offense that supersedes any playbook. Either way, Hokies will be better at QB in 2023.

Lastly, the defense shouldn't take any steps back. Top recruits like Mansoor Delane, Keli Lawson, and Jalen Stroman should develop into star players. Lack of production on the defensive line, mainly QB hurries and sacks, is tanking the defensive TARP metric. But from a micro viewpoint, the Hokies should be just as good if not better on defense.

The market expects Virginia Tech to improve:

The consensus win total for the Hokies sits at around 6. Regardless of what the outcome would have been against Virginia in the final game, that's still a multiple game increase in wins. Even a vastly underwhelming 5-7 season would see the Hokies as a more competitive team, especially with the lack of an FCS team on the schedule this season.

I would also throw in the nugget that the Hokies are 14 point favorites against Old Dominion to start this season compared to 8 point favorites last season. That may say more about the Monarchs entering a rebuilding season and the game being in Blacksburg, but a touchdown difference year over year is a significant change nonetheless.

The schedule provides challenges, but fewer 'unwinnable' games:

Well, no game is 'unwinnable', but I think you get the point. Outside of a road trip to #7 Florida State, the Hokies won't be outmatched physically in any game. Combining TARP and recruiting like we did in the previous section (which is about as back of the napkin as analysis can be) the Hokies would have a 'better roster' than 6 of their opponents (Virginia, Purdue, Pittsburgh, Wake Forest, Marshall, and Old Dominion) and be just one spot behind Boston College. On top of that, I don't think anyone is arguing that Rutgers is significantly more talented than the Hokies.

Louisville, Syracuse, North Carolina State, and the aforementioned Florida State all have meaningful talent advantages on the Hokies. Only one of those games (Syracuse on a Thursday night) is at home. I'd have no issue calling that a 1-3 group of games at best.

Road trips to Marshall and Rutgers provide challenges against stingy defenses, but poor offenses. Purdue has massive question marks after a coaching overhaul and losses in the transfer portal, and similar questions exist with Louisville and Syracuse. Pittsburgh lost 7 defensive starters to the NFL, Wake Forest lost maybe their best player in school history, and North Carolina State is in the midst of a massive offensive overhaul themselves. Old Dominion and Virginia are simply bad football teams.

Conclusion: lots of toss up games.

Here's what our Program Power Rankings look like for the ACC and the Hokies 2023 schedule (reminder that being on or above the line indicates probable progression and below the line means probably regression).

You may have noticed in the national view that above the line was mostly Power 5 teams and below the line was mostly Group of 5 teams. This somewhat surprised me at first, but once you dive into the recruiting numbers it becomes obvious to see that on a national scale, Group of 5 teams are fighting an impossible uphill battle.

As for the Hokies 2023 opponents, the Hokies score significantly higher than Old Dominion, Rutgers, and Virginia, and are virtually tied with Purdue, Boston College, and Wake Forest. Marshall's top 10 defense nationally last season should see some serious regression, but the Thundering Herd still come in slightly higher than the Hokies (their adjusted EPA of 0.299 was 13th highest in the country last season, boosting their 2023 ranking to 47th overall).

Florida State, Louisville, North Carolina State, Syracuse, and Pittsburgh all post higher scores than the Hokies and should expect minimal to no regression, with the exception of Pittsburgh. That all checks out.

What all this means for the Hokies moving forward:

The Hokies are heading in the right direction. When I run this same analyses next season, the 2024 recruiting class and a bulk of the roster returning will bolster the expectations for the 2024 season to what we saw at the beginning of the Fuente era (7.5-8.5 Vegas win total). Barring any major transfer portal mishaps, that's a guarantee.

If we look at 2023, I struggle to see any analytical measure that would point to the Hokies being as awful as they were in 2022 again. On a macro level, the talent increased, the depth increased, and the schedule is manageable at worst. On a micro level, the Hokies were unfortunate to lose a few of the games they did last season. The opener vs Old Dominion saw the Hokies rack up a +0.333 EPA margin, the 4th highest of any losing team in 2022. Games against Georgia Tech and North Carolina State provided a similar story.

My record prediction actually bumped up a game after completing this: 7-5 (4-4 ACC)

I think Marshall, Pittsburgh, and Wake Forest are set for steps back from 2022 performance that may be greater than what I previously thought. The Hokies will win a game or two they aren't supposed to, and lose a game they should win. That's just the nature of the sport, and the nature of building a program the way Brent Pry has to this season.

I look forward to updating this weekly this season. If you'd like to follow along, check out my Twitter @SamOfSaturday where I'll post this content and much more like it!