Predicting Win Totals
Is 247's Team Talent metric good at predicting Hokie win totals?
I have written before about the 247Sports metric Team Talent, and it has been on my mind lately as I think about a framework that could help us form reasonable expectations for the 2025 Virginia Tech Hokies.
Recently, I went back and pulled 10 years’ worth of Team Talent data on Virginia Tech and its opponents.
Before I looked at the data, I didn’t think the metric would be very good at predicting the outcome of individual games, but I did believe it would prove surprisingly accurate at predicting Virginia Tech’s regular season record.
In this article, I will present my findings, and where the metric was not predictive of Virginia Tech’s record, I will examine possible reasons why.
Improving over time
In the 10 years since 247Sports introduced Team Talent, it has proven a good, if not great, predictor of Virginia Tech’s success on the gridiron.
The method of prediction here is simple - the team with the numerically lower talent ranking is predicted to win every time. No partial wins are allotted for games between teams ranked closely to one another.
Although one could create such a system, the idea here is that over the course of a season and in the aggregate, it should be a wash.
What I found surprising is that the metric became a better predictor over time.
In the first four years, it correctly predicted the 2017 team’s 9-3 record (I looked only at regular season and conference championship games) but overshot the 2015 and 2018 teams’ records by two wins each while undershooting the 2016 team’s win total by two.
Those results are nothing special. Most casual fans could have matched, or exceeded, that level of performance.
However, it accurately predicted the 6-6 records in 2023 and 2024, while overshooting the 2021 and 2022 teams’ win totals by just one each.
For the 2021 team to win seven regular season games, it would have had to beat West Virginia or Syracuse. The Hokies offense stalled inside the five yard-line in the waning seconds at Morgantown and gave up a nearly 50-yard touchdown pass in the very end of the Syracuse game at home.
The Syracuse loss is the one that, for all intents and purposes, ended the Justin Fuente era.
The numerically poor 2020 and 2021 recruiting classes weighed down Team Talent since 2022, enabling this one metric to predict a poor 2022 squad and middling teams thereafter.
The 4-7 prediction for 2022 would have been accurate if the Hokies had not completely melted down in the opening game at Old Dominion (17 penalties, botched field goal snap returned for a touchdown, game ending TD drive allowed).
The 2020 (Covid) team finished two wins shy of the prediction, but that was obviously a very strange season, one in which the pandemic had an outsized impact on Virginia Tech.
Justin Hamilton did not have a real offseason to install his defense, which was different in many ways from Bud Foster’s scheme. Midway through the season, Hamilton had to revert back to a hybrid system that incorporated more of the Foster scheme, which Hamilton had played and coached in some, but not a lot.
Tech also appeared to suffer disproportionately from injuries and subpar performances driven by poor physical conditioning. Each state had different rules reflecting their voters’ sentiments toward the broader pandemic mitigation measures (masking, lockdowns, etc.), and Virginia’s state government did the Hokies no favors.
Overshooting the mark by one win in 2019 is likely the result of anomalous injuries. The Hokies were down to their third string quarterback at Notre Dame and still had a chance to win. It took an 18-play Irish drive at the end of the game to seal a one-point victory.
That was also the year that the Hokies lost their winning streak over UVA. Both of Tech’s starting cornerbacks were injured and did not play. Bryce Perkins and the Wahoo offense took advantage.
Hendon Hooker and the Hokies kept pace in a back-and-forth affair until the very end of the game, when the UVA defense made a key stop to seal the victory.
Important variables
In reflecting on the differences between predicted and actual win totals by year, a few variables come to mind that deserve further elaboration.
First, good ole fashioned luck can still throw off predictions.
The 2015 team lost both home games that went to overtime, otherwise it would have finished 8-4, as predicted.
Conversely, the Hokies lost a couple of rain-soaked one-score games that, had they gone the other way, would have also gotten the team to eight wins.
To lose all four games in which luck played an outsized role is unfortunate. Had the team won all four of those games, an equally unlikely result, team talent would have undershot the team’s win total by two.
Second, leadership from upperclassmen appears to be a major driver of team success.
The near misses in the 2015 season produced a battle-hardened 2016 team, which featured great leadership from its oldest players. That helped the team generate results that were two wins more than the sum of their parts would have led us to expect.
The third factor that impacts accuracy is character among the non-team leaders. While the Sam Rogers’ of the world outplay their recruiting ranking and make others around them better, players like Trevon Hill have the opposite impact.
The 2018 team featured too many low-character guys. When the going got tough, they either imploded or quit. Often, they did both.
Brent Pry’s increased focus on team culture, including a concerted effort to recruit high school team captains, has reduced the variability in performance from year to year.
Now, Pry just has to recruit better players.
So, if you want to know how many games the 2025 Hokies will win, check back here in late August when this season’s team talent rankings are published. At the very least, those numbers should give us a good idea of where the Hokies are likely to land this year.