Not sure if you’re eager for another look back at Georgia Tech’s season, but a story I wrote for myajc prompted me to try to take a different look at the numbers.
You may be dismayed to learn that, upon further review, the Yellow Jackets, as suspected, were not a very good football team. Prompted by a story I wrote for myajc regarding the College Football Playoff selection process (more below), I measured Tech’s performance against its 11 FBS opponents using those teams' season averages. The idea being that, for instance, the Yellow Jackets’ defensive play in their 30-22 loss to Notre Dame looks a little different considering that Tech held the Irish under their season averages for third-down efficiency and points per possession, but still allowed their potent offense to hit for more big plays than usual, as judged by yards per play.
I used four categories – yards per play, third-down conversion rate, points per game and points per possession. Numbers are from cfbstats.com and fbsdrivestats.com. The statistics were only against FBS competition. There’s a game-by-game comparison, a “win-loss” record (win if Tech was better than the season average, loss if it was poorer) and a season percentage (for example, Tech’s total points divided by the sum of each team’s season points-per-game average).
Georgia Tech offense vs. opponent defense
("wins" in bold)
Georgia Tech defense vs. opponents' offense
Offensively, there were just three games played well – Tulane, Pittsburgh and Florida State, and the FSU game is debatable considering the two interceptions. But in each, Tech performed better than those teams’ season averages. The rest were hit and miss.
One interesting game to me was Miami. Tech had its third highest yards-per-play average of any game besides Tulane and Alcorn State (5.7) and converted 58.3 of its third downs, highest of the season outside of Tulane and Alcorn State. (Though the Jackets couldn't manage to match Miami's yards-per-play average, which was among the highest in FBS.) But, obviously, the nine fumbles and four turnovers completely wrecked any chance to stay in the game, by killing drives, taking points off the board in one instance and by giving Miami extra scoring opportunities.
Ultimately, particularly looking at win-loss, it obviously was an ineffective unit in most games. But the record for each category also points to the sense that breaks went against Tech in the fall to 3-9. (Not that you're going to see "We should have been 6-6" on any Tech billboards next summer.)
Also, the Tulane game weighs heavily. If you toss out the results from that game, points-per-drive percentage falls from 101 to 87.5. Then again, if you toss out the Notre Dame game, the percentage goes from 101 to 107.8.
Defensively, the third-down performance (7-4 and 96 percent of combined opponents’ average) stands out. And, it’s interesting – going back to last spring, improving on third down was perhaps the highest priority for defensive coordinator Ted Roof and his unit. For the season, Tech improved from 46.2 percent (117th) to 38.9 percent (tied for 65th). But a) improving to 65th probably isn't what the Jackets had in mind; b) in yards per play, the Jackets were better than opponents’ season averages just three times out of 11, reflecting the number of big plays surrendered and occasional inability to get teams to third down.
As has been mentioned before, it was not the sort of play expected from a unit with eight returning starters.
The two points categories illustrate how statistics that don’t take possessions and other factors into account can sometimes be misleading. Tech’s defense gave up 94.6 percent of the combined scoring average of its 11 FBS opponents. However, using the fbsdrivestats’ numbers, which measures teams by productivity per possession and also discards what it considers statistically irrelevant drives, like possessions in blowouts or half-ending kneeldowns, Tech (which plays at one of the slowest tempos in FBS) was at 108.6 percent.
Regarding the methodology (such as it is), I borrowed liberally from an idea generated by SportSource Analytics, a football data analytics company founded by two brothers from metro Atlanta, Scott and Stephen Prather, and their friend Drew Borland. (Scott played baseball at Georgia Tech and then was in the St. Louis chain for five years) They won the contract to provide the data analytics platform for the selection committee, and one of the metrics was an opponent-adjusted way of looking at yards per play, points per game, third down efficiency and the like. It’s basically what I used, but presumably much more technically sound – a percentage score that takes into account the team’s performance in a game measured against the opponent’s season averages.
I don’t say this often, but I hope you'll take a look at it if you haven't. It was a fun story to write, first for their unlikely story and second for their products. Unrelated, but interesting: Scott Prather is a big Tech fan. He shares a suite with friends at Bobby Dodd Stadium and claims to have missed one Tech home game in the past 10 years. (This is funny: the one game he missed was this year’s Florida State game, which, it perhaps goes without saying, is probably the one game you wouldn’t want to have missed in the past 10 years.) He said that his group has become so large, in fact, that then-AD Dan Radakovich moved out of his own box to accommodate Prather and his buddies.
He also said that Roof is a big user of SportSource’s Coaches By the Numbers product.
“I don’t know that I’ve ever met Paul Johnson,” he said. “But I would tell him, ‘Look, besides your wife and Joel Langsfeld, I’m probably your third biggest fan.’”
(Langsfeld is an alumnus, a major donor and friend of Johnson’s who can often be found at Tech football practice.)