Analytics is quickly becoming the go-to tool for any business. It has permeated professional sports and now college football, with schools like Ohio State using it to help win games by focusing on opponent’s tendencies. With a new season of games just around the corner, how far will analytics take teams?
College football is a huge sport with many different aspects. One of the most important is analytics, which has been used to create a better understanding of how teams are performing. Read more in detail here: analytics in football.
Last December, Mike Lombardi, a writer, broadcaster, and former NFL general manager, tweeted, “An almost playoff game and you decide to turn down points because something called analytics says it’s wise.” “You deserved to lose last night — further evidence that each game is unique and each scenario is unique; no chart or Madden-playing youngster in the basement can assist.”
It was the morning after the Los Angeles Chargers had been defeated by the Kansas City Chiefs in a heartbreaking home setback. The game itself was a thriller, with the Chiefs scoring the first 10 points and the Chargers scoring 21 of the next 24 before the game went to overtime after a back-and-forth fourth quarter. Kansas City won 34-28 after a 34-yard Patrick Mahomes-to-Travis Kelce pass 75 seconds into overtime.
The teams combined for 924 yards and 54 first downs, but the Chargers’ inability to move the chains or put the ball in the end zone was the story of the game. They were 2-for-5 on fourth downs, turning the ball over on downs in Chiefs territory at the 5-yard line in the first quarter, the 1-yard line on the last play of the half, and the 28 early in the second half.
This was nothing out of the ordinary for Brandon Staley’s Chargers. In 2021, they attempted multiple fourth-down conversions in seven games. “That’s the way we’re going to play,” Staley told the reporters after the Chiefs game. In midseason victories against the Las Vegas Raiders, Cleveland Browns, and Philadelphia Eagles, they went a combined 7-for-10, including an astounding 6-for-7 in their wild Week 18 defeat over the Raiders. However, Lombardi and other detractors mostly ignored the accomplishments.
A sporting debate is never, ever resolved. If you believe that geeks are destroying [insert sport here], you will probably always believe that. However, the outpouring of opposition to Staley appeared almost like a “final throes” moment, and the traditional nerds-in-basements stereotype seemed outdated. If the previous several seasons of pro and college football have taught us anything, it’s that the game is paying far more attention to the geeks than it used to.
How Lane Kiffin boosted the geeks
A decade or so ago, Michael McRoberts was working at a credit agency when his curiosity got the better of him.
“Some of the judgments that coaches were making simply didn’t seem right to me when watching games,” he stated. “I was on a business trip and got into Excel and began doing some numbers. The arithmetic didn’t add up with the options.”
He obtained some college football play-by-play information and began crunching stats. “What occurred to me was that there are a lot of tools out there,” he said, “but there weren’t many people attempting to assist the coaches make judgments” on fourth downs and 2-point conversions. “Could we put something together where, if a circumstance arose during the game, we could offer a coach a mathematical guideline to utilize in their decision-making process?”
The answer to this query formed the impetus for the formation of Championship Analytics, or CAI.
It began with only one school, then a few more, and then a few more. CAI now offers decision books to more than half of the FBS teams, as well as four NFL teams and teams from minor colleges and high schools.
“At the outset, there wasn’t a massive purpose,” McRoberts said. “When I was stranded in a storm in Buffalo, it was literally simply calculating some figures on a spreadsheet. And now we’ve arrived.”
Army, a CAI customer, went a combined 21-5 in 2017 and 2018, which was one of CAI’s first public triumphs. The Black Knights were masters of the margins, converting 50 of their 66 fourth-down conversion attempts and were 10-3 in games decided by one score. In an interview with the media, head coach Jeff Monken freely acknowledged the assistance offered by CAI to his club.
Only two teams attempted more fourth-down conversions than Army during this time period, but with fewer successes: Navy and Florida Atlantic, who combined for 83 fourth-down conversions. At the time, who was FAU’s head coach? Lane Kiffin is a college football coach.
When Kiffin earned the offensive coordinator job for Nick Saban at Alabama in 2014, he had already had a complete football life — USC offensive coordinator at 30, Oakland Raiders head coach at 32, dismissed USC head coach at 38. The Tide had become a CAI customer by the conclusion of Kiffin’s three seasons in Tuscaloosa, and he had had the opportunity to examine their materials, immerse himself in their game manuals, and understand just how many of his, and most coaches’, assumptions were wrong.
He immediately came to the conclusion that if and when he was offered another head-coaching position, he would commit to innovation in two ways. First, he’d blend a lot of pro-style principles and modern bells and whistles (motion, etc.) with what he dubbed “Baylor pace,” which was inspired by Baylor’s 85-play-per-game offensive in the early 2010s. Second, when it came to math, he would go all-in. He’d swiftly establish himself as CAI’s most outspoken customer.
Kiffin won a return to the SEC as head coach at Ole Miss after winning two Conference USA championships in three years at FAU, and he has stayed loyal to those early ideals. His Rebels have averaged more snaps (78.9) and fourth-down conversion attempts (3.6) per game than anybody else in the FBS over the last two seasons. When Ole Miss went 4-for-4 on fourth downs against eventual national champion Alabama and had the Tide on the ropes well into the fourth quarter, it worked splendidly. Sometimes it fails in front of all eyes, like when the Rebels went a Chargers-like 2-for-5 against Alabama in 2021, failing on the game’s opening two drives and being blown out.
“In scenarios that people aren’t accustomed to seeing,” Kiffin said, “you’re expected to go for it a lot.” “That makes it difficult because if you don’t understand it, not only does it seem awful, but it looks horrible in front of 100,000 people and on national television, and you’re facing questions from the media afterwards, and the announcers are saying how stupid that is.”
He compared the danger to blackjack: knowing the odds, when to hit on 16, and so on is one thing, but really pulling the trigger with real money on the line is another. “You’ve put thousands of dollars on that hand, and it’s a lot more difficult to hit.”
That has been true at both the collegiate and professional levels of football, but, like Staley, Kiffin has shown a willingness to hit, and it has paid off. He took over an Ole Miss team that had gone 20-28 in four seasons while working its way back from NCAA penalties, and his Rebels finished 10-3 in 2021 after a 5-5 start. Their AP poll ranking of No. 11 was their highest since 1969.
The use of fourth down by Ole Miss remains an exception, but the trends are obvious at both the collegiate and professional levels — where every NFL club now employs analytics specialists.
Fourth-down attempt rate, 2017-21
|Fourth-and-1 situation (NCAA)||60%||67%||68%||75%||74%||+14|
|Fourth-and-1 situation (NFL)||44%||57%||59%||66%||70%||+26|
|4th-and-3 or 4th-and-4 (NCAA)||22%||23%||25%||27%||28%||+6|
|4th-and-3 or 4th-and-4 (NFL)||12%||18%||19%||18%||19%||+7|
|a fourth-and-five or a sixth-and-six (NCAA)||14%||16%||15%||17%||17%||+3|
|a fourth-and-five or a sixth-and-six (NFL)||10%||11%||11%||13%||13%||+3|
FBS teams moved from going for it 60 percent of the time on fourth-and-1 to 74 percent over the course of five seasons, 92 percent of the time in opponents’ territory and 48 percent in their own. Meanwhile, NFL clubs increased their go-for-broke percentage from 44% to 70%. College teams went for it 13 percentage points more on fourth-and-2, while NFL teams went for it 23 percentage points more. In many instances, the math favors going for it on longer fourth downs even more than that, so there’s still opportunity for improvement, but that’s a significant movement in a short period of time.
Aaron Schatz of Football Outsiders observed, “The entire fourth-downs situation is visually evident.” “That shifted significantly in 2018, and things haven’t been the same since.”
“It’s exploded, frankly,” said Brian Burke of ESPN Stats & Information. “There’s lots of ways to measure fourth-down stuff, and it’s so situational that you can’t just look at raw fourth-down go-for-it rates, but even if you just look at that, it’s doubling, if only because the base rate was so low.”
“But in all of these areas, fourth downs, two points, teams don’t act optimally on them things,” ESPN’s Seth Walder remarked.
“Some [NFL] clubs have the capacity to construct their own models, while others rely on suppliers to offer models,” Burke said. “It’s being used for more than simply fourth downs. Obviously, 2-point conversions, but also things like committing deliberate penalties on the goal line, onside kicks, and time management. On the headsets, you have certain teams with an analytics expert in the booth “which may assist teams avoid wasting timeouts, among other things.
“I believe another significant shift is that teams are considering their third-down playcall in tandem with the fourth-down advice,” McRoberts said. “Third-and-7 isn’t always a throw to the sticks and a binary zero-one outcome — they could throw a pass short of it or even run to attempt to get into that fourth-down ‘go’ range, if nothing else.”
In the NFL, 35 percent of fourth downs now take 4 or less yards to complete, compared to 31% in 2017, indicating that professional teams are doing a better job of setting up these chances. (At the college level, the data hasn’t altered much.)
Kiffin said, “We teach the mindset to the guys early on here.” “They understand our aggressive style — whether it’s fakes, fourth downs, or a lot of blitzes, they realize it’s all about believing in the guys. That’s faith in your ability to make a play.”
It’s never too early to get pumped up for college football season. Here are some things we’re already looking forward to this fall. Bill Connelly
Way-too-early top 25 « | SP+ forecasts
“The discrepancy between the data from the shoulder pads and the data from the charts is massive,” Schatz added. “With the pass rush win rate and pass block win rate, for example, you can see when blockers are being beaten. The quantity of data generated and the many things you can do with it — for example, predicted running yards, which is dependent on the location of blocks and defenders at the time of the handoff — is incredible. It’s much beyond anything I could have imagined.”
When it comes to awarding credit (or blame) to a rusher and a blocker on a run play, or an offensive lineman and the quarterback on a sack, such data might be quite useful.
“I believe the most important thing that teams will strive toward in the long run is attributing value to each player on each play,” Seth said. “In soccer, there’s a common measure called xG Buildup that I particularly enjoy. It involves chaining events together to indicate how much a player contributes to his team’s predicted goals as they move the ball down the field. Consider a situation in football where you take an offense’s anticipated points added [EPA] for a play and split it by how much each player contributed to that EPA. That might have a significant impact on how we assess players on the field as well as positional value.”
Burke concurred. “Any form of individual player measure that can identify a player’s impact to a play’s success is akin to the Holy Grail. Because football is the ultimate team sport, it’s difficult to isolate individual contributions. However, tracking allows us to do just that.”
Burke was instrumental in the development of ESPN’s victory rates, and he has additional ideas in the works. “”Receivers will definitely come next,” he continued, “and defensive backs will probably come after that.” I believe that’ll be fantastic.”
Applying the same statistics to college football would be an even greater Holy Grail.
“I believe college athlete monitoring will come sooner rather than later,” Walder remarked. “There’s no way it won’t have a major influence, and a number of firms are vying for it: StatsBomb, Telemetry, SportLogiq, and Slants, to name a few. I believe all four are trying computer vision for college football. It’s possible that there are more.”
“”All I have to do is take their data and put it through the algorithms I currently have when these other firms develop a viable solution with college video-based monitoring,” Burke said. Win rates are win rates, and the X and Y data [for field dimensions] are the same. It might be for college, high school, or a father with a 4K camera and an app that allows him to follow his children’s progress in peewee football. Imagine what teams might do if that day came.”
Moneyball has come in football.
The Utah Utes, led by Kyle Whittingham, were one of CAI’s earliest customers, and one of the first big FBS clubs to get on the fourth-down bandwagon. They were down 27-24 in the final minute of a game against USC in 2016, and faced a fourth-and-1 from the Trojans’ 23. They went for the victory instead of attempting to take the points and send the game to overtime with a field goal. With 16 seconds remaining, Zack Moss gained 5 yards to move the chains, and Troy Williams connected with Tim Patrick for an 18-yard go-ahead touchdown.
“It was incredible to watch,” McRoberts remarked. “You never know what they would have done if CAI hadn’t intervened, but I’m quite confident they would have chosen to kick that field goal in the past. Moments like that mean a lot to us.”
As Army rose to prominence, CAI discovered additional opportunities like this, and Kiffin’s recent accomplishments in the SEC have offered yet another boost.
Virgil Carter, a Cincinnati Bengals quarterback and part-time math instructor, established his version of an anticipated points model more than 50 years ago. The pioneering “Hidden Game of Football,” co-authored by Bob Carroll, John Thorn, and Pete Palmer, was published over 35 years ago. It’s been almost 20 years since Schatz founded Football Outsiders (in 2003) and Burke launched his Advanced Football Analytics blog (2007). It’s been a few years since full play-by-play data became accessible for anybody with a passing understanding of Python, R, or even Excel to begin playing with and experimenting with, due to the work of The Athletic’s Ben Baldwin and others.
One might argue that football’s “Moneyball” moment had long been overdue. On a session at the recent Sloan Sports Analytics Conference, Baltimore Ravens analyst Sarah Mallepalle characterized herself as being from “Analytics Twitter” and said she came from “Analytics Twitter.” Companies like CAI and SportSource Analytics, as well as student clubs like the Michigan Football Analytics Society, which Seth used to be a member of, have found their way into the ears of more trusting college coaches. Around 2018, a sudden spike in (some) coaches’ willingness to go for it on fourth down gained traction. And, once again, we’re just seeing the visible development.
“I’ve spoken to individuals on the personnel side,” Burke said, referring to Mallepalle, “and although they’re not doing anything cosmic, they’re putting up the scouts and decision-makers with the knowledge they need.” You can go deeper [in your scouting], rule individuals out faster and save time, and that makes the front office’s work easier and more efficient.”
“”What we found early on was simply talking to coaches about their own games and scenarios that have come up in the past — like, well, here’s what you did, here’s what the stats would have advised, and here’s why,” McRoberts said. They may not really understand what you’re doing, but they’re curious to see what the figures indicate and how it relates to what they’ve done in the past “e in the past.”
Seth alluded to other sports’ tendencies, such as baseball’s greater focus on on-base percentage (or barrels, if you’re into Statcast statistics) or basketball’s growing emphasis on 3-pointers and layups, as a result of the growth of analytics.
“I’m extremely interested to observe how these new advantages are acquired in football as more and more teams learn to behave more optimally,” he remarked.
While some have claimed that these changes have made baseball and basketball more dull, football may not suffer from the same problem; few casual spectators would object to games including more high-leverage fourth-down tries and 2-point conversions, as well as maybe more passing. That, though, would just be the start of the cat-and-mouse game.
“What additional benefits do teams that are more analytically minded find?” Seth was the one who inquired. “How will they fare once the rest of the league catches up with them?”
I published a column for SB Nation seven years ago on where I believed football analytics (in particular, college football statistics) would go. I’m not sure we could have ticked many boxes a couple of years after authoring it. But, whether the old-schoolers and point-takers like it or not, the pace is quickening.
The “birkbeck football analytics” is a state of analytics in college football. The article discusses the importance of analytics, and how it can help improve performance for teams.
- football analytics websites
- college football data analytics
- data analytics in soccer
- data development in football
- what is a football data analyst