It’s no question that last season was a landmark one for Dartmouth football. The Big Green reached the FCS Top 25 rankings, had nationally acclaimed offensive and defensive play, eventually had players sign with or try out for professional teams and, above all else, secured a share of the Ivy League championship for the first time in 19 years. But at a more technical and even philosophical level, the team also took an innovative step.
Starting in 2015, Dartmouth partnered with the company Championship Analytics and began to receive analytics-driven advice for help with game strategy. As football continues to play catch-up to baseball and basketball in the sports analytics arena, this integration of a statistics-centered approach marks a key point of development for the program.
Central to CAI’s philosophy is simplifying the offered statistical advice, but at the same time thoroughly preparing the team for every unique circumstance it encounters on the field. Along with feedback on past games and observations from across college football, CAI sends Dartmouth a book of recommendations regarding fourth down decisions, two point conversions, penalty acceptance/decline decisions and other game aspects customized for each week on the schedule.
“We want to prepare our teams for any situation that comes up from the opening coin flip to the end of the game,” CAI founder and president Michael McRoberts said. “From a statistical perspective, they have the insight about the strategy that would help them win the game most often.”
McRoberts also mentioned that the firm takes into account whether the upcoming game will be high-scoring or low-scoring, who is the favorite and who is the underdog, any changes in confidence of the punter and kicker and the quality of the other team’s special teams returners.
During games, these charts — specific to down and distance, field positon, time and point differential — are handled by assistant coach Chad Nice, who relays the information to head coach Buddy Teevens ’79 and offensive coordinator Keith Clark through headsets.
“It allows us to make data-driven decisions and not [purely] emotional [ones],” Nice said. “It makes sure everyone is on the same page.”
Among the most fundamental components of its analytics-driven recommendations, CAI often pushes for teams to take more chances on fourth downs. For those familiar with football analytics, going for it more on fourth down represents a central tenet. Examining expected point totals of punting, field goal attempts and going for it on fourth down by field position, analyses unanimously show that teams ought to try to convert on fourth downs more often than they usually do. What CAI does differently is in advising that aggressive approach in specific situations based on opponent, score, game time, down and area of the field.
“We may reach some of the same conclusions [as the studies regarding expected points], but you still have to customize that to specific situations,” McRoberts said. “Everything we do as part of our models is associated with which [decision] leads to winning most often.”
Part of this disposition to go for it more on fourth down stems from maximizing a team’s opportunities, especially when the better team in a contest clearly exists. The favorite ought to be more aggressive and thus extend the game more by using more offensive plays, creating a larger sample for its actual ability, and the disparity of talent on the field, to manifest. This dynamic became particularly relevant for Dartmouth this past season, when the team was a favorite in every game except against Harvard University.
“If you’re the favorite, you want to play at a higher pace,” McRoberts said. “The more possession you have, the more chances you have to assert your dominance.”
The methodical preparation that CAI provides often reaps benefits at the most crucial of situations. While key moments such as fourth downs force most teams to tensely deliberate during timeouts, programs supported by CAI — granted the foresight found in the in-game recommendations — can avoid this.
“Once you get to a [third down situation], everyone on the team knows what’s going to happen on fourth down,” McRoberts pointed out. “The outcome from that is the team will be very decisive and will make very mathematically sound choices with those critical decisions.”
CAI worked with 20 teams last year, including one NFL and eight FBS teams, and is primed to double its client list for the year ahead. CAI first got in contact with Dartmouth through then-Montana State University football head coach Rob Ash, who had worked with CAI and knew Dartmouth defensive coach Don Dobes. When CAI took a road trip up north to meet with teams, that connection then lead to a meeting with Dartmouth.
Yet within Dartmouth football, consideration for incorporating analytics had already been burgeoning beforehand according to Drew Galbraith, senior associate athletics director for Peak Performance who is involved with the team.
Though the topic had been discussed for some years, the team had not been able to break through with real information to truly adopt an analytics mindset.
“This is a conversation [about analytics] that coach Teevens and I have been having off and on for [years] about how we can be more precise in our decision-making,” Galbraith said. “Oftentimes we end up doing things based on gut. So we were interested in what the math really is on this. And serendipitously the guys at [CAI] had reached out.”
Moreover, Galbraith views this integration of an analytics approach as part of a broader ethos that typifies the football program.
“It was a mindset of pushing ourselves to think about the game as creatively and innovatively as possible,” he said. “Obviously, there is a very visible manifestation in what the program has done with the non-tackling protocol and the mobile tackling dummy. This is another manifestation of that. We want to think about whether we’re doing things because of rote practice, or are we doing them because they’re the best informed decisions we can make at that time.”
In 2014, the Mobile Tackling Target was introduced as an alternative to the static dummies. Dartmouth cut back on tackling in 2010, and in February, the Ivy League officially voted to eliminate all full-contact hitting from regular season practices.
Part and parcel of actually implementing these strategies is becoming convinced in the first place. For Galbraith, that took the form of reading “Scorecasting,” a popular book that evaluated common beliefs in sports with data. Galbraith also mentioned that he follows The New York Times fourth-down bot. This automated device, popular on Twitter, judges each fourth down play in the NFL based on analysis of 10 years’ worth of NFL data, usually determining that coaches are too conservative in this aspect of the game.
Though the reaction has abated as of late, efforts to embrace a statistical focus often meets resistance in sports. Yet by all accounts — save for the kickers and punters who have seen some of their chances displaced by fourth down attempts — the reception by Dartmouth to the CAI services has been largely positive.
“Our service is really a tool for the head coach,” McRoberts said. “These are the type of decisions that fall back on them. The head coach always has to be on board with the analytics. They don’t necessarily have to follow what we’re recommending. But certainly the [Dartmouth] staff has been great and has really got into it.”
While players don’t involve themselves too much with what CAI offers, Galbraith noted that the more aggressive offensive approach — rooted in the data — bolstered the team on both sides of the ball.
“Coaches generally really positively received it, [and] same thing with the players,” he said. “There was a strong feeling among the offense that, ‘We’re the type of the team that goes for it.’ And similarly for the defense, [they] believed [they] could stop teams. There’s a layer of confidence it provided. The vast majority of players had nothing but positive things to say about it.”
For Nice, who most directly works with the analytics recommendations, the process over the last season has proved both fascinating and advantageous.
“From the top down, [people on the team] think it’s a no-brainer,” he said. “They’re just going off stats, it’s like ‘Moneyball’ [a seminal book for sports analytics]. Not everybody agrees with it, but you have to be able to take coaching and be able to adapt to what the data is telling you. It’s very interesting to see what conventional wisdom tells you to do sometimes aligns with the data, and sometimes it completely does not.”
The final decision on whether to integrate analytics rested on Teevens’ shoulders. In accordance with his track record of introducing innovative approaches to a game as rigidly and traditionally structured as football, Teevens thus deserves much of the credit for taking this path.
“Coach Teevens is an innovative head coach [and] pretty unique to find in [college] sports, particularly in the sport of the football where a lot of decision-making is made ‘just because,’ not necessarily based in science,” Galbraith said. “These things can’t ever be all science, but there’s a lot of times we’re doing things just because that’s the way they’ve been done, instead of thinking, when you play the [strategy] out over time, you can create advantages for yourself. If we can help our decision-making, why shouldn’t we?”
To some degree, those playing for Teevens also appreciate such a mindset in coaching football. Jack Heneghan ’18, a backup quarterback involved with signaling offensive plays, observed that the analytics recommendations influenced the team’s strategy, and noted the importance of his coach adopting them.
“He’s constantly looking for an edge for our team,” Heneghan said in reference to Teevens. “He’s a great coach in that respect that he’s willing to open up his mind and learn and improve. He’s been [coaching] for so long, he was a coach here in the 80s. Some coaches of that age are close-minded, but he’s really progressive, which is fun to play for.”