- Joined
- Apr 6, 2013
- Messages
- 307
Number crunching a growing craze in the NFL
Jarrett Bell, USA TODAY Sports 9:48 p.m. EDT May 15, 2013
(Photo: John David Mercer, USA TODAY Sports)
Story Highlights
Demoff is looking for an edge to accelerate the rebuilding.
Get a franchise quarterback? Check. Sam Bradford is preparing for his fourth season.
Hire a new coach? Check. Jeff Fisher, a proven winner, is changing the culture in his second year at the helm.
Develop an analytics department? Coming.
The Rams are among a growing number of NFL teams embracing the notion of incorporating advanced statistics into their football operations to supplement the make-or-break equations of the NFL: X's and O's, salary-cap dollars and W's and L's.
NFL team, meet Moneyball.
"There is so much more data out there than ever," says Demoff, 35, in his fifth year with the Rams. "So there's a hunger to evaluate it. Somewhere, there's a lot of secrets in the data. Maybe it's finding the next Russell Wilson or Jerry Rice. Or maybe it's the key to reducing a certain type of injury."
SHAKEUP: Jets make another front office move
NEW HIRE: Bills tab Doug Whaley as general manager
Within the last few months, new analytics departments in the NFL have been popping up like spring crab grass. The Cleveland Browns, Jacksonville Jaguars and Baltimore Ravens have launched efforts to merge deeper data analysis into football decisions. And Buffalo Bills President Russ Brandon says starting an analytics department is on his agenda, too.
New Kansas City Chiefs coach Andy Reid recently lured Mike Frazier from the Philadelphia Eagles as his statistical analysis coordinator, Frazier's second NFL job. Before his nine years with the Eagles, he had internships with Smith Barney and Wachovia Securities. Finance is not always the path to a key NFL front office job, but it's increasingly representative of a skill set being tapped to influence football moves.
The Eagles, meanwhile, are continuing an investment into analytics that has existed for nearly a decade, with full-time staffers employed in a department that falls under the coaching and scouting umbrella.
"Every team in the league is going to spend the next two or three years to see if they can build a better analytics department," Demoff says. "Everybody's trying to see if there's a silver bullet."
Eliminating guesswork
How are analytics most relevant in the NFL? It varies, depending on the philosophies of any given team's power brokers. Generally, they apply in:
The draft process, in sorting through biomechanical metrics to rank prospects.
Salary-cap management, to project, for instance, the long-term impact of contracts.
Game management, providing statistical probability for key decisions.
Free agency, to account for variables behind statistics.
"It's a lot more widely used than most people would advertise," Atlanta Falcons general manager Thomas Dimitroff says. "It's been around for a long time. We use analytics to eliminate as much guesswork as we possibly can."
For generations, NFL teams analyzed the tendencies of opponents to, say, determine where they stood a better chance for success on a third-and-3 pass. For decades, they have timed prospects in the 40-yard dash, for instance, as part of scouting evaluations. And for years, Bill Belichick and the New England Patriots have flourished behind a cap management system that often has maximized the value of veteran free agents.
Such longstanding practices are why several NFL executives scoff at comparisons to Moneyball, the best-selling book published in 2003 and ensuing film that hailed Oakland Athletics GM Billy Beane's use of advanced statistics to help build a winner despite limited cash.
But there's at least a philosophical correlation to Moneyball in the idea of getting more bang for the buck, always a factor in the salary-cap system of the NFL. And there's an undeniable evolution within the NFL, fueled by technology and a younger generation of GMs and top executives who are seemingly more open-minded than many of their tradition-steeped predecessors.
"It's becoming more about what you know, not who you know," says Paraag Marathe, chief operating officer for the San Francisco 49ers.
It's no wonder that 18 NFL teams — including already-established analytics-influenced operations such as the 49ers and Falcons — were represented at the MIT Sloan Sports Analytics Conference in Boston in March.
The tech-aggressive 49ers are hellbent on applying new methods. This year, the 49ers struck a deal with SAP Software Solutions to develop a revolutionary draft app, SAP Scouting, that allows information within a central database to be processed in real time.
If a scout visiting a pro day at, say, a Southeastern Conference school enters a 40-yard time of a running back while another scout is observing another running back at, say, a Big Ten school and has a 40 time to report, the information can be shared immediately. Furthermore, the 40 times can be stacked up against every other running back — or every other player — in the nation.
The same concept applies to any type of quantitative data for an infinite number of prospects, and data can be used throughout a player's career.
Marathe broke into the NFL while working as a financial analyst for Bain and Company, which late 49ers architect Bill Walsh contracted for a project to evaluate the value of draft slots — something like an exchange rate — before the 2001 draft. Later that year, Walsh hired Marathe, who began as a salary-cap researcher.
Now Marathe is the chief negotiator and administrator of the team's cap.
A glance at the 49ers' salary structure provides clues about why the team seems well-equipped to remain a consistent contender after advancing to the NFC title game and the Super Bowl, respectively, over the last two seasons. The 49ers' cap is largely balanced and manageable, without exorbitant cap numbers for key players.
Although all-pro inside linebacker Patrick Willis had a $17.76 million cap number last year, his contract was structured to lower the figure to $3.65 million in 2013. With the trade of backup quarterback Alex Smith ($9.75 million) to the Chiefs, the highest cap number on the books is tight end Vernon Davis' $8.74million.
It's unrealistic to think they can retain every productive veteran who becomes a free agent, but Marathe vouches for analytics as a tool that can be used for timing some contracts to stretch cap dollars and get maximum returns.
"We don't want to be the Florida Marlins of 1997 or 2003, when you win and then break up the team," Marathe says.
New ways of looking at stats
That Marathe referenced a Major League Baseball franchise brings other parallels to mind. In Moneyball, a key statistical distinction weighted on-base percentage with a higher value than the traditional measure of a batter's success, batting average.
In an NFL context, Browns CEO Joe Banner says analytics influenced him, for example, to read less into a player's sacks and more into an equation of sacks plus quarterback hurries as a starting point to incorporate other variables. Opponents' schemes, teammates and circumstances such as stunts and one-on-one matchups also factor in.
"That's the value of it," Banner says. "It's one thing to say a guy had nine sacks. How many times was he blocked by the tight end, rather than the right tackle?"
When he conducted the search for a Browns coach, which ultimately led to Rob Chudzinski's hiring, Banner says he discussed analytics with each candidate, wanting to gain a sense of how they would be incorporated into game planning, the draft and free agency.
Banner contends there were robust exchanges during interviews about methods to apply advanced statistics. He chuckled when asked to recall the mixed messages he received from Eagles coaches and other staff when he started emphasizing analytics around 2004 while serving as the team's president.
"Some people embraced it," Banner says. "Some people understood that it was an extension of quality control. And some thought it was a foreign object."
Jaguars general manager Dave Caldwell sees analytics as a piece of the puzzle. Caldwell aims to build the old-fashioned way, through draft picks and the continuity of keeping the best talent that they can develop. Yet he embraces new tools.
The Jaguars have committed two full-time staffers to an analytics department headed by Tony Khan, the tech-savvy son of owner Shahid Khan and senior vice president.
"They are going to give us a lot of support to think outside the box," Caldwell said.
To that degree, Caldwell has had some experience.
Jones trade as an example
Caldwell, previously a personnel director with the Falcons, says a classic example of the impact of analytics occurred when Dimitroff swung the draft-day trade that netted big-play receiver Julio Jones with the sixth pick in 2011. Atlanta gave up a bundle to Cleveland to move up from the 27th slot: two first-round picks, a second-round pick and two fourth-round picks.
The analytics included statistical research weighing the risk of how high to take a receiver in the draft and whether blue-chip wideouts were more likely to become draft-day busts when compared to other positions. Dimitroff sought Jones to increase explosive plays that he felt his offense lacked.
"We knocked off statistic after statistic," Dimitroff says.
He says the offer didn't hinge on the first- and second-round picks dealt. "Everybody understands the first- and second-round picks," he says. "We needed to figure out the value of the other picks."
What did the research reveal? Less than 15% of fourth-round picks become starters.
In two seasons, Jones has blossomed into one of the NFL's budding stars on one of the league's most productive offenses. He earned his first Pro Bowl berth after the 2012 campaign.
Dimitroff knows that advanced metrics are hardly foolproof. Yet they can go a long way toward identifying which of the five linebacker prospects on the board has a better shot at succeeding in the Falcons system or what value to place on a veteran.
"In the end, you still have to make the call for the right reason," Dimitroff says.
COACHES VIEW: In games, backlash can outweigh odds
It's fourth-and-3, and you're at the opponent's 37-yard line.
Go for it? Try a 54-yard field goal?
According to the statistical model developed by Brian Burke, it's better to keep the offense on the field. The math suggests teams have a 56% chance of converting on fourth down in that situation vs. a 47% chance of making the field goal.
Burke, an ex-Navy pilot, has crunched the numbers to such a degree that he has computed the odds for decisions based on a wide range of situations that make up his Win Probability Model. He has written extensively about his findings since 2007 on his website, AdvancedNFLStats.com, and says he has consulted for three NFL teams.
Still, while many teams have become progressive in applying analytics in areas such as the salary cap, game management is a tougher sell.
Playing the odds might work on paper, but for coaches whose jobs can be on the line, the risks have an added dimension.
Brian Billick interviewed with the Philadelphia Eagles.(Photo: Matthew S. Gunby, AP)
Brian Billick, who won Super Bowl XXXV coaching the Baltimore Ravens and now works as an analyst for Fox Sports, recognizes the value mathematical equations can offer coaches with decision-making.
But he points out researchers don't deal with the risk. "If I go for it on fourth-and-3 from my own 20 and it doesn't work, I want that son of a (gun) standing next to me on the sideline when people are throwing things," Billick said.
Denver Broncos defensive coordinator Jack Del Rio bristles when thinking of the second-guessing he endured when game-management decisions backfired during his tenure as Jacksonville Jaguars coach.
"You may have statistical probability on your side, but if it doesn't work, you're going to get ripped for not going by the book," Del Rio said. "The book is being rewritten all the time. There's an old book that we grew up with in the '70s and '80s. I don't think that book applies in the NFL anymore."
http://www.usatoday.com/story/sports/2013/05/15/advanced-statistics-nfl/2164723/
Jarrett Bell, USA TODAY Sports 9:48 p.m. EDT May 15, 2013
(Photo: John David Mercer, USA TODAY Sports)
Story Highlights
- Numerous teams are creating analytics departments to support coaches and front office staff
- Analytics played a role in the Falcons decision to trade up for wide receiver Julio Jones
Demoff is looking for an edge to accelerate the rebuilding.
Get a franchise quarterback? Check. Sam Bradford is preparing for his fourth season.
Hire a new coach? Check. Jeff Fisher, a proven winner, is changing the culture in his second year at the helm.
Develop an analytics department? Coming.
The Rams are among a growing number of NFL teams embracing the notion of incorporating advanced statistics into their football operations to supplement the make-or-break equations of the NFL: X's and O's, salary-cap dollars and W's and L's.
NFL team, meet Moneyball.
"There is so much more data out there than ever," says Demoff, 35, in his fifth year with the Rams. "So there's a hunger to evaluate it. Somewhere, there's a lot of secrets in the data. Maybe it's finding the next Russell Wilson or Jerry Rice. Or maybe it's the key to reducing a certain type of injury."
SHAKEUP: Jets make another front office move
NEW HIRE: Bills tab Doug Whaley as general manager
Within the last few months, new analytics departments in the NFL have been popping up like spring crab grass. The Cleveland Browns, Jacksonville Jaguars and Baltimore Ravens have launched efforts to merge deeper data analysis into football decisions. And Buffalo Bills President Russ Brandon says starting an analytics department is on his agenda, too.
New Kansas City Chiefs coach Andy Reid recently lured Mike Frazier from the Philadelphia Eagles as his statistical analysis coordinator, Frazier's second NFL job. Before his nine years with the Eagles, he had internships with Smith Barney and Wachovia Securities. Finance is not always the path to a key NFL front office job, but it's increasingly representative of a skill set being tapped to influence football moves.
The Eagles, meanwhile, are continuing an investment into analytics that has existed for nearly a decade, with full-time staffers employed in a department that falls under the coaching and scouting umbrella.
"Every team in the league is going to spend the next two or three years to see if they can build a better analytics department," Demoff says. "Everybody's trying to see if there's a silver bullet."
Eliminating guesswork
How are analytics most relevant in the NFL? It varies, depending on the philosophies of any given team's power brokers. Generally, they apply in:
The draft process, in sorting through biomechanical metrics to rank prospects.
Salary-cap management, to project, for instance, the long-term impact of contracts.
Game management, providing statistical probability for key decisions.
Free agency, to account for variables behind statistics.
"It's a lot more widely used than most people would advertise," Atlanta Falcons general manager Thomas Dimitroff says. "It's been around for a long time. We use analytics to eliminate as much guesswork as we possibly can."
For generations, NFL teams analyzed the tendencies of opponents to, say, determine where they stood a better chance for success on a third-and-3 pass. For decades, they have timed prospects in the 40-yard dash, for instance, as part of scouting evaluations. And for years, Bill Belichick and the New England Patriots have flourished behind a cap management system that often has maximized the value of veteran free agents.
Such longstanding practices are why several NFL executives scoff at comparisons to Moneyball, the best-selling book published in 2003 and ensuing film that hailed Oakland Athletics GM Billy Beane's use of advanced statistics to help build a winner despite limited cash.
But there's at least a philosophical correlation to Moneyball in the idea of getting more bang for the buck, always a factor in the salary-cap system of the NFL. And there's an undeniable evolution within the NFL, fueled by technology and a younger generation of GMs and top executives who are seemingly more open-minded than many of their tradition-steeped predecessors.
"It's becoming more about what you know, not who you know," says Paraag Marathe, chief operating officer for the San Francisco 49ers.
It's no wonder that 18 NFL teams — including already-established analytics-influenced operations such as the 49ers and Falcons — were represented at the MIT Sloan Sports Analytics Conference in Boston in March.
The tech-aggressive 49ers are hellbent on applying new methods. This year, the 49ers struck a deal with SAP Software Solutions to develop a revolutionary draft app, SAP Scouting, that allows information within a central database to be processed in real time.
If a scout visiting a pro day at, say, a Southeastern Conference school enters a 40-yard time of a running back while another scout is observing another running back at, say, a Big Ten school and has a 40 time to report, the information can be shared immediately. Furthermore, the 40 times can be stacked up against every other running back — or every other player — in the nation.
The same concept applies to any type of quantitative data for an infinite number of prospects, and data can be used throughout a player's career.
Marathe broke into the NFL while working as a financial analyst for Bain and Company, which late 49ers architect Bill Walsh contracted for a project to evaluate the value of draft slots — something like an exchange rate — before the 2001 draft. Later that year, Walsh hired Marathe, who began as a salary-cap researcher.
Now Marathe is the chief negotiator and administrator of the team's cap.
A glance at the 49ers' salary structure provides clues about why the team seems well-equipped to remain a consistent contender after advancing to the NFC title game and the Super Bowl, respectively, over the last two seasons. The 49ers' cap is largely balanced and manageable, without exorbitant cap numbers for key players.
Although all-pro inside linebacker Patrick Willis had a $17.76 million cap number last year, his contract was structured to lower the figure to $3.65 million in 2013. With the trade of backup quarterback Alex Smith ($9.75 million) to the Chiefs, the highest cap number on the books is tight end Vernon Davis' $8.74million.
It's unrealistic to think they can retain every productive veteran who becomes a free agent, but Marathe vouches for analytics as a tool that can be used for timing some contracts to stretch cap dollars and get maximum returns.
"We don't want to be the Florida Marlins of 1997 or 2003, when you win and then break up the team," Marathe says.
New ways of looking at stats
That Marathe referenced a Major League Baseball franchise brings other parallels to mind. In Moneyball, a key statistical distinction weighted on-base percentage with a higher value than the traditional measure of a batter's success, batting average.
In an NFL context, Browns CEO Joe Banner says analytics influenced him, for example, to read less into a player's sacks and more into an equation of sacks plus quarterback hurries as a starting point to incorporate other variables. Opponents' schemes, teammates and circumstances such as stunts and one-on-one matchups also factor in.
"That's the value of it," Banner says. "It's one thing to say a guy had nine sacks. How many times was he blocked by the tight end, rather than the right tackle?"
When he conducted the search for a Browns coach, which ultimately led to Rob Chudzinski's hiring, Banner says he discussed analytics with each candidate, wanting to gain a sense of how they would be incorporated into game planning, the draft and free agency.
Banner contends there were robust exchanges during interviews about methods to apply advanced statistics. He chuckled when asked to recall the mixed messages he received from Eagles coaches and other staff when he started emphasizing analytics around 2004 while serving as the team's president.
"Some people embraced it," Banner says. "Some people understood that it was an extension of quality control. And some thought it was a foreign object."
Jaguars general manager Dave Caldwell sees analytics as a piece of the puzzle. Caldwell aims to build the old-fashioned way, through draft picks and the continuity of keeping the best talent that they can develop. Yet he embraces new tools.
The Jaguars have committed two full-time staffers to an analytics department headed by Tony Khan, the tech-savvy son of owner Shahid Khan and senior vice president.
"They are going to give us a lot of support to think outside the box," Caldwell said.
To that degree, Caldwell has had some experience.
Jones trade as an example
Caldwell, previously a personnel director with the Falcons, says a classic example of the impact of analytics occurred when Dimitroff swung the draft-day trade that netted big-play receiver Julio Jones with the sixth pick in 2011. Atlanta gave up a bundle to Cleveland to move up from the 27th slot: two first-round picks, a second-round pick and two fourth-round picks.
The analytics included statistical research weighing the risk of how high to take a receiver in the draft and whether blue-chip wideouts were more likely to become draft-day busts when compared to other positions. Dimitroff sought Jones to increase explosive plays that he felt his offense lacked.
"We knocked off statistic after statistic," Dimitroff says.
He says the offer didn't hinge on the first- and second-round picks dealt. "Everybody understands the first- and second-round picks," he says. "We needed to figure out the value of the other picks."
What did the research reveal? Less than 15% of fourth-round picks become starters.
In two seasons, Jones has blossomed into one of the NFL's budding stars on one of the league's most productive offenses. He earned his first Pro Bowl berth after the 2012 campaign.
Dimitroff knows that advanced metrics are hardly foolproof. Yet they can go a long way toward identifying which of the five linebacker prospects on the board has a better shot at succeeding in the Falcons system or what value to place on a veteran.
"In the end, you still have to make the call for the right reason," Dimitroff says.
COACHES VIEW: In games, backlash can outweigh odds
It's fourth-and-3, and you're at the opponent's 37-yard line.
Go for it? Try a 54-yard field goal?
According to the statistical model developed by Brian Burke, it's better to keep the offense on the field. The math suggests teams have a 56% chance of converting on fourth down in that situation vs. a 47% chance of making the field goal.
Burke, an ex-Navy pilot, has crunched the numbers to such a degree that he has computed the odds for decisions based on a wide range of situations that make up his Win Probability Model. He has written extensively about his findings since 2007 on his website, AdvancedNFLStats.com, and says he has consulted for three NFL teams.
Still, while many teams have become progressive in applying analytics in areas such as the salary cap, game management is a tougher sell.
Playing the odds might work on paper, but for coaches whose jobs can be on the line, the risks have an added dimension.
Brian Billick interviewed with the Philadelphia Eagles.(Photo: Matthew S. Gunby, AP)
Brian Billick, who won Super Bowl XXXV coaching the Baltimore Ravens and now works as an analyst for Fox Sports, recognizes the value mathematical equations can offer coaches with decision-making.
But he points out researchers don't deal with the risk. "If I go for it on fourth-and-3 from my own 20 and it doesn't work, I want that son of a (gun) standing next to me on the sideline when people are throwing things," Billick said.
Denver Broncos defensive coordinator Jack Del Rio bristles when thinking of the second-guessing he endured when game-management decisions backfired during his tenure as Jacksonville Jaguars coach.
"You may have statistical probability on your side, but if it doesn't work, you're going to get ripped for not going by the book," Del Rio said. "The book is being rewritten all the time. There's an old book that we grew up with in the '70s and '80s. I don't think that book applies in the NFL anymore."
http://www.usatoday.com/story/sports/2013/05/15/advanced-statistics-nfl/2164723/
