E-Mail
Print
The Buffalo Bills announced earlier this week they would be implementing a "robust analytics department." It's impossible to know what "robust" means in this case, but we can dream, right?
After some thought, several diet cokes and an informal Twitter poll, here's what WGR's vuluptious analytics department (me) came up with as a wish list - or maybe an outline of hope - for the Bills' new analytics deparment.
Before the list, here's what I hope the analytics department isn't: A tool for Russ Brandon to use to tell everyone (media, fans) they have no idea what they're talking about. Fans are smart. They know C.J. Spiller should be getting the ball. They know Rian Lindell can make a 52-yard field goal. They can see the truth through 20 feet of crap, so just saying "we have a stat department, so you can't argue with math" won't fool anyone. I also hope it isn't a group designed to make Brandon or his GM feel better about their decisions.
Now that that's out of the way, here's the list:
Coach and GM evaluation:
An analytics department should absolutely work with a coach and GM to assist them in decisions, but it should also deal with Brandon separately. The team's president should receive reports grading choices from punting on fourth-and-1 at the 34 to deciding to sign Shawn Merriman to paying Ryan Fitzpatrick to drafting C.J. Spiller. Their opinion on all moves should be known and respected.
The department should be constantly providing Brandon with information about his team and other coaches and GMs around the NFL. If another team is doing something sabermetrically savvy, the he should know about it.
In-game decisions:
At advancedNFLstats.com, there is a win probability calculator. It uses situations (down, score, quarter etc.) to calculate the percentage that a team increases their chances to win with each decision. The Bills should be using this or something of the like during every single game. The head coach should be transmitted information by the analytics department from the press box to assist him with decision making. Of course, whether the coach chooses to use the information will be up to him, but just because you sort of know the way doesn't mean a GPS won't help.
The in-game decisions shouldn't always be about field goals, punts and two-point conversions. It should also provide relevant information about what types of adjustments the coach should make based on success or failures of certain strategies. Say the Rams have not targetted one of their wide receivers in the first half, but in past 10 games, receiver X has the majority of his receptions in the second half, the Bills' defense should emphasize stopping receiver X. That's simplifying it to some extent, but you get the idea.
Player Evaluation and Contracts
Ryan Fitzpatrick is a great example of Buddy Nix and Chan Gailey being fooled by small sample size. They evaluated Fitzpatrick's abilities to lead a winning franchise based on seven good games in which Fitzpatrick's QB rating was around a 92 and the Bills were 5-2. But if you looked at his entire career leading up to those seven games, his career rating was in the mid-70s. What were the chances that Fitz all the sudden became a top-tier quarterback? Very small. An analytics department would have been strongly against a long-term contract.
Also consider the case of Jairus Byrd. Statisticians should be able to predict his future performances. Based on age and numbers, they should be able to judge the probability that he will continue to perform at a high level. Same goes for Andy Levitre. How well did the Bills do when running his way? When they threw screen plays behind him? How did other lineman in the NFL do by comparison? All these questions can be answered.
It isn't just about contracts, it's about game-to-game, season-to-season. You don't evaluate Mario Williams' season based on four games following his wrist surgery, you judge him based on his entire season. You don't just judge him on sacks, you evaluate him on pressures, QB hits, pass deflections, successful plays he's involved in, opponents' success rushing to his side, QBs rolling to his side and so forth. Same with a pass rusher like Kyle Moore. If he was strong statistically, it might be a good idea to keep him around.
Also, strengths and weaknesses could be evaluated easier. Which direction does C.J. Spiller run with the most success? Your eyes probably have a good idea, but the numbers will tell you for sure.
Drafting/College Scouting/Positional values
The Bills' analytics department should be able to assist scouts - much like baseball - in predicting which prospects will succeed and which will fail. Like traditional scouting, projections have flaws. However, they can help clear up a blurry picture. For example, Football Outsiders predicted Russell Wilson would have a strong season in the NFL and they were right.
Stats should also make it clear which positions have the most effect on winning. For example, AdvancedNFLstats.com calculates Aaron Rodgers, the league's best quarterback, has a much higher Win Probability Added than J.J. Watt, the NFL's best defensive lineman. The Bills should use this as an outline when considering draft picks. They should ask: Which players could we replace with a free agent instead of wasting a draft pick? For example, the Bills could have easily found an equal for Aaron Williams even if he reached his full potential compared to a quarterback, who may have impacted the franchise much more if potential was reached.
Free Agents
The Bills have hit the jackpot on undrafted free agents and under-the-radar free agents in the past. There's no better example than Fred Jackson. It doesn't take long to find many examples of Jackson-like players who weren't considered to be future stars but ended with with solid careers. A solid stats department should be able to dig deep to find players who might be candidates for becoming the next Arian Foster or Danario Alexander. It should also be able to predict
whether a player is "washed up." The New York Jets signed Thomas Jones and Ladanian Tomlinson, both of whom were said to be dusted, but turned out to be quality players for the Jets. The Bills have tried that, but have too often
misdiagnosed see: Merriman, Shawn.