YOU HEARD WHAT what Sir Charles said -- that analytics don't matter. That they don't impact the game in any meaningful way. Well, with all due respect to Chuck, the folks at MIT disagree. Here are the eight finalists -- SparkNotes-style -- for the MIT Sloan Sports Analytics Conference's research paper of the year. And for all you teacher's pets out there: You'll find the unabridged versions here. The winner will be announced at this year's conference, Feb. 27-28 in Boston. Now, get reading.
WHO IS RESPONSIBLE FOR A CALLED STRIKE?
By Joe Rosales and Scott Spratt
What you need to know: Yeah, you've heard that Brewers catcher Jonathan Lucroy is one of the majors' best pitch framers. But here's a new one: Dustin Pedroia. Says who? Says "strike zone plus/minus," a new statistic that, unlike current pitch-framing methodologies that account only for catchers' abilities, divides credit for getting more strikes and fewer balls among the four actors in play on any given pitch: the catcher, pitcher, batter and umpire. So does a catcher really excel at pitch-framing, or does he benefit from a good framing pitcher? Or hitter, for that matter? (Given his plate discipline, Pedroia alone is said to have saved the Boston Red Sox four runs last season.) Strike zone plus/minus tells the comprehensive tale.
The research says ... "Sixteen runs [saved] is worth about $11 million to $12 million in the current market ... [but] the only catchers in baseball that approach that kind of annual salary are the ones who are valuable in multiple ways, such as offensively."
The big number: 16 -- Mariners catcher Mike Zunino's "strike zone runs saved" in 2014, MLB's top mark, also recorded by the Astros' Hank Conger.
COUNTERPOINTS: ADVANCED DEFENSIVE METRICS FOR NBA BASKETBALL
By Alexander Franks, Andy Miller, Luke Bornn and Kirk Goldsberry
What you need to know: Steals. Blocks. Rebounds. They make for great highlight-reel fodder, but as proxies for defensive effectiveness they're woefully inadequate. Based on tracking data from every defensive matchup in the 2013-14 NBA season, this study introduces a suite of five groundbreaking metrics to quantify defensive performance -- a notorious shortcoming in basketball analytics -- including "volume score" (how often a defender's matchup takes the shot) and "disruption score" (how much a defender reduces his opponents' shot efficiency).
The research says ... "We stand in a muddled state where offensive ability is naturally quantified with numerous directly measured numbers, yet we attempt to explain defensive ability through statistics only loosely related to overall defensive ability."
The big number: 10.8 -- The average number of points Clippers guard Chris Paul gave up per game on average last season -- the league's best mark among backcourt defenders.
GRAPHICAL MODEL FOR BASKETBALL MATCH SIMULATION
By Min-hwan Oh, Suraj Keshri and Garud Iyengar
What you need to know: News flash: The likely outcome of any given Cavaliers game ebbs and flows with LeBron James' ability -- or inability -- to take the court. And not just James but teammates Kevin Love, Tristan Thompson and J.R. Smith. So why don't current basketball game simulations account for sensitivity to lineup changes? It's a glaring hole in today's models and one this study aims to remedy. By analyzing the NBA's player tracking data (via SportVU) for the 2013-14 season, the authors create an infrastructure that simulates ball movement for every play, based on teammates' pass interaction, how likely any given player is to shoot and shooting efficiency. In the end, you'll know the macro: Which team will win? And the micro: How will LeBron James perform?
The research says ... "[In conventional game simulation models], questions such as 'With Tim Duncan and Tony Parker out tonight, will the Spurs win against the Rockets?' still remain unanswered."
The big number: 25 -- The number of times, per second, that SportVU's software tracks the movements of the basketball and every player on the court.
ALLEVIATING COMPETITIVE IMBALANCE IN NFL SCHEDULES: AN INTEGER-PROGRAMMING APPROACH
By Mark Karwan, Murat Kurt, Niraj Kumar Pandey and Kyle Cunningham
What you need to know: Conspiracy theorists, rejoice: The NFL does play favorites. From 2003 to 2012, Arizona played just three games against teams coming off a bye. Atlanta? Eighteen. That's a lot more rested legs the Falcons faced over 11 seasons. It's a problem -- but it doesn't have to be, according to this study, which aims to eradicate the league's competitive imbalance by minimizing the number of games that any team will play against "more-rested" opponents in a season. Factoring in bye weeks, Thursday games and home and away streaks, the authors introduce a programming model that outperforms the NFL's when it comes to striking competitive balance.
The research says ... "Dissatisfied by their 2013 season schedule, the Bills raised a complaint ... stating, 'It's very difficult to call the NFL a league of parity when there's one team with half of their division games against clubs with extra time to rest and prepare, while another in the same division has none.' "
The big number: 40.6 -- The average winning percentage of teams that faced opponents with extra rest in 2013.
ASSESSING THE OFFENSIVE PRODUCTIVITY OF NHL PLAYERS USING IN-GAME WIN PROBABILITIES
By Stephen Pettigrew
What you need to know: With all due respect to the guys on the ice, the NHL is stuck in the analytics stone age. This study aims to help pry it out, introducing a win probability metric that assesses the likelihood, second-by-second, that a team wins its game, accounting for current score, penalty situation and home-ice advantage. From there, the author uses his model to manufacture a brand-new statistic, "added goal value," a quantifiable look at exactly how much a player's goal increases his team's win probability. Turns out, all goals are not created equal -- and it's AGV that unveils which players notch the most game-changing ones. Which, if you ask us (or NHL GMs, probably), is handy to know.
The research says ... "[AGV] can be incredibly useful in evaluating the skill of non-elite players. For a general manager looking to add scoring depth, these are the players to whom he may look."
The big number: 17 percent -- The average increase in win probability from a goal since the 2004-05 lockout.
"QUALITY VS. QUANTITY": IMPROVED SHOT PREDICTION IN SOCCER USING STRATEGIC FEATURES FROM SPATIOTEMPORAL DATA
By Patrick Lucey, Alina Bialkowski, Mathew Monfort, Peter Carr and Iain Matthews
What you need to know: How much of a premium are goals in English football matches? In 85 percent of all games dating to 1888 -- yes, 1888! -- neither squad scored more than three times. This study quantifies a variety of match factors, such as in-game scenarios like corner kicks, free kicks and open play, defender proximity, and shot location and speed of play to measure the likelihood of any shot finding the net. This "expected goal value" then can be used to evaluate a team's offensive and defensive efficiency. And when scoring goals is this tough, well, that just makes data like this all the more valuable.
The research says ... "In the 2014 FIFA World Cup, Germany blitzed Brazil 7-1, but Brazil actually had more shots and shots on target, 18 versus 14 and 13 versus 12, respectively. When the statistics do not tell the full story of a match, our analysis can be used to give a better indication of whether a team was 'dominant' or 'lucky.' "
The big number: 9,732 -- The number of shots analyzed using player- and ball-tracking data for this study.
DIAMONDS ON THE LINE: PROFITS THROUGH INVESTMENT GAMING
By Clayton Graham
What you need to know: How to be a strategic sports gambler, in three easy steps: 1) Devise a forecasting model that accurately predicts a baseball team's chance of winning a game; 2) Pair that probability with the betting line to determine an investment edge; and 3) Create a risk-return calculation to quantify optimal bet size. Lather, rinse, repeat.
The research says ... "Caution, just because there is a large [expected value of return on investment] or [investment edge], one must not conclude that it is a good investment. There are situations where bets appear to be too good to be true, and in fact are just that."
The big number: 1.273 -- The "runs park factor" for Coors Field (home to the Colorado Rockies) in 2013, highest -- and therefore most hitter-friendly -- in the league. This study integrates "run factor" in its game forecasting model.
EVALUATING THE EFFECTIVENESS OF DYNAMIC PRICING STRATEGIES ON MLB SINGLE-GAME TICKET REVENUE
By Joseph Xu, Peter Fader and Senthil Veeraraghavan
What you need to know: Peanuts, Cracker Jacks and dynamic ticket pricing. These are your national pastime staples. The flexible pricing plan is the preference du jour among MLB teams -- yet they have little to no guidance on how to design those plans. This study amends that, developing a single-game demand model that can improve revenue by 2.36 percent.
The research says ... "Estimates ... suggest that an extra loss for the home team has the same effect on demand as raising the ticket price by $1.09 across the entire stadium."
The big number: 21 -- The number of MLB teams (out of 30) that integrated some form of dynamic pricing strategy for single-game tickets as of 2013.