BRAIN GAMES: THE SCIENCE OF COACHING
Data is dominating sports, but assessing which coaching staffs are effective is still inexact.
Link: http://www.sportsonearth.com/article/225628760/coaching-analytics-advanced-metrics
In this era of big data, we've been able to advance our knowledge of every professional game, with advanced metrics now part of roster building in most front offices. Scouting is no longer a feel, but a science, and we have a fuller picture of which players truly rise above their peers. But, despite all this, we tend to rely on just two numbers to assess the best coaches and managers: wins and losses.
Team records, of course, do not tell the full story. And smart fans understand this. Thus, with this week marking the rare time of year when five major team sports converge in the news cycle -- the NFL Draft on Thursday, the NBA and NHL playoffs in full swing, the MLB season blossoming and the English Premier League campaign winding down -- we thought it would be a good time to audit the measures of coaching, specifically when it comes to in-game strategy.
Which sports require the most hands-on approach when it comes to managing each minute, pitch and possession? How do they compare? What are the strategies that affect the outcomes of games the most? And, of course, which bosses are the best? Let's go to the clipboard.
MLB
The number of coaches it takes to win
The team with the fewest coaches (not including the manager) on staff is the Blue Jays with nine. The teams with the most, at 13 coaches, are the Diamondbacks and Giants, with most of the league somewhere in-between. The variations are mainly a result of some teams not carrying an assistant hitting coach (Blue Jays, Phillies and Padres), and some teams carrying more than one bullpen catcher (Indians, Tigers, Astros, Angels, Rays, Diamondbacks, Mets and Giants). Given the 25-man active roster for MLB teams prior to Sept. 1, when active rosters expand to 40, the ratio of coaches to players is roughly one coach to every two players.
Most areas of coaching expertise are compartmentalized, with pitching coaches and hitting coaches handling the bulk of the instruction. Most managers will handle key decisions during a game -- when to bunt, when to call in a reliever, when to attempt a steal, when to intentionally walk a batter -- with much of the micromanaging handled by bench coaches or the players themselves (for instance, pitch selection will depend on many factors, including the rapport and experience of any given battery). A catcher like Yadier Molina -- who knows how to deal with a staff, where to position fielders, when to call pickoffs -- is basically a manager on the field.
How the schedule affects strategy
Managers have to make sure to protect players over a 162-game season, which means trying to avoid overburdening relievers and veteran bats. That will certainly affect in-game decision-making during the regular season, though all bets are off in the postseason. As we saw prominently last October with the use of relievers like the Indians' Andrew Miller and the Cubs' Aroldis Chapman, earlier and more frequent use of late-inning specialists in high-leverage situations is becoming more common.
Of the top 11 teams in MLB last season in terms of innings pitched by relievers, just one -- the Dodgers -- made the postseason. The Dodgers led all of baseball with 590 2/3 innings pitched out of the bullpen. The next four clubs, in order, were the Pirates, Reds, Twins and Braves.
Of the bottom five teams in terms of relievers' innings pitched, four reached the postseason -- the Giants, Cubs, Red Sox and Blue Jays. The only team in the bottom five that didn't make the playoffs in 2016 was the White Sox (78-84). Blue Jays relievers pitched the fewest innings of any team's relief corps last season, with 464.
A key element of this is that lower usage of the bullpen indicates starting pitchers are going deeper into games, and teams in that situation tend to have a better chance to make the postseason. But another element is the strategy involved based on pitcher-hitter matchups, past history and the grind of a 162-game season with the strain it places on bullpen arms.
There does appear to be some correlation between innings pitched and effectiveness. In 2016, the Astros' bullpen had the lowest Fielding Independent Pitching (FIP) at 3.32, and was 10th in innings pitched at 551. Meanwhile, of the top five teams in reliever FIP last season, only the Dodgers were in the top nine in terms of innings pitched.
Biggest game-changer: defensive shifts
There's no doubt that defensive shifts have proliferated in MLB during this decade: During the 2010 season, 3,323 batters faced a defensive shift, per Fangraphs. By last season, that number had soared to 34,801 (a 947 percent increase). But how does that translate into the games themselves?
"It's the analysts' process, where they take the data and they make adjustments based on what they see based on the data," Pirates general manager Neal Huntington told Sports on Earth. "Then it's our advance scouts that make adjustments based on what they're seeing in their current look. It's our coaches going through the process, and are they comfortable with where we are? Then it's an in-game adjustment with [catcher Francisco Cervelli] and the pitcher and the infielder and the outfielder, with what they're seeing.
"It's not like there's one mandate that comes out of the computer that, 'This is where you're going to stand.' It's a six-person process."
While it's evident that defensive shifts have exploded in usage, it's also clear that teams are still grappling with how often to use them, as well as what kind of shifts to deploy.
1) Traditional vs. non-traditional configurations
The Marlins and Pirates were among the top 12 teams who used non-traditional defensive shifts most frequently in 2016. A "non-traditional" shift, per Baseball Info Solutions, is one involving positioning other than the "Ted Williams shift" -- having three infielders on one side of the infield -- or "partial Ted Williams shift," where two players are significantly out of position at one time, including a second baseman in short right field.
Miami, 26th in total number of shifts (of any kind) last season, deployed a non-traditional shift against 298 batters in 2016, second only to Oakland's 299. Pittsburgh employed a non-traditional shift against 234 batters last season.
But the two clubs have seen significantly different results: opponents' batting average on balls in play (BABIP) when being shifted against by the Marlins was .289, whereas the BABIP against Pirates' non-traditional shifts was .244. Both clubs were more successful than the MLB average BABIP allowed against non-traditional shifts, which was .293.
The Marlins' utilization of non-traditional shifts jumped 35 percent from 2015 to 2016. But the opponents' success rate increased, as well -- in 2015, opponents' BABIP was just .221 against Miami in non-traditional shift situations, 68 points lower than last season. Though admittedly a small sample size, that could indicate the club is still wrestling with mastering shift strategy.
"It was one of those things we looked into this winter," said Marlins manager Don Mattingly. "As much as anything, the bottom line of it isn't about how much you shift or how much you don't, it's really about turning ground balls into outs."
The Pirates went in the opposite direction with respect to non-traditional shifts from 2015-16. In '16, they deployed non-traditional shifts against 59 fewer batters than they did the prior season and the BABIP of Pittsburgh's opponents in those situations also went down significantly, from .273 to .244. Meanwhile, the Pirates moved from 27th in the Majors with 971 traditional shifts in 2015, to fourth last season with 1,483, signaling a significant change in strategy.
"We have to be cognizant of when the shift is beat," said Huntington. "The challenge becomes remembering the several hundred other times they hit the ball right into the shift and [infielder] Josh Harrison's standing in a spot that conventional wisdom wouldn't have him standing in, and he makes a play and he makes it look really easy."
Other teams have also drastically increased the number of traditional shifts they utilize. One of these clubs is the Cardinals, whose use of traditional shifts jumped 161 percent from 2015 to '16, from 312 instances to 813. In 2015, opponents hit .295 on balls in play against the shift, and last season, .296 (MLB average in 2015 was .290, and last season it was .298).
"Some of it has to do with the more information we get," Cardinals manager Mike Matheny said. "Getting buy-in from our pitchers, too, and showing them and continuing to educate them on, 'Here's what we see. Here's what the information tells us. How comfortable do you feel?' If the pitchers don't feel comfortable with it, we're going to shy away. When they do, we're going to take advantage of it."
Early on, some Cardinals pitchers were hesitant when it came to defensive shifts, and as Matheny said, pitcher buy-in with this process is paramount. As that increased, so has the number of Cardinals defensive shifts. The number of non-traditional shifts St. Louis used also increased moderately from 2015 -- from 240 to 276 -- but, notably, the BABIP shot up from .204 to .348.
2) Quality, not quantity
The World Series champion Cubs were, by far, the most successful team in the Majors when it came to deploying defensive shifts in 2016. Opponents' BABIP against shifts (both traditional and non-traditional) put on by Chicago was .239; the American League champion Indians were second, at .272.
The Cubs also shifted against the fewest batters in all of baseball last season: 603 (299 of which were non-traditional shifts).
Bench coach Dave Martinez handles the bulk of the defensive positioning duties and said that the Cubs will continue doing what they did last year. Martinez also echoed what the Rockies' Cole said about player input.
"In-game, you see how guys are swinging the bats, so we might move them, and it depends on the wind here [at Wrigley Field]" Martinez said. "You might move them a few more steps back, or cover the lines more. For the most part, those guys do a great job. Last year, we followed [what they suggested] and it worked."
The Cubs, under the guidance of manager Joe Maddon, are also known for unorthodox defensive strategy, such as moving a pitcher into the outfield or having first baseman Anthony Rizzo crash in hard on an opposing hitter's bunt attempt while the second baseman covers first base. The process has even involved Rizzo trading his first baseman's glove for a middle-infielder's glove.
"It's starting to catch on," Martinez said. "It won't be long before everybody does what we do with the bunt plays where we have Rizzo [crash in towards the plate]."
As teams around baseball continue to wrestle with strategy on defensive shifts, working to refine the process and achieve better results, executives, managers and coaches alike appear to share one common kernel of wisdom behind all of those efforts.
"This is still a game played by human beings, and there are times the shift will not work," Huntington said. "The reason you shift is the probability tells you this is where he's going to hit the ball, but occasionally, it does work against you.
"That's the challenge: Do you feel better when a play's not made conventionally, because, well, that's just where you're supposed to play? Or do you feel better when you make a play that nobody realizes you made?"
No matter what, the utilization of shifts is not only one of the biggest in-game management strategies that affects outcomes, but one that we can assess with raw data to determine success or failure.
-- By Manny Randhawa
NBA
The number of coaches it takes to win
Roster size vs. number of coaches is approximately 3-to-1 or 2-to-1. Most NBA coaching staffs comprise of the head coach and 5-7 assistants in various roles. For example, the Spurs have employed Gregg Popovich as their head coach for 20 full seasons (plus an interim stint during the 1996-97 season). This season, their coaching staff included six assistant coaches (Ettore Messina, Ime Udoka, James Borrego, Chip Engelland, Becky Hammon and Will Hardy). The two teams who have met in the NBA Finals the previous two seasons -- the Cavaliers and Warriors -- have similar-sized staffs as well.
The Warriors have Steve Kerr as their head coach along with three assistant coaches (Ron Adams, Mike Brown and Jarron Collins) and three assistant coaches for player development (Bruce Fraser, Willie Green and Chris DeMarco). That coaching depth will be tested soon, as lingering back issues could force Kerr to miss the entirety of the playoffs.
The Cavaliers are led by Tyronn Lue, who has five assistant coaches (Jim Boylan, Larry Drew, James Posey, Phil Handy and Mike Longabardi) on his staff. Teams often employ player development coaches who are not listed on the official coaching staff. For example, Steve Nash is a player development consultant for the Warriors and has worked with extensively with Kevin Durant and other players on the roster. Tim Duncan has worked with the Spurs this season. His title is less defined. Popovich once jokingly called him the "coach of whatever he feels like."
How the schedule affects strategy
The league modified its schedule this season to scale back situations where teams play four games in five nights, but teams are still weary about how much their star players play throughout the season. Here's a look at all 30 teams' records this season without a day's rest between games (i.e. the second night of a back-to-back). The top records belong to the top three teams in the West: Houston (13-3), Golden State (13-4) and San Antonio (12-4). The worst teams are three teams that missed the playoffs this season: Dallas (3-13), Detroit (3-14) and Brooklyn (1-14). Another impact of scheduling is that one would think older rosters might be more susceptible to losses on the second half of back-to-backs, but that doesn't appear to be the case either. Per RealGM, the average age of Houston's roster is 26.4, the Warriors have an average age of 28.1, the Spurs 29.1. All of them were elite teams without any rest this season. Scheduling doesn't affect a team's performance so much as simply having a roster full of talented players.
Biggest game-changers: Shifting lineups, out of TO possessions
1) Managing minutes
One of the ways for coaches to impact outcomes of games is to maximize optimal five-man lineups and manage their rosters to ensure that there is at least one superstar on the floor at all times. Per NBA.com, among lineups that played at least 300 minutes, the best five-man lineup in terms of net rating in the league belonged to the Warriors' unit of Stephen Curry, Klay Thompson, Kevin Durant, Draymond Green and Zaza Pachulia, who outscored opponents by 23.1 points per 100 possessions. Kerr's task in Golden State might sound simple, but coaches with multiple superstars have to ensure that their superstars are on the floor at all times. The Warriors are a good example for this. This season, Curry played 2,638 total minutes, Thompson played 2,649 minutes and Durant logged 2,070 minutes. Durant missed the most time with injury, so if we look at Curry and Thompson's minutes, they spent 1,773 minutes on the floor together, which means approximately 67 percent of the time they played, the two were on the floor together. The Warriors outscored opponents by 19.0 points per 100 possessions when Curry and Thompson were on the court together. For NBA coaches, success in the regular season is tied to how well they can keep their stars on the floor, and find the best five-man lineups and use them with frequency.
In-game and in-series adjustments during the playoffs is where head coaches are most scrutinized, though. Kerr established himself as a tactician in the 2014-15 playoffs when the Warriors were down 2-1 to the Grizzlies, and he started Game 4 by having starting center Andrew Bogut defend Tony Allen -- a below-average shooter on the perimeter -- which allowed Bogut to act as a roamer on defense, and ignited the team to three straight wins to take the series. In the NBA Finals against the Cavs, the Warriors were down 2-1 once again when Kerr removed Bogut from the starting lineup and went with a smaller starting five with Andre Iguodala. The move also triggered three straight wins for the Warriors and the championship that year. (You could argue that a lack of adjustment cost Golden State the title last year, when LeBron James and Kyrie Irving were eating the Warriors alive in Games 5-7).
2) Out-of-timeout possessions
Teams this season averaged between 93 to 103 offensive possessions per game. Even if a coach can positively contribute to, say, three offensive possessions per game out of timeouts, that's 246 offensive possessions over the course of 82 games. Synergy Sports provided some data on each team's points-per-timeout last season. Kevin McHale, who was coach of the Rockets at the time, was bottom of the list at 0.50 points per possession. At the top was Alvin Gentry of the Pelicans, averaging 0.976 points per possession. Using those metrics as an example, over a course of an entire season, assuming the 246 offensive possessions out of timeouts, McHale's teams would have 123 points over those possessions, compared to 240 points for Gentry. The spread between those numbers would have an incremental impact on a team's win total.
-- By Alex Wong
NFL
The number of coaches it takes to win
The ratio of players to coaches can be anywhere from 2-to-1 to 3-to-1. The Patriots have a staff of 18, but the gameday coaching falls heaviest on the shoulders of Bill Belichick, offensive coordinator Josh McDaniels, defensive coordinator Matt Patricia and special teams coach Joe Judge. Not that assistants don't have their own influence after kickoff, but with an active gameday roster of 46 and three or four coaches doing the heaviest lifting, then maybe the "real" ratio is closer to 7.5-to-1, or 15-to-1. In the broadest terms, it's one head coach and 46 active players.
Not that 18 is the standard-sized staff. The Seahawks have 26 coaches. Pete Carroll even has two assistant head coaches: Tom Cable and the recently promoted Michael Barrow. There are more coaches in Seattle to bear the load, but perhaps the single-minded view of the Pats also adds to their successful model; it's just hard to argue that one way is good and one way is bad because Belichick and Carroll are two of the most successful coaches since 2012.
The smallest current staff belongs to the Steelers, where there are 17 coaches, led by Mike Tomlin. The largest staff is under another veteran head coach, Andy Reid, who has a staff of 29 in Kansas City. Since taking over a 2-14 team four seasons ago, Reid has gone 43-21 with three playoff appearances. Over those same four seasons, Tomlin is 40-24, also with three postseason trips.
Head coaches who do play-calling on game day:
Ben McAdoo, Giants (offense)
Mike McCarthy, Packers (offense)
Sean Payton, Saints (offense)
Bruce Arians, Cardinals (offense)
Hue Jackson, Browns (offense)
Andy Reid, Chiefs (offense)
Sean McVay, Rams (offense)
Adam Gase, Dolphins (offense)
Mike Zimmer, Vikings (defense)
Kyle Shanahan, 49ers (offense)
McAdoo was promoted by the Giants from offensive coordinator to head coach and understandably didn't want to hand over the job of calling plays to someone else. This is despite the fact that New York, like most teams on this list, has someone with the title of offensive coordinator: Mike Sullivan. The Giants were 26th in scoring and 25th in total yards last season and McAdoo won't say if he'll give up play-calling duties in 2017.
Of course, the most important play-caller on any NFL team might be the quarterback. As Belichick said this March, the biggest thing Tom Brady does for the Patriots is getting them out of a bad play call once he recognizes the defensive alignments before the snap. Like Molina on the St. Louis Cardinals, he's the field manager.
How the schedule affects strategy
The NFL has by far the fewest games of any professional American sport. Because of the number of games and preparation that can be as short as four days and up to two weeks between games, coaching becomes quite concentrated and preparation is key. Coaches can lay out offensive "blueprints" to open every game based on their research and film study leading up to the game, but even those may last two-to-three series at most. Now consider what a "halftime adjustment" basically is: It's that one week of film study and preparation boiled down to basically trying to accomplish the same task in 15 minutes.
A study by Pro Football Focus in 2013 showed that scoring went down on Thursday Night Football 0.68 touchdowns per game on average, as teams often had just three days to prepare and the play has always seemed noticeably sloppy. A 2014 study by Timothy Vaughan showed the wide variance of outcomes on TNF, noting that the first five Thursday games of that season had ended in a blowout; however, six of the next 12 TNF games were decided by six points or less. Home teams also win at a higher rate on Thursday than they do on all other days, possibly due to the quick turnaround for travel.
Pete Carroll is 4-1 on Thursday, with all four wins coming by double digits. The Ravens are 6-1 on Thursday under John Harbaugh. Meanwhile, Andy Reid is an incredible 16-2 (.888 winning percentage) following a bye week. That's better than Belichick's 12-5 record following a bye, including a 31-24 home loss to the Seahawks last season.
Biggest game-changers: clock, stops and rushes
Football has to be the most coach-designed major American sport, and for the players it mostly comes down to execution. That's why the NFL Scouting Combine matters so much to these coaches and the draft is so important. Sometimes it's just about finding the players who are physically capable rather than the ones who succeeded in systems that don't translate to the NFL. Coaches don't care if a receiver is gifted at making up a route; coaches care if he can run the exact route they game-planned, can memorize the plays and execute them to perfection. Compare that to basketball where freelancing and improvisation in a fast break is one of the most valuable skills to possess.
In 2016, teams ran an average of 57.4 offensive plays per game on the low end (Dolphins) to 69.1 (Saints) on the high end. New Orleans played an NFL-high 2,112 offensive and defensive plays last season, while the Lions ran an NFL-low 1,944 plays.
In general, the gap does not seem that wide, but in those 10-11 extra plays per game, a lot could happen. Here's what coaches can control:
1) Clock management
Perhaps every coach has been criticized at some point for clock management, but one head coach who took it very seriously was Tampa Bay's Dirk Koetter. After the Bucs promoted Koetter in 2016, he gave offensive assistant Andrew Weidinger a new position: "game management." Weidinger's job is to manage the clock and prepare for late-game situations and challenges. Tampa Bay improved from 26th in time of possession in 2015 to eight in 2016. That's an extra two minutes of possession per game.
Four of the top five teams in TOP made the playoffs, while the bottom three -- 49ers, Browns, Bears -- are also the top three teams picking in the 2017 NFL Draft. The Eagles went from last in TOP in 2015 under Chip Kelly (25:51) to first in 2016 under Doug Pederson (32:31), giving them almost six-and-a-half extra minutes per game, though they actually scored 10 fewer total points under Pederson for the season.
2) Third-down stops
Perhaps no stat is more widely attributed to play-calling than third-down conversions and stops because of how much it impacts the outcome of a game.
Last season, the Saints were first in the NFL in third- and fourth-down conversion rates: 48.6 percent on third and 13 of 15 on fourth. Rounding out the top five offenses were the Packers, Titans, Patriot and Redskins. The bottom five in third-down conversions, starting with 32 and working upward: Rams, Broncos, 49ers, Jaguars and Giants.
The top five defenses in stops on third down: the Bucs, Ravens, Giants, Dolphins and Broncos. The most surprising is that Tampa Bay finished first in that category (34.4 percent) despite finishing 15th in points allowed and 23rd in yards allowed. Defensive coordinator Mike Smith, former head coach of the Falcons, did draw some interviews and interest in January for head coaching positions with other teams.
The worst, starting with 32, were: Washington, the Lions, Browns, 49ers and Saints.
So which coaching staffs were the best on both sides of the ball? Tampa Bay, Tennessee and New England were the only teams in the NFL last season to finish in the top seven in both offense and defense for percentage of third-down conversions and stops. Of course, the Pats won a title. Keep an eye on the Bucs and Titans in 2017. Cleveland and San Francisco were the only teams to finish in the cellar in both rankings.
3) Going all-in
Another thing that coaching can often be accountable for is how much pressure is being applied to the quarterback. This is because defensive alignments and blitz packages are often called out by coaches, though defensive leaders can still make adjustments if they notice a gap in the offensive alignment. Per Football Outsiders, the Broncos led the NFL in pressure in 2016 at 31.8 percent. Denver finished first in passing yards allowed, passing touchdowns allowed and net yards per pass attempt, and sacked the quarterback 42 times. Rounding out the top five were the Cardinals, Eagles, Vikings and Dolphins.
The worst team in the NFL at pressuring the QB in 2016 was the Colts, who did so on just 18 percent of drop-backs. They have somehow decided to hold steady with head coach Chuck Pagano and defensive coordinator Ted Monachino anyway.
-- By Kenneth Arthur
NHL
The number of coaches it takes to win
Only a fraction of a team's coaching staff is stationed on a bench during a game. Typically, a head coach will have two assistants on the bench with him, but that's not set in stone. Blue Jackets coach John Tortorella, for instance, prefers to have just one assistant on the bench with him (as an uninformed reporter learned the hard way some years back). But a team's overall coaching staff is much larger than that. The Rangers' staff -- in addition to head coach Alain Vigneault -- has an associate coach, two assistants (one in the press box who relays in-game observations during intermissions), a goaltending coach, a video coach, a video assistant, a skills coach and a skills consultant.
A head coach may have to make any number of decisions during the game: choosing whether to challenge a goal (with the assistance of his video team), or whether to call a timeout after an icing when his team is gassed, or when to pull his goalie, or which players to use in a shootout. There's also work to be done between periods, from speaking to his team to watching video. And, of course, coaches communicate with their team throughout the game, offering criticism, or encouragement, or instructions, or whatever else is needed.
How the schedule affects strategy
A coach may consider a team's opponent when deciding on the last couple of spots in the lineup for a given game. Against a physical opponent, for instance, he may opt for a little more so-called sandpaper to better match up. And games on back-to-back nights may mean tapping the team's starting goalie for one game and the backup goalie for the other. This season has also brought unique challenges, because of a schedule condensed by the preseason World Cup of Hockey and the new mandatory bye weeks. Tortorella killed off the Blue Jackets' morning skate on game days (something he'd wanted to do in previous jobs), while per the Toronto Star, other teams are mixing in more optional practices or telling players who log big minutes to stay away from the rink as much as possible, with some teams making the morning skate optional.
Biggest game-changer: bench management
This is a deceptively complex aspect of the game that only looks routine and machine-like because of the skill and preparation of all involved. In a nutshell, bench management involves not just setting the lines, but adjusting them when needed and deploying them at the right times.
There's considerable strategy involved: A coach may want to get certain defensemen on the ice when an opponent's most dangerous scorer is on the ice. Similarly, a coach might want to protect his own big scorer from shifts against an opponent's best defenders. (The visiting team's coach must decide first who to put on the ice prior to a face-off, but once the puck drops, lines can be changed on the fly.) Double-shifting a star might throw off an opponent looking to match up lines, but it also complicates things on the other end. And a coach may instead want to dictate the tone or pace by choosing a certain combination of players. Which is to say, there's a complex calculus involved in bench management.
There's also the matter of making sure the right players get the right amount of ice time. In a 2007 interview, then-Flames coach Mike Keenan spoke about Scott Bowman, a firm believer that ice time should be based on merit:
"Bowman was the master coach. What I learned from him was, and it may sound simple, it's essential to have the right people on the ice at the right time. Ask someone how long a hockey game is, and most everyone will say 60 minutes. But from a coaching perspective, it's 720 minutes -- that's six players, or positions, on each side, and two teams over 60 minutes. What you do as a coach, how you [apportion] those minutes, is essential."
"You try to get your best offensive player away from a particular matchup, even for one shift," said Eddie Olczyk, an NBC analyst and former NHL player and coach, in 2015. "One shift can change everything."
-- By Joe DeLessio