Article I wrote for my HS Newspaper on Stats in Sports

[WizSkinsO's]

Just wanted to share an article I recently wrote for my high-school student newspaper on statistics in sports. It may interest Skins' fans a bit as Redskins Offensive Assistant Chris Meidt is a source in the football section of the article. He's a very nice guy, spoke to me on the phone for ten minutes, and I really appreciated his time.

Other sources I'm expecting to add to this article include Bill James, Keith Woolner, and Football Outsiders.

Any thoughts/edits would be much appreciated. It's part of a series I've been writing called "What's Wrong with Sports?", taking an in-depth look at problems either inside the world of sports or by casual fans out of it.

WWWS: You Know Nothing.

Box at top: You may think you're a big sports fan. You watch all the games, play fantasy football, and can name the 1969 Mets' roster without batting an eye. In truth, as these respected sources in professional sports will tell you, you know nothing. Perpetrated by irresponsible reporting by some of the mainstream media, you've grown up thinking strikeouts in baseball are the worst thing in the world and points per game are a great indicator of the top basketball players. You're not the only ones misinterpreting statistics. A number of coaches, executives, and especially media pundits in these sports aren't doing what they should be. This is WWWS #4.

Start article:

Recent decades have seen tremendous improvements across sports. Players are bigger, faster, and stronger. Players are better-trained, and are returning from previously-debilitating injuries quicker than ever before, thanks to advances in sports medicine and nutrition. But one facet of the sports industry has been relatively static: the knowledge of casual fans. The truth is that you, the fan, actually knows very little. Is it your fault? Not always. Media pundits like Joe Morgan, Tim McCarver, and Stephen A. Smith certainly aren’t helping the problem. But as huge fans of a sport or team, shouldn’t you make it your goal to learn for yourself and not believe what everyone tells you? Here, we’ll tackle some of the myths, held by both coaches and fans alike, and explain why just about everything you think you know about sports is wrong. No hard feelings though.

Rounding the Bases with Sabermetrics

Paul DePodesta is currently the Executive Vice President of the San Diego Padres. Previously, he had served as the General Manager of the L.A. Dodgers, and before that the Assistant GM to legendary Billy Beane in Oakland, during which time he became a main character in the best-selling book “Moneyball” by Michael Lewis.

Moneyball is about how the low-budget Athletics, with a payroll as low as $30 million while the Yankees weighed in at over $200 million, were able to win about as many games as them over a long span. How did they do it? By realizing that everything about baseball—from players to managers to other executives—was lagging way behind where it should have been with progress. Teams were still relying on stats like batting average, RBI, and saves to determine which players to sign.

The A’s, meanwhile, had DePodesta, a cum laude graduate of Harvard University who had never played professional baseball, plugging away numbers constantly as Beane’s Assistant GM. He found a number of flaws with the conventional wisdom of the day—most notably, that on-base-percentage, which includes walks, was a far better indicator of success than batting average, which does not. Walks, he realized, were a big part of individual ability as a hitter.

For the first time, the term ‘sabermetrics’ became well-known in baseball. Sabermetrics is the analysis of baseball through objective research, mainly statistics. While average, homers, RBI, wins, ERA, and saves may have been commonplace in the past, sabermetricians use such numbers as EQA, VORP, DIPS, and NZR.

Don’t be confused by the weird-looking acronyms, though. Sabermetric stats are, quite simply, better. Check for yourself. Look at team batting average and how well it correlates with the teams that score the most runs. Then look at on-base-percentage. OBP is far-and-away more important; batting average, in ignoring walks, doesn’t tell nearly the whole story, which DePodesta helped originally find out when he joined the A’s and was able to parlay into a job as the fourth-youngest GM in baseball history at 31.

Common Sense spoke with DePodesta to get some insight into his stat-oriented mind.

Why is on-base-percentage better than average, we asked? “Outs are the time clock of baseball, so you need to do your best to avoid them,” he says. “Think of this extreme case – which team will score more runs: A) a team that hits .400 with no walks (four hits and six outs for every ten plate appearances) or a team that hits .200 but has a .700 on-base percentage (two hits, five walks, and just three outs for every ten plate appearances)?”

Batting average isn’t the only stat under scrutiny from brilliant minds like DePodesta’s. Runs batted in (RBI) “depends on opportunities, which vary dramatically due to teammates and lineup composition.” Wins for a pitcher also depend greatly on their team—a pitcher who pitches 8 shutout innings could get a no-decision if his offense can’t score, while another could go five innings, give up six runs, and get a win because of a potent offense. “We score each pitcher on their outing regardless of the overall outcome,” he says. Saves are flawed because “sometimes the most critical situation comes before the 9th inning,” and saves are relatively easy to build up, even for poor pitchers, because of relatively easy requirements—a pitcher who gives up two runs with a 3-run-lead would still earn a save, despite an ERA of 18 for that inning.

Tom Tango is a similar outside-the-box thinker, and a leading sabermetrician. Recently, he was hired as a consultant for the Seattle Mariners.

“Our job is to separate the signal from the noise,” says Tango. “RBIs are clouded with noise (lineup spot, teammates being on base). W/L record for a pitcher has similar noise (if the team doesn’t hit for you, you can’t win as much). Batting average excludes walks, and treats a single as much as a HR. That may make sense to somebody, but I have never met the person that can justify its existence when asked.”

Moving the Chains with Numbers

Chris Meidt just finished his first year as Offensive Assistant Coach for the Washington Redskins. Prior to that, he was the Head Coach at Division-III St. Olaf College in Minnesota for six seasons, when they went 40-20. In 2007, his offense’s 49.2 points per game ranked third in the nation. With the Redskins, he spends most of his time coaching quarterback Jason Campbell.

He also is in charge of everything numbers-related. As a college coach, Chris Meidt was a big proponent of surprise onside kicks, going for it on fourth downs, and going for two-point conversions.

The NFL is a different beast, obviously. Only about 20% of onside kicks are successful, and the more frequently you use them the less likely you can catch the other team off guard.

So Meidt’s primary objective during his rookie season as a coach was to listen and learn. Still, that didn’t stop him from looking at the numbers a little bit.

“I love numbers,” he says. “I’m a huge numbers guy.”

One thing Meidt studied in particular was the ‘ZEUS’ system. ZEUS is a computer program that models and predicts outcomes of coaching decisions in a given NFL game. It is capable of simulating 1,000,000 NFL games in only a few seconds, and is based on extensive research of historical data from NFL games. Primarily, it evaluates coaches based on what it calls ‘binary’ decisions involving just a couple options, such as going for it on a 4th down, trying a field goal, or punting. Meidt says the Redskins were ranked 6th in ZEUS’ ratings in terms of good decision making.

Meidt thinks average fan knowledge of the NFL varies greatly. There are some who watch film and can really break everything down, (“I don’t know what they do for a job,” he jokes,) and others that “the only football they know is what they played in Pop Warner and what they watch on Sunday.”

Meidt’s career as a coach will invariably be linked to the numbers. Following an extremely successful career as a high school quarterback and the son of a Minnesota Hall of Fame coach himself, Meidt went on to earn a bachelor’s degree in mathematics from Bethel College, and then a MBA at the Carlson School of Management.

“My background is really the variable data. Trying to find trends, trying to find some type of system that can help you improve,” Meidt says. In fact, at the time of the interview, Meidt was doing just that for the Redskins. The interview had to be limited to 5-10 minutes because the Redskins were in 8 a.m. to 8 p.m. meetings in which Meidt primarily presided over the numbers.

So what’s the stat in football that, like batting average in baseball, is used way too much?

“Quarterback rating,” Meidt says. “It puts so much emphasis on throwing touchdowns that, if your team decides to run the ball in instead, your QB rating takes a real big hit.”

For one thing, does any football fan actually know what QB rating is? The formula is so abstract and unwieldy that no-one can place any real value on it. Check out the computation:

There are four different sub-components. A is: (((completions/attempts)*100)-30))/20. B is: ((TDs/attempts)*100)/5. C is: (9.5-((INTs/attempts)*100))/4. D is: ((Yards/attempts)-3)/4. Each of A, B, C, and D must be within 0 and 2.375. Now that you have those four numbers, you can find the QB rating. Add up A, B, C, and D, and divide by .06 (obviously).

Now, you too can evaluate quarterbacks!

There isn’t a perfect cut-and-dry way to evaluate QBs, but it certainly isn’t QB rating.

Consider this: AFC Divisional Playoffs. Baltimore vs. Tennessee. Kerry Collins completes a check-down pass to tight end Alge Crumpler. Crumpler rumbles towards the endzone, but coughs up the football after absorbing a big hit. The Ravens recover on the 1 yard line. Collins finished the game with a QB rating of 71.6. Had Crumpler made a move and gotten into the end zone, it would have been 79.6. A difference of eight points in QB rating, determined by an event (Crumpler fumbling) that had nothing to do with the quarterback. An eight point swing, on just one play. This season, eight points in QB rating was the separation between Seneca Wallace and Peyton Manning. Convinced?

Kevin Kelley is a similar thinker to Meidt. He’s the Head Coach at Pulaski Academy, a high school in Arkansas.

Kelley has gone against conventional wisdom to a striking extent—he never punts. Even inside his own ten-yard-line.

The numbers, though, back up what he’s doing. With a fourth-and-eight at his own five-yard-line, Kelley says his team’s offense is strong enough to convert at least 50% of the time. If they fail to convert, the other team will score a touchdown 90% of the time. If they chose to punt, say, with a net of 30 yards, the other team would start on the 38-yard-line and have a 77% chance at a touchdown—only a 13 percent swing.

Kelley’s influences sounds like something out of an economics textbook. He has read several studies, including “Do Firms Maximize? Evidence from Pro Football” from University of California-Berkeley economics professor David Romer. He has studied ZEUS, the same program Meidt relies on, that was developed by an astrophysicist and game theory expert.

“I like numbers and believe in statistics,” says Kelley. “Overall, it was my desire to win and to find things that would help us win, even if that meant out of the norm.”

In addition to never punting, Kelley’s team uses an onside kick about 75 percent of the time. His reasoning? The difference between an onside and standard kickoff recovered by the other team is only about 14 yards.

“I came to the perception based on my knowledge that it would help our percentage chances of winning a football game. THE BEST WAY TO DO SOMETHING ISN’T NECESSARILY THE WAY IT HAS ALWAYS BEEN DONE,” Kelley shouts for emphasis. “To implement the strategy, you have to convince your coaches and your team that it is the best way to help win and that although there will be times it doesn’t work, in the long run it will pay off.”

Kelley has the success to support his ideas, too. This season, Pulaski won the 5A state title. After losing their first game, they won thirteen straight. Part of Kelley’s success may have just been how much his outlandish strategies annoyed other coaches.

“Opposing coaches hate it. It makes preparing for us far different than anyone else. Each down is less predictable. There is a psychological factor when we do convert, too.”

Swoosh with Statistics

Kevin Pelton is a writer at the leading basketball statistical analysis site, basketballprospectus.com. Among others, he has written for 82games.com, Hoopsworld.com and SI.com. Pelton is one of the leaders of the basketball version of baseball’s sabermetric movement, a movement from hunches and guesswork to reliance on advanced statistical data.

While basketball sabermetrics is still in its infancy, many important tenets have already been formed. One of the most basic ideas is the influence of possessions on statistics. Cursory metrics like points per game on both individual as well as team levels don’t tell nearly the whole story.

Don’t take our word for it, though. Here’s an excerpt from one of the first articles published by Basketball Prospectus:

Like no other coach, (Joe) Scott, the head man for the Air Force Academy, insists that his offense use as much of the shot clock as possible. For the past four seasons, a Joe Scott-coached team, either at Air Force or at Princeton, has been the slowest-paced team in the nation…Last season's Princeton team led the nation in fewest points allowed per game at 53.3. In fact, over the past five seasons, a Scott-led team has finished first in this category three times, and second in the other two years.

So is Joe Scott some sort of defensive genius? Not really…By virtue of spending so much time with the ball, Scott's teams give opponents fewer opportunities to score. Last season, Princeton averaged about 53 possessions per game, essentially 53 opportunities to score.

Each team's number of possessions [is] essentially the same in each game. This is important because it means that it really doesn't matter how many opportunities a team gets, its opponents will get the same number. So while Princeton was able to keep their opponents off the scoreboard, with only about 53 opportunities to score each game, the Tigers also kept themselves from putting up a lot of points. They ranked dead last in the nation in points scored per game. So you don't learn much about Princeton after having examined the NCAA's scoring offense and scoring defense statistics, other than that the Tigers were involved in a lot of low-scoring contests. The games weren't low scoring because of Princeton's ability; they were low scoring because of Princeton's style.

By the same logic, more possessions yield more chances to rack up gaudy stats, but does that really indicate that the player is playing better?

A great real world example is two-time MVP Steve Nash of the Phoenix Suns. Last year, under Mike D’Antoni frenetically-paced offense, Nash averaged 17 points and 11 assists per game. This year—with D’Antoni taking his offense to the Knicks—Nash is slightly over 14 points and nine assists per game. Did Nash’s skills erode over the summer? It’s possible aging had a bit to do with it, but more than likely, his stats are down simply because the Suns offense has fewer possessions per game.

“Sometimes fans (and all of us) can get seduced by impressive-looking traditional statistics without considering their context,” Pelton says. “For example, a player scoring 20 points per game sounds good, but if it's a starter logging 35 minutes a night on a fast-paced team like the Knicks and using a high percentage of the team's possessions, it's certainly possible the player may not be anywhere near as productive as a typical 20-point scorer.”

So how do we neutralize the influence of possessions? The answer is simple on the team level. Instead of points per game, the real measure of an offense is points per possession. This is usually given as a number per 100 possessions and called offensive efficiency.

For example, last year’s Washington Wizards’ offensive efficiency was 105.7. They scored 105.7 points per every 100 possessions. On the other side of the ball, they allowed 106.0 points per possession on defense. Now, if you subtract these two numbers, you come up with efficiency differential. For the Wizards, their efficiency differential was -.35, suggesting that they were a very slightly below average team. Not surprisingly, the Wizards won a middle-of-the-pack 43 of 82 games in the weaker Eastern Conference.

The correlation between efficiency differential and success is undeniable. Who was the best in the league if efficiency differential last year? None other than the NBA champion Boston Celtics. Second best was the Detroit Pistons (Eastern Conference finalist) and third best was the Los Angeles Lakers (Western Conference champs).

On the player level, the critical first step is to change per-game stats to per-minute stats. From ESPN basketball analyst John Hollinger's Pro Basketball Forecast:

"It's a pretty simple concept, but one that has largely escaped most NBA front offices: The idea that what a player does on a per-minute basis is far more important than his per-game stats. The latter tend to be influenced more by playing time than by quality of play, yet remain the most common metric of player performance."

This is an idea that Pelton understood from an early age.

“When I was a kid I used my basketball cards to calculate per-minute statistics and rank players,” he says.

Thanks to people like Pelton, stats are beginning to evolve. Three stats in particular, when used in harmony, can evaluate a player’s performance.

“I think the three key stats are True Shooting Percentage, usage rate and rebound percentage,” Pelton says. “The combination of True Shooting Percentage, measuring how efficiently players use possessions, and usage, measuring the size of their role in their team's offense, tells you much more about a player's effectiveness on offense than any combination of traditional statistics. Rebound percentage puts players on teams with varying opportunities for rebounds on an equal playing field, improving upon rebounds per game.”

Pelton believes that the mainstream media is hit-or-miss in terms of educating fans.

“In general, I think they do a pretty good job,” Pelton says. “There are some terrific color analysts who really teach the game, including Hubie Brown and Jeff Van Gundy. I'm also partial to Doug Collins. One of the easiest ways to learn the game is to pay attention to these titans of coaching.”

This contributes to the knowledge of traditional fans.

“I think the average fan is pretty informed these days. The number of games on TV and the availability of information on the Internet make it easy to get a broad understanding of the game,” Pelton says.

There is still a long way to go, however.

“The downside is when these sources offer the game's traditional axioms that can be misleading, along the lines of defense winning championships, the importance of clutch play and the tendency to blame a star player for his team's failings.”

“[some color commentators] tend to lean on stereotypes and clichés, like criticizing the defense of any fast-paced team with little regard for how well they defend on a per-possession basis. Over-hyping high scorers is something of a two-way street. In part, the networks are responsible, but they're also responding to fans' interest in these players. The tradition of marketing a game as Player X and his team vs. Player Y and his team may frustrate purists and coaches, but fans respond to it.”

All in all, Pelton believes that people who want the new information can find it, although it has not quite become mainstream.

“I think the information is out there if you look for it and tune out the empty hype. ESPN in various different formats as well as Sports Illustrated have made a bigger priority of including advanced stats as part of the stories they tell to offer additional perspective, and ESPN.com gives prominent placement to John Hollinger's work.”

Behind the scenes, basketball front offices are certainly utilizing the new data.

“We have yet to see a lot of GMs who have staked their reputation to analytics in the same way the Moneyball GMs have in baseball, but there is a trend similar to baseball of a new breed of general managers who are comfortable with both scouting and statistical analysis and want to make sure both are incorporated into their decision-making process. Houston's Daryl Morey is the leader in this regard and the Rockets use statistical analysis more extensively than any other team. I would also include Oklahoma City's Sam Presti and Portland's Kevin Pritchard. Most teams have at least one person on staff or as a consultant offering statistical analysis by this point.”

[MattFancy]

thats awesome! i am a numbers guy myself and love messing around with numbers. thats a damn good article for a hs newspaper! good work!

[WizSkinsO's]

thats awesome! i am a numbers guy myself and love messing around with numbers. thats a damn good article for a hs newspaper! good work!

Thanks very much. I don't really expect anyone in my HS to actually read it, haha, but I appreciate people like you reading it and letting me know what you think.

[Bang]

That's an excellent piece. You should maybe see if one of the local newspapers would have interest in it.

Very well done.

~Bang

[WizSkinsO's]

That's an excellent piece. You should maybe see if one of the local newspapers would have interest in it.

Very well done.

~Bang

Thanks, Bang. After I actually finish the thing, I wonder if I can convince the Post to give me a full page for my statistical ramblings lol.

One question actually along this note--does anyone know anything about hockey? We were going to include that but then realized we don't know anything about statistical analysis in hockey. Does it exist? Does George McPhee use it by any chance? Didn't mean to leave out a sport but couldn't find anything about that.

Thanks again.