LeBron vs. MJ: A Statistics-Based Analysis

LeBron James has now scored more points in the postseason than any player in NBA history. Because he just passed Michael Jordan to get the record, and because the internet loves debating Michael vs. LeBron nearly as much as it loves cat pictures, it seems as if the debate about who’s the better player is raging more than ever. As usual, the Jordan fans are yelling “6 ringzz!” from the rooftops while LeBron fans claim that LeBron is the better all-around player. As is my custom, I went into my sports nerd cave (basketball-reference.com) to find the truth. What I found was shocking! (Ok, not really, but I did think it was very interesting. Hyperbolic use of the word “shocking!” is the one thing that the internet likes more than cat pictures).

As a preface, I feel compelled to point out that this entire conversation is premature. LeBron is only 32 years old. Considering how well he has played this year, especially in the playoffs, I wouldn’t be shocked if he puts up five more All-NBA seasons. However, LeBron has played in 14 seasons, 1061 regular season games, and 41,272 regular season minutes, while MJ played in 15 seasons, 1072 regular season games, and 41,011 regular season minutes. Thus, the length of their careers have been almost identical.

Below, I have tables comparing regular season and postseason stats. The player with the better stat will be in bold. Choosing which stats to use is arbitrary, but I’ve tried to use a sufficient variety of stats so we can get a complete picture. I tend to prefer advanced stats, but I included more traditional stats as well. Enough chitchat, let’s get to it.

Regular Season Statistics


LeBron James

Michael Jordan

Points Per Game 27.1


Rebounds Per Game

7.3 5.9
Assists Per Game 7.0


Steals Per Game

1.6 2.2
Blocks Per Game 0.8


Field Goal %

.501 .497
True Shooting % .584


Player Efficiency Rating

27.6 27.9
Offensive Rating 116


Defensive Rating

103 103

Win Shares



Win Shares/48 Minutes .239


Box +/-



Value Over Replacement Player 115.9


My takeaway: Wow, are those numbers close. LeBron wins six categories, Jordan wins six categories, and two categories are ties. Jordan fans will say that his Wizards years are pulling down his numbers, but don’t forget that LeBron entered the league as a teenager. If you were to remove the first one or two years of LeBron’s career, his numbers would go up.

Let’s look at the postseason. Surely the six time champion and the founding father of clutch will pull away here, right?

Postseason Statistics

Category LeBron James

Michael Jordan

Points Per Game

28.3 33.4
Rebounds Per Game 8.8


Assists Per Game

6.8 5.7
Steals Per Game 1.8


Blocks Per Game

1.0 0.9
Field Goal % .483


True Shooting %

.573 .568
Player Efficiency Rating 27.8


Offensive Rating

115 118
Defensive Rating 101


Win Shares

44.9 39.8
Win Shares/48 Minutes .242


Box +/-

10.7 10.1
Value Over Replacement Player 28.6


Nope. LeBron wins eight categories and Jordan wins six. (To be fair, the Win Shares Category is not a perfect comparison because LeBron has had 33 extra games to accrue win shares. But if you don’t consider that category, LeBron would still lead seven categories to six). But again, look at how similar those stats are. The shooting percentage and true shooting percentage are nearly identical, as is the player efficiency rating, win shares/48 minutes, and box +/-. Really, it comes down to which stats you think are the most meaningful. Do you value scoring more than you value rebounding and passing? Is win shares/48 the better advanced stat, or is it Box +/-?

Looking only at these regular season and postseason statistics, if you were to ask me to choose the better player, my answer would be a giant ¯\_(ツ)_/¯. As much as I love stats, they simply aren’t sufficient to differentiate in this scenario.

So if stats aren’t enough, maybe personal/team awards can be the tiebreaker. Let’s look at their achievements.


LeBron James Michael Jordan
MVPs 4


Defensive Player of Year

0 1
All-NBA Selections 13


All-Defense Selections

6 9
All-Star Selections 13


Jordan clearly has the edge here with more MVPs, more DPOYs, more All-Defense selections, and more All-Star selections. But LeBron is still only 32 years old. What if LeBron brings home two more All-NBA selections and four more All-Star selections? What if LeBron wins another MVP? Jordan won his final MVP at the age of 35, so it isn’t inconceivable. Karl Malone, whose body type is often compared to LeBron’s, won the award just a few short months before his 36th birthday. LeBron may have 3-4 more seasons where he has a legitimate shot at winning the MVP. Everyone knows he’s still the best player in the world. He would just need to decide to take the regular season a bit more seriously and his 5th MVP would be all but guaranteed.

Even if LeBron doesn’t bring home another MVP, a few more All-NBA and All-Star selections seem like a safe bet considering LeBron’s age and durability. Would you rather have an extra MVP award and a DPOY award or four extra All-NBA seasons? In other words, what’s better: (1) one extra MVP season or (2) four extra All-NBA seasons? Again, it comes down to preference and how much you value peak performance versus longevity. I’ll give the slightest of edges to Jordan for the time being, but I have a feeling that I’ll change my mind as long as LeBron can keep up his historic pace for a few more years.

After all that, there’s only one more thing to look at: championships. As everyone knows, Jordan has six and LeBron has three.* Likewise, Jordan has six finals MVPs and LeBron has three.α But here’s the thing that always bothers me about the “ringzz” argument. The Bulls were a legitimately good team even without Jordan. During the 1992-1993 season, Jordan carried the Bulls to a 57-25 regular season record and their third straight championship. Following Jordan’s first retirement from basketball, the Bulls plummeted to 55-27. The year after, with Jordan still playing baseball, the Bulls went 47-35. Even without Jordan, the Chicago Bulls were a team that could win somewhere in the vicinity of 50 games. After LeBron took his talents to Miami, the Cavaliers went from a 61-win team to a 19-win team! And when he headed back to Cleveland, Miami went from a 54-win team to a 37-win team. There is no debate; Jordan had a better surrounding cast than LeBron, even after LeBron formed his Miami “super team”.β If their careers had been switched, would LeBron had won six championships with the Bulls? We’ll never know. But it seems entirely possible considering how talented those Bulls teams were even without Jordan. In fact, LeBron almost assuredly wouldn’t have retired early and the Bulls could’ve competed for the 1994 and 1995 championships as well.

Considering all of the above, if I were forced to choose, I’d probably still go with Jordan for now. But another championship, MVP, or All-NBA selection by LeBron might be enough to change my mind. Regardless of what you think, let’s just enjoy watching LeBron while we have him, because we’ll never see a player quite like him again. Besides, you need to stock up on these memories now so in twenty years from now you can tell your kids how much better the NBA used to be and how the players of the 2030’s never would’ve been able to compete in a league with the likes of LeBron, Kevin Durant, and Kawhi Leonard.


*Don’t even get me started on the “Jordan is a perfect 6-0 in the finals.” Would you think more highly of LeBron if he were 3-0 in the finals rather than 3-4? Those four finals losses signify the crazy amount of success that LeBron has had against the rest of the Eastern Conference. Would you regard him more highly if he had not carried the 2007 Cavaliers to the finals despite a poor surrounding cast? This is your friendly reminder that the second highest scoring player on that 2007 Cavaliers team was Larry Freaking Hughes.

α Until my dying day, I will argue that LeBron should have won the Finals MVP in 2015 despite losing to the juggernaut Warriors. With Kevin Love and Kyrie Irving injured, LeBron carried Tristan Thompson, Matthew Dellavadova, and company to two wins against a budding dynasty. LeBron averaged 35.8 points, 13.3 rebounds, and 8.8 assists for the series

β As a side note, this is why I always thought that the criticism that LeBron got for leaving Cleveland was unfair. Jordan never had to leave Chicago because he was eventually surrounded by stellar players.


The Best and Worst NBA Draft Picks of the Past 25 Years

Today I’d like to look at the best and worst NBA lottery draft picks over the past 25 years. One way to determine whether a certain draft pick was good or bad is to compare a player to other players taken in the same spot in other years. For example, after we have determined the average value of a player taken 5th, we can quantify how smart/lucky the Heat were when they took Dwyane Wade 5th in the 2003 draft.

If you’ve been following the most recent posts, you’ll know that I have gathered data on every lottery pick for the past 25 years. For this post, I will exclusively be looking at Win Shares/48 (WS/48) because this is a statistic that shows the overall value of a player while also taking out the effect of playing more minutes/having a longer career. My method was simple; I figured out the average value and the standard deviation of a player taken at each spot in the draft. From there, I determined which players were the farthest from the average on both ends of the spectrum. I also used the normal distribution function to determine each player’s percentile. For example, a player in the 95th percentile is better than 95% of the players taken at that spot in the draft.

Let’s start with the best 10 picks.

Year Pick Number Name WS/48 Std Dev Above Mean Percentile



Chris Paul






Kobe Bryant






Brandan Wright






Kevin Durant






Dirk Nowitzki






Brandon Roy






Stephen Curry






Dwyane Wade






Lebron James






Kevin Love




Let’s briefly talk about each pick.

1. It turns out that Chris Paul is really good at basketball. He has yet to win an MVP in his 10-year NBA career, but his all-around game has propelled him to several top five finishes. Getting him at any spot in the draft would have been a good value. Getting him fourth was a steal (which is ironic because Paul has led the league in steals five times).

2. One of the all-time Lakers greats, and he was drafted 13th. Incredible. Just more proof that the Lakers seem to get all the breaks.

3. Ugh. Brandan Wright ending up 3rd makes the whole list seem a little less valid. But Wright is a pretty good player on a per-minute basis, and I chose to use a statistic that relies on per-minute production. Let’s just move on.

4. Even as the second pick, Kevin Durant was a great value. Including this year, he has lead the league in scoring four of his seven years in the league. And he’s still only 25. You can’t help but be excited about Durant’s future.

5. Like Kobe, Dirk Nowitzki is a league MVP and an NBA champion that was picked outside of the top five. They are the only such players in the past 25 years.

6. Multiple knee injuries prematurely ended Roy’s career, but when he was healthy, he was one of the best 10-20 players in the league. Considering how surprisingly awful the 6th pick usually turns out (see my “Is Tanking Worth It?” post), getting Roy at 6th was a great bargain.

7. Looking back, it’s hard to understand why Curry was the 3rd point guard taken in the draft that year. Besides being one of the best shooters in the NBA, Curry is currently 3rd in assists per game this season and he is a surprisingly good rebounder for his size.

8. Ever since he led the Heat to the championship in his third season, Wade has been one of the most feared & respected players in the game.

9. Love Lebron or hate him, you have to recognize that Lebron is one of the greatest to ever play the game. Even amidst his #1 pick peers, Lebron stands far above the crowd.

10. For some inexplicable reason, Kevin Love wasn’t a full-time starter until his 3rd season in the NBA. Ever since his coaches wised up, he’s established himself as one of the best players in today’s NBA. Just like Durant, Love is only 25 and his best is likely still ahead of him. (As an aside, if you were hoping to someday be an NBA superstar, you should have been born in September 1988 and your parents should have named you Kevin. Hindsight is 20/20, I guess.)

Now, let’s get to the fun stuff. Here are the worst 10 picks of the past 25 years. (Note that I excluded all players taken in the past four years because I’m sure that many of these players will improve their numbers as their careers progress. Why four years instead of three or five? Because four felt right and because this is my blog.)

Year Pick Number Name WS/48 Std Dev Below Mean Percentile



Aleksander Radojevic






Nikoloz Tskitishvili






Bobby Hurley






Terrence Williams






Fran Vasquez






Adam Morrison






Jonny Flynn






Rafael Araujo






WillIam Avery






Michael Olowokandi




1. Never heard of Aleksander Radojevic? You’re probably not alone. In his brief NBA career, he had nearly as many turnovers + personal fouls (57) as points + rebounds (62). Even as a 12th pick, those are atrocious numbers. His WS/48 are more than 3.5 standard deviations below the average 12th pick, meaning that his WS/48 were only better than .02% (1 in 5,000) of all players picked 12th! According to Wikipedia, he couldn’t even find success in his stints in the European leagues. On the bright side, he earned nearly $5 million during his NBA career. Also, at 7 foot 3, he’s probably never had to ask a store associate for help reaching the top shelf . So don’t feel too bad for Radojevic.

2. Please allow me a personal aside. As John Stockton and Karl Malone neared the ends of their Hall of Fame careers in the summer of 2002, I decided to switch my loyalty from the Jazz to my hometown Nuggets. After acquiring a new fan, the Nuggets’ next order of business was to select Nikoloz Tskitishvili 5th overall in the 2002 draft. I was hoping that the 18-year old, 7 foot forward would become the next Dirk Nowitzki after he had a few years to develop. Needless to say, that didn’t work out. As a rookie, he only averaged 3.9 PPG with a 29% shooting percentage even though he started about a fifth of the Nuggets’ games that year. My younger brother and I quickly assigned him the clever nickname “Stink-ishvili.” A top 5 pick shouldn’t have negative win shares every year of his career. The Nuggets would have been better off if they had told the NBA “we’re actually not going to use our 5th pick this year.”

3. Bobby Hurley, the two-time All-American from Duke, could never translate his skills to the professional level. He contributed negative Win Shares in four of his five years in the NBA. He was one of the most disliked players in college, so I’m sure that many people were happy to see him struggle.

4. Terrence Williams is surely the only guy on this list who has recorded a triple-double in the NBA. Williams is no longer in the NBA, but just 5 weeks ago, he set the Los Angeles D-Fenders (D-league) single game scoring record. I wonder if Jack Nicholson was courtside.

5. Fran Vasquez had indicated a desire to play in the NBA, but after being drafted 11th by the Magic, Vasquez decided to keep playing basketball in Spain. If only Stinkishvili had taken the Fran Vasquez route.

6. Adam Morrison was another college star that didn’t pan out in the NBA. Fans who were hoping that tube socks and mustaches would be the next NBA fad are sorely disappointed.

7. Flynn was ok as a rookie, but he never really seemed to recover from a hip surgery that took place the summer after his rookie year.

8. This one hurts a little bit because Rafael Araujo is from BYU, my alma mater. As much as I’d like to think otherwise, Araujo was an unequivocal bust.

9. We don’t expect a lot from a 14th pick, but the 2.7 points per game that Avery averaged over his career is a disappointment.

10. Olowokandi is often listed as the biggest bust of the past several decades because such lists tend to focus on #1 overall picks. The Candy Man’s career was no doubt a disappointment, but his solid defense earned him a roster spot in the NBA for a decade (not to mention $37 million in career earnings. If that’s what it looks like to be the biggest bust of the past 25 years, count me in!). I hope that this post has convinced you that there have been many players who were more disappointing than Olowokandi.

Is Tanking Worth It?

In years like this one when college basketball is full of enticing NBA prospects, many NBA teams seem to throw away their seasons in an attempt to receive a higher pick. General Managers trade away their best players before the season begins (the 76ers and Celtics last summer are both great examples), teams allow some of their best players to leave in free agency (such as the Utah Jazz last summer), and some teams have key players miss games for suspicious injuries (I seem to recall the Warriors employing this tactic a couple years ago). The strategy is certainly tempting when you consider how the right draft pick can alter the course of a franchise. The Spurs have been arguably the best franchise in basketball since drafting Tim Duncan; Lebron took the Cavaliers to the finals just four years after the team finished 17-65. But these are the success stories. Other teams continue to struggle even though they get a high lottery pick every year. So is tanking worth it? What is the expected value of getting a higher draft pick?

As I mentioned in my “Best & Worst NBA Draft Classes” post, I have assembled a large data table with stats for each player taken in the lottery in the past 25 years. I will be using win shares (WS) and win shares/48 (WS/48) to measure the overall success of a player. WS approximates the overall value of a player, while WS/48 approximates the overall value of a player per 48 minutes played (please see the previous post if you need a little more explanation on these stats). Because we are trying to determine the expected value of each pick in the lottery, we will be looking at the average stats of every player chosen in that slot over the past 25 years.

The charts below illustrate the average statistics of a player taken at a certain spot in the lottery. The X-axis is the pick number and the Y-axis is the average statistic for a player selected with that pick. (Click on a chart to see a magnified version)

Average WS

Average WS-48

Several interesting facts emerge from these charts. The first pick in the draft has an expected value that far exceeds any other pick. Picks 2-5 are nearly identical. Surprisingly, pick 6 has one of the lowest expected values of the entire lottery, and picks 7-8 aren’t much better. However, players picked 9th have an expected value nearly equal to players picked 2-5 (if you look at WS/48). After the 9th pick, the expected value gradually declines, except for another unexpected increase at pick 13. (Lest you think that Kobe being picked 13th is skewing the data, I tried taking his stats out of the equation and the result wasn’t much different.)

Let’s look at the data a little differently. The following charts show the percentage decline from one pick to another. For example, both charts show that a player picked second has an expected value that is about 20% less than a player picked first. A negative percentage means that a player drafted in that spot actually has a higher expected value than the player selected one spot previously.

% Decline (WS)

% Decline (WS-48)

Both charts show similar results, but for simplicity’s sake, let’s just refer to the WS/48 chart for a moment. *If the end of the season is near and your team has the 6th worst record in the league, you may want to lose a few extra games because moving from the 6th pick to the 5th pick increases the expected value of your pick by nearly 30%! However, if you already have the 5th worst record, the chart shows that moving from the 5th pick to the 4th pick will only increase the expected value of your pick by about 2%. The answer to whether tanking is worth it depends on where you expect to fall in the lottery. Getting a top 5 pick is very valuable, but pick 6 isn’t noticeably better than pick 14. And the 14th worst team got to enjoy a season in which it nearly made the playoffs, while the 6th worst team endured an excruciating season (unless you’re a Bobcats fan, who would consider a 28-54 season a raging success!)

* My analysis here ignores some of the complexity of the lottery system. The worst team in the league isn’t given the first pick; it is merely given the highest probability of receiving the first pick. A team that is tanking needs to consider whether it is worth losing more games in exchange for the mere probability of the highest pick. On the other hand, the lottery is structured so that a team can confidently predict what range its pick will fall into. For example, the worst team in the league is assured of receiving a top four pick. For more information about how the lottery works, look here: http://en.wikipedia.org/wiki/NBA_draft_lottery.

Best and Worst NBA Draft Classes of the Past 25 Years

During my week off from school, I have assembled a large data table with stats for each player taken in the lottery in the past 25 years. I am hoping to use this data for several research projects in the coming month. The first (and easiest) project is to look at which years of the NBA draft produced the best players (lottery only*-I only had a week off from school, not a month).

I looked at win shares (WS) and win shares/48 (WS/48) to measure the overall success of a player. As a reminder, WS attempts to approximate the number of wins that a player is worth to his team. WS/48 divides a player’s WS by every 48 minutes that the player played, thus taking out the effect of playing more minutes. If two players are identical and Player 1 plays twice as many minutes, Player 1 will have twice as many WS as Player 2 but both players will have the same WS/48. WS/48 is critical in this study because we are comparing players that have had varying lengths of careers. Kobe Bryant and James Harden have nearly identical WS/48, but Kobe has 173 WS while Harden only has 41 because Kobe has been in the league for thirteen additional years.

For the above-mentioned reasons, looking at overall WS isn’t terribly helpful. The five draft classes with the fewest combined win shares are the five most recent draft classes. The five draft classes with the most combined win shares are all from the 90’s because they contain players who are retired or nearing the end of their careers. But if we look at WS/48, a much more useful picture emerges. Without further ado, here are the five best and five worst classes.

Five Best Draft Classes

1. 1996-Kobe, Ray Allen, Allen Iverson, Marcus Camby, & Peja Stojakovic anchor the best draft class of the past 25 years.

2. 2003-This class is really top-heavy with Lebron James, Carmelo Anthony, Chris Bosh, and Dwyane Wade. But a lack of depth pushed 2003 to #2.

3. 1997-Tim Duncan, Chauncey Billups, and Tracy Mcgrady are the highlights here.

4. 1998-Dirk Nowitzki, Paul Pierce, Vince Carter, Antwan Jamison, and Mike Bibby are enough to overcome a horrific top pick (I’m looking at you, Olowokandi)

5. 2007-Kevin Durant, Joakim Noah, and Al Horford headline a deep class (and Greg Oden’s WS/48 numbers are actually very good! Who knew?)

Five Worst Draft Classes**

1. 2000-Kenyon Martin was the first overall pick and after looking at the rest of the class, I think that was the right choice. That’s all you need to know.

2. 1990-Gary Payton is the only name in this draft that many younger NBA fans have ever heard of.

3. 2002-Yao Ming, Amar’e Stoudemire, and Nene were solid picks, but this is probably the worst top 5 that you’ll ever see (spoiler alert for future blog posts: Nikoloz Tskitishvili, taken 5th, may very well be the worst pick of the past 25 years)

4. 1993-There’s really nothing notable about this class, and that’s the problem.

5. 1994-After a decent top 5 that included Jason Kidd and Grant Hill, this lottery contains a bunch of players with very short NBA careers.

For those of you who need to see to believe, here’s a PDF chart to show the year-by-year breakdown: Yearly Average WS per 48

*25 years ago, the lottery only consisted of 11 players because there were only 27 teams in the league. As the league has expanded, so has the lottery. I looked at the first 14 players taken every year.

** 2013 and 2011 should be #1 and #3 on this list, respectively, but I didn’t think it was fair to include them. As players from these years reach their primes, their productivity will presumably increase and so will their WS/48.

A Few Websites Worth a Look

I thought that I’d list some of my favorite websites that deal with sports statistics. You’ve probably already heard of most of these, but maybe there’s one or two that you haven’t seen before.

If you like testing your knowledge of sports trivia, I highly recommend Sporcle. (http://www.sporcle.com/games/category/sports) It has a bunch of quizzes covering every sport, and new quizzes are being added daily. While your there, try a quiz that I made: http://www.sporcle.com/games/Drewmoney/nba-win-share-leaders-by-position

ESPN recently published an excellent article by Nate Silver (the guy who correctly predicted all 50 states in the last presidential election). It’s a good primer to how analytics is changing sports. Here’s the link: http://espn.go.com/espn/story/_/id/10476210/nba-mlb-embrace-analytics-nfl-reluctant-espn-magazine

I think that Grantland (grantland.com), which is associated with ESPN.com, has some great journalists that use a lot of data analytics. If you haven’t been on Grantland before, I recommend you look up Bill Barnwell (NFL), Kirk Goldsberry (NBA), Zach Lowe (NBA), and Jonah Keri (MLB).

The Wages of Wins journal (http://wagesofwins.com/) has some interesting articles. Their focus is pretty similar to the focus of this blog.

For pure data, sports-reference.com is the best site I’ve ever seen. It has websites devoted to MLB, NBA, NFL, NHL, NCAAB, NCAAF, and more.  A lot of the data that I use for my research projects comes from these websites.

If you’re interested in salaries, spotrac.com has lots of good information.

If you have any good websites that I haven’t listed here, I’d be glad to hear about them in the comments section.

What is the right amount to pay a quarterback?

Are teams with highly-paid quarterbacks more likely to win games? Russell Wilson had the third lowest salary of any starting quarterback in the league (Mike Glennon of the Buccaneers and Nick Foles of the Eagles had the lowest salaries), and his team just won the Super Bowl in convincing fashion. Many other cheap QBs, such as Foles, Andrew Luck, Andy Dalton, Colin Kaepernick, and Cam Newton also made the playoffs.  On the flip side, some highly paid quarterbacks such as Matthew Stafford, Joe Flacco, Eli Manning, Sam Bradford, and Ben Roethlisberger had disappointing season. You may also remember Jim Irsay taking some not-so-veiled shots at Peyton Manning by suggesting that paying a lot of money to a quarterback is not the best way to build a team.

People are naturally wondering if Jim Irsay is right. Is it worth it to throw a lot of money at the most important player on the field, or would that money be better used elsewhere? I decided to examine this problem by looking at the correlation of a QB’s salary and the team’s number of wins during the regular season. I ran a regression using data from the 2013 NFL season. I looked at the salary of the highest paid QB on the team, not necessarily the player who started the most games. (Even though Michael Vick only started six games for the Eagles in 2013, the Eagles’ decision to give him a big contract made an impact on what other players the team could sign.) Also note that I used a player’s cap hit, rather than his base salary, in my regression. I’m not going to pretend like I know the intricacies of the NFL salary cap, but a player’s cap hit seems like a more relevant piece of information if we’re trying to decide how a team can optimally allocate its money. All salary information is from Spotrac.com.

If you’re like me, you’re not going to like the answer to the question that I posed at the beginning of this post. The regression was statistically insignificant, which means that it doesn’t really matter whether you have a cheap or an expensive QB. The salary of the QB does not have a correlation with the success of the team. There are multiple ways to put together a successful team. Maybe it’s not a coincidence that the Super Bowl pitted one of the cheapest quarterbacks against one of the most expensive quarterbacks.

For those of you who still want a little more info even though you know that it’s not statistically significant, a QB’s salary is positively correlated with the number of regular season games that his team wins. If Team A pays its QB $12 million more than Team B, Team A is expected to win one additional game. The R-squared is .023.

Is passing or running more correlated with success in college football?

Near the end of the 2012 college football season, I got a little bored in my Federal Income Tax class, so I decided to run some regressions to determine whether a good passing game or a good running game had a stronger correlation to success (points scored and games won). What I learned was a little surprising.* I learned that a team’s rushing yards per game has a higher correlation than the team’s passing yards per game with (1) scoring points and (2) winning. Despite the popular belief that modern college football is all about the passing game, a good running game is more likely to lead to success.

That was the sparknotes version of the research; bear with me if you want more of the details.
Here’s the correlation and statistical significance breakdown of six of the regressions I ran:

  1. Pass yards-Points R squared=.22 (statistically significant)
  2. Rush yards-Points R squared=.31 (statistically significant)
  3. Pass yards-Wins R squared=.03 (not statistically significant)
  4. Rush yards-Wins R squared= .21 (statistically significant)
  5. Total yards-Wins R squared=.28 (statistically significant)
  6. Points-Wins R squared=.52 (statistically significant

Look at that difference in R squared between pass yards-wins and rush yards-wins. Rushing yards per game (ypg) can explain 21% of a team’s win total while passing ypg can only explain 3% of a team’s win total. If you knew nothing about a team except how many ypg the team rushes, you should have a hazy idea of how many wins the team will accrue. If you knew nothing about a team except how many ypg the team passes, you might as well just guess a random number if you’re trying to predict how many wins the team has.

Now look at the difference in R squared between total yards-wins and points-wins. Unsurprisingly, scoring points is a much better indicator of winning that gaining yards. This affirms the importance of a good red zone offense. Running up and down the field won’t do you much good unless you can convert those yards to points. (I was also interested to see that the R-squared for points-wins was so close to .5. I’m guessing that if I ran a points allowed-wins regression, the R-squared would be very close to .48. If this is true, it would suggest that scoring and stopping the other team from scoring are nearly equal in importance to winning)

By looking at the coefficients of each regression, we can learn a few more interesting facts. 

  1. If Team A has 40 more total ypg than Team B, Team A is expected to win one more game than Team B (during a 12 game regular season schedule).
  2. If Team A has 40 more rushing ypg than Team B, Team A is expected to win one more game than Team B.
  3. If Team A has 118 more passing ypg than Team B, Team A is expected to win one more game than Team B
  4. If Team A scores 4 more points per game (ppg) than Team B, Team A is expected to win one more game than Team B
  5. If Team A has 10 more ypg than Team B, Team A is expected to score 1 more ppg.
  6. If Team A has 12 more ypg than Team B, Team A is expected to score 1 more ppg.
  7. If Team A has 16 more ypg than Team B, Team A is expected to score 1 more ppg.

These facts only show correlation, not causation, but they suggest that a team hoping to improve its offense should focus on the running game rather than the passing game. I think that point 4 is intriguing. It suggests that (all else being equal) a team that hopes to improve its record will need to improve its offense by 4 ppg for every additional win the team hopes to achieve.

Next time you hear a football analyst proclaim that college football is all about the passing attack, just remember that rushing yards are still far more indicative of how well a team is playing.

*It was surprising to me, at least. My friend, Barry, responded to my research by saying “this fact is a little obvious. Teams that can run are more likely to win because they are more likely to have a two-dimensional offense. They are more likely to have the ball longer. Also, teams pass more to catch up when they’re losing and run more when they’re winning to run out the clock. Any way you put it, running the ball is important.” Barry and I disagree about how surprising my result was, but I think his analysis is a good explanation for why the running game is still important in football. 

An amateur's attempt to explain sports through statistics