WIN PROBABILITY GRAPHS

Check out the Win Probability graphs and play-by-play of your favorite team's biggest comebacks and most exciting games since 2000. An explanation can be found here. Just select a year, a team, or 'any', and start clicking:

Or search for all the games for your favorite team:

Or browse the current season by week:

Jul 11, 2007

What Makes Teams Win? 1

This is the first part of a four part article summarizing the relative importance of each phase of the game in winning.

INTRODUCTION

Passing is much more important than running, and offense appears to be important than defense. Turnovers matter a great deal, interceptions more than fumbles. Penalties matter too, but not like you’d think.

We've been told by NFL analysts for years that defenses win championships and that a solid running game is the key to winning. I'd agree that things appear that way. Teams with lots of passing yards aren't always winners, but teams with lots of running yards almost always are. But, as you're about to find out, appearances can be deceiving.

One way to measure the relative importance of passing vs. running, or offense vs. defense, is to measure their statistical correlations with season wins. For example, the correlation coefficient of a team's total rushing yards correlates with its number of wins in a season is 0.45. A correlation of 1.0 would be perfect correlation, while a correlation of 0.0 would indicate no relationship. So 0.45 implies that total rushing yards is a moderately strong indication of how many games a team would win.

In comparison, total passing yards correlates with wins at 0.31. Compared with total rushing yards, total passing yards is less important in terms of winning games. This is what most fans and analysts notice when watching games or glancing at team stats. But does this mean that passing is less important than running in the NFL?

Before we settle on an answer, we need to consider the number of attempts of runs and passes. In baseball, the statisticians understood this 150 years ago when they created the batting average. If one player has 300 hits and another player has 350 hits, who is the better batter? The answer is we don't know until we divide the number of hits by their at-bats.

RUN AND PASS EFFICIENCY

How often have you heard an NFL commentator say, "When running back X gets at least Y carries, his team wins?" The clear implication is that the team in question should feed a steady diet of carries to the running back, and this will cause his team to win. What if we stated the same observation this way, "When his team is winning, running back X receives at least Y carries?" And by the way, why don't we ever hear, "When QB X passes at least Y times, his team wins?" Here's why:





Stat Win Correlation
Rush Attempts0.58
Pass Attempts-0.17


The negative correlation for pass attempts means that the more often a team passes, the less likely it is to win. The correlation of rush attempts with wins (0.58) is even stronger than that for total rushing yards (0.45). This is a curious result, and it’s where conventional NFL analysis begins to crumble.

When we see two things that appear correlated, it is natural for us to say that one causes the other. The runs come during the game, and the win comes at its conclusion. Therefore most fans and analysts assume the running causes the winning. The problem is, it usually doesn't. It's the winning that causes the running. Teams that are ahead, and likely to win, run the ball to take time off the clock and to minimize the risk of a turnover. Teams that are behind, and likely to lose, abandon the run in favor of the pass. Statistics can measure the correlation, but it can't determine the direction of causation.

The critical question then becomes: how can we truly measure a team's passing and running abilities and their respective contributions to winning? The answer is football's equivalent to the batting average--efficiency stats. If we want to know how good a team is at running, the best way is to know how many yards it tends to gain each time it runs the ball--yards per rush attempt. The table below lists basic passing statistics and their correlation with season wins.







StatWin Correlation
Pass Yards0.31
Pass Attempts-0.17
Pass Yds/Att0.61


Yards per pass attempt is merely pass yards divided by pass attempts. So we have a relatively weak statistic (0.31) divided by an even weaker one with a negative correlation with winning (-0.17). We would expect to have a fairly meaningless result, but we don't. Passing efficiency turns out to be strongly correlated with winning (0.61). And unless having a lead in a game ‘causes’ a team’s passes to be more successful, we can safely say that passing efficiency leads to winning.

Because sacks are an important factor in the passing game, I include plays that result in sacks as pass attempts for the purpose of calculating efficiency. Likewise, I also subtract sack yards from total passing yards. I call this true pass efficiency and have found it correlates better with both offensive points scored and wins.

Continue reading part 2 of the article.

17 comments:

Derek said...

As someone who was avid TMQ reader (not so much now), I was of the opinion that teams need to run to win the game. The reasoning seems sound because of play action and keeping the defense guessing. You need to at least balance things out and not go "pass-wacky." Looking at box scores validated that when I saw winning teams were well balanced in terms of rush and pass attempts.

Then Football Outsiders shoots back with this really solid bit of logic: Teams that are ahead will run the ball more to run out the clock. Teams actually call more passes early in the game. So it seems that passing game opens up the running game rather than vice versa, the common wisdom.

I guess we're grounded in the tradition of teams like the 70s Dolphins, who could get away with 8 pass attempts in a Super Bowl because they had 2 HOF RBs. But the Dolphins were also #2 in pass efficiency in 1972.

Anonymous said...

The Dolphins of that era did not have 2 HOF Running Backs. The HOF players were QB-Bob Griese, FB-Larry Csonka, WR-Paul Warfield, C-Jim Langer, G-Larry Little, MLB-Nick Buoniconti. Someday we will likely see either Jake Scott or Dick Anderson in the HOF, these gentlemen were the Dolphins safeties.

Anonymous said...

I doubt running efficiency and passing efficiency are independent of each other enough to provide a stronger causal relationsip between one of the two and winning. For example, at the end of the article above, you shrug off the concept of "leading a game 'causes' a team to pass more efficiently". But isn't that what would happen if a team leading were running the ball more (so as to move the clock) and the defense was playing that team to run more - i.e., of the few(er) passes that the leading team would throw, they would be more likely to be completed as the defense would be looking for a run.

That said (and not having read past part 1 yet), I would think that games are won with a balance of both - that may sound like a cliche, but it is up to the coaches and offensive and defensive coordinators to prepare their players before the game and then make the right calls from the booth's by adapting and reacting accurately to the patterns they observe on the field during the game.

Brian Burke said...

You'd be surprised. Over the past 6 seasons ('02-'07), off running and passing efficiencies correlate at 0.12 (p=0.08).

I've tested a regression model that uses an interaction of run and pass efficiency, hoping to find the synergy of run/pass balance. The interaction's coefficient was actually negative! (As in: being good at both hurts. But I would tend to believe the interaction effect is actually very close to zero, and the negative non-significant coefficient was just noise.)

Anonymous said...

The Titans are the only undefeated team yet they have one of the worst "Yards per Pass Attempt" in the NFL....

Brian Burke said...

Actually, the Titans have a very solid 6.1 net yards per pass attempt. That's exactly middle of the pack for the '08 NFL season to-date. The main reason that they are undefeated so far is that they've had the softest schedule by a significant margin. Plus, their net defensive YPA is second best in the league. (Perhaps you've mistaken that for offensive YPA?) The blown roughing the passer call in Baltimore didn't hurt either.

Anonymous said...

After all you got that right. I do that passing research since i read a article in the early 90´s. PASSING efficent wins. Here is the shocker: You even can COMPLETLEY abondon the run and win. Why?
1.) Martz once called 18 consecutive passes. Nothing the D could do. He did call 31passes of 35 plays in 1st half of SB 1999. Nothing the D could do.
2.) Dungy´s (he is conservative) Indy once called 31 (!!) consecutive passes vs. GB. Nothing the D could do.
3.) Most important: I tried to guess the play-calling in Live-Games. On 1st & 2nd Down i was at 50%, meaning i could have rolled the dice. But on 3rd downs i called the play correct in 70%+ of the cases. Mostly i called a pass and was wrong when a run came.
So important is this: when the Defense KNOWS a pass is coming, nothing they could do about it. Evidence: Overall-League-Passing-Efficiency does NOT go down on 3rd Downs. Prediction: The NYG and Titans will NOT win the SB, if some strong opposing Play-off-Defense together w/an opposing efficient passing-offense come their way. Those teams can be: AFC; Pittsburgh or SD (if they make the Play-offs). SD has a strong Run-Defense and the most efficient passing-offense, a great point differential but a idiot coach, thats the problem here (see also the other post here, he is one of the worst coaches in history).
NFC: Minnesota and Philadelphia (if they make the play-offs).

P.S.: I would have liked to give my name, but problems with URL

Anonymous said...

Here is the actual stats for this season (w/o MNF):
Y/PP 182-57 (.762 Winning-Pct.)
Y/PP (Differential more then 0,75 Y/PP = equals a turnover) 162-27 (.857)
Y/R 117-121(.492, one tie)
Total Yds 173-64 (.728, two ties)
TO-Differential: 143-46 (.757)
TO-even 47 Games
if TO even:
Y/PP 36-11 (.766)
Y/R 23-24 (.489)
Total Yds 37-10 (.787)
I do this stats since the early 90´s. Its always the same as this season (only total yards have a higher then normal Wng.-Pct. this year): You win with efficient Passing, it does not matter how efficient you run, you can overcome minus 1 turnover and still win if you pass or defend the pass eficient. Exeption to the rule: NE Patriots. They always won, no matter how "unefficient" their passing was. It was real ugly in the season they upset the SL Rams. Like the author of this page, since then i have my doubts about this team. There is something REAL wrong.

theladyinspring said...

I'm not really mathematically-minded and I'm more just a fan than anything, but as an Eagles fan I am scared to death that all of the local media pressure to have a "balanced run/pass ratio" will impact coaching decisions in the playoffs and lead to an early exit. The Philadelphia media and fanbase have gone crazy over the notion that they need to "run as much as they pass", even though the running game is not very effective (especially in short yardage), they've never won a game because of their running though it can be helpful to set up play/action, and their offensive scheme is built around passing. I guarantee if you read the Philly sports pages over the last few weeks nearly every Eagles article will have something in there about "balance", and the coach is constantly questioned about it. I just hope he ignores it all.

Players need to hold onto balls and run their routes, quarterbacks need to throw accurately, and offensive line needs to protect. That's how you get effective yardage. It's certainly true of the Eagles.

Anonymous said...

Question for the people...I am working on a college assignment regression.

I want to predict WINS like this article. However, my variables are in per game averages and my dependent variable (WINS) are based on the season. For example AZ cardinals have 74 Rushing yards per game average and 9 wins for the season.

What should I do to correlate the two and be able to interpret my regression results.

here is a brief snapshot of one variable rushing yds (per game) the coefficient is 0.017174406.

Does this mean for every additional rushing yard in a game the number of WINS in a season increases by .0171 or should I multiply the coefficient by 16 or divide by 16 to have the same units?? I hope this question makes sense.

Anonymous said...

Brian sent me an answer to the question above so I am posting his response here...

The coefficient for any linear regression is in "units of the dependent variable per unit of the predictor variable."

Let's take RYDS in your model. The coefficient is 0.017. So for every additional running yd/game a team averages, it should expect to win an additional 0.017 wins in the season. (I'm assuming season wins is the dependent variable.) So if a team averages 10 yds/game running more than average, it should expect to win an additional .17 games more than if they were average in running. If they were average in everything else, you could say they'd win 8.17, which the average # wins + 0.17. This is what statisticians mean when say "holding everything else equal..."

You can take the entire set of coefficients and build one master equation to calculate any team's expected wins based on their per game stats. It's a simple linear equation, just like from algebra (y = bx+ a). But this time there are a whole bunch of 'bx's.

y = a (constant or intercept) + b1x1 + b2x2 + b3x3....bnxn

So, in your case, for any given team:

wins = 16.6 + 0.017*RYDS - 0.227*PCMP - 0.028*RYDSA ... + 6.996*AFC

Ketch Rudder said...

The creator of this site abuses statistics routinely.

A good example would be applying the past to an open system (football) to make specious prediction.

The story above gives us another fine example of this abuse.

The writer writes, "... how can we ... measure a team's passing and running abilities and their respective contributions to winning? The answer ... efficiency stats. If we want to know how good a team is at running, the best way is to know how many yards it tends to gain each time it runs the ball--yards per rush attempt."

The writer above expresses a false belief.

In truth, accumulated yards beyond those needed to score amount to inefficiency -- the inability to score.

One team with a higher yards per rush attempt could be a loser, if those rushing yards accumulated without scoring.

Only those yards accumulated during a scoring drive count.

As always, the correct way to measure is to establish the correct context -- the correct ratio.

The correct ratio is the run yards during scoring drives per attempts on those drives to run yards during failed drives per attempts on those, respectively.

That measure shall give you an effectiveness measure.

The same holds true for passing. What counts is passing per completed pass on scoring drives to passing per completed pass on failed drives.

With passing effectiveness, what counts is a completed pass, not an incomplete nor interception since the rules specify that a team must possess the ball, except for a safety.

Probability and prediction can take place only in closed systems, e.g., cards, dice and roulette.

It's impossible to predict any future in any open system.

At most, all that can be said is what were the chances had you engaged in some act during some past time frame.

Anonymous said...

Ketch's claim that "Only those yards accumulated during a scoring drive count" is clearly rubbish. If, in the context of a tied game, you drive from your own 10 yd line to the opponent's 10 yd line, you've clearly produced quantifiable value, even if you lose possession without scoring.

Ketch Rudder said...

Our befuddled "Anonymous" writes, "If, in the context of a tied game, you drive from your own 10 yd line to the opponent's 10 yd line, you've clearly produced quantifiable value, even if you lose possession without scoring."

Poor-minded "Anonymous" suffers from false beliefs, as many do.

What has happened in the story world of "Anonymous", even if he lacks the ability to perceive it, is that the first team turned over the ball (presumably on downs) without a score. The first team racked up 80 useless yards.

If yards accumulated not in scoring drives counted as Anonymous believes, falsely, of course, then all teams would accept the kick-off, advance the ball to the one-yard line and instantly take a knee, thus giving themselves the potential maximum number of yards to accumulate in a possession.

Moreover, teams would catch punts and run them backward to their one yard line, again, taking a knee, and again to give themselves the potential maximum number of yards to accumulate in a possession.

Because most football fans lack understanding of the design of the game, they get confused and thus do not get the game they watch and perhaps love.

In the Science of Winning, scoring counts. Only those teams that score more points than their opponents can win.

Only those yards accumulated during a scoring drive count. Accumulated yards beyond those needed to score amount to inefficiency -- the inability to score.

Brian Burke said...

Ketch-Very untrue. Those yards set the opponent back further from scoring, plus they make it that much easier to score on the next possession, given the defense makes a stop. Those yards have value.

Anonymous said...

Hmm . . . so a team takes the ball from its own 1 to the opponent's 1 in 7 plays, but then fumbles, for 7 yards per rush on an "unsuccessful" drive. The opposition (now on its own 1) fumbles the ball back, and the team drives the final 1 yard on the subsequent play, for 1 yard per rush on a "successful" drive. That's a 1:7 successful-unsuccessful ratio! Such waste! Such inefficiency! This team has forgotten the goal of the game!

I doubt that I've ever read anything so mind-numbingly stupid couched in such pompous, self-assured language. I even visited "Hail the Ale" to learn more about such a confused mind. I left . . . confused.

Anonymous said...

That's 14 yards per play in the former case of course, not that it matters for the illustration . . .

Post a Comment