How much of the game is passing, running, kicking, or punting? How much of football are interceptions? Fumbles? How much do penalties impact the outcomes of games? One way to answer that is to simply add up how many of those kinds of plays occurred and divide by the total number of plays. But that’s not going to tell us much, because each kind of play tends to have a different magnitude in terms of its effect on the outcome of games.

When we talk about “how much of football is” something, we need a good definition of what football is. The essence of football, like any other sport, is about competing and winning. So when we ask how much of football is passing, we want to know the impact of passing plays on the outcomes of NFL games compared to other types of plays.

Win Probability can provide the answers. In each game, the WP of each team starts at 0.5, but must end at 1 or 0, for a net total of 0.5 WPA. But between the first kickoff and the final whistle, the WP can swing up and down, traveling far more than the net 0.5. To calculate the total movement, we just need to add up the absolute value (7th grade flashback) of each play’s Win Probability Added (WPA).

Some of the more exciting NFL games feature WP graphs that travel a total of 7 or 8 or more “wins.” In the nearly 2,900 games in my database, there is a total of about 12,000 “wins” worth of total WP travel, for an average of about 4.3 per game. In other words, the WP needle moves, either direction, a total of 4.3 “wins” between the beginning and end of the game.

Now we just need to see how much of that total 12,000 WP movement was due to each type of play. For example, passing plays account for 6,600 of the total WP movement, which equates to 55%. Here are the other play types. (Note that there will be some overlap, so the total will sum above 100%. For example, pass plays also include sacks, penalties, and interceptions.)

Play Type | % |WPA| |

pass | 55.0 |

run | 24.6 |

penalty | 9.1 |

kick off | 4.1 |

punt | 3.7 |

field goal att | 3.5 |

other | 1.3 |

sack | 4.7 |

interception | 4.7 |

fumble lost | 3.7 |

touchdown | 8.8 |

We can see why passing has bigger coefficients in the efficiency regression models. It’s more than twice as important in terms of determining game outcomes than running. I was surprised that the % of WP due to penalties wasn’t higher. Often it seems like a few critical flags in high leverage situations can heavily sway a game. Penalty |WPA| does not include plays that featured declined penalties.

Turnovers, both interceptions and fumbles lost, account for 8.4% of game outcomes.

In total, special teams plays—kick offs, punts, and field goal attempts, comprise 11.3% of game outcomes.

Touchdowns plays themselves account for 8.8% of game outcomes. Together with field goals, scoring plays comprise 12.3% of “football.” Most of the other 87.7% doesn’t make the highlight reel.

‘Other’ plays consist of aborted plays, kneel downs, spikes, and plays that could not otherwise be classified.

We can have even more fun with this. Let’s break down |WPA| by quarter.

Quarter | % |WPA| |

1 | 23.4 |

2 | 24.8 |

3 | 18.9 |

4 | 30.9 |

OT | 2.1 |

It’s understandable that the 4th quarter is more determinative than the other three. At first it appears there is something odd with the third quarter, accounting for an unexpectedly low proportion of game outcomes. But it’s actually in line with the other quarters. It's actually the 1st quarter that has a higher |WPA| per play than the 2nd or 3rd quarters. It’s just that teams tend to play at a relatively slow pace in the 3rd quarter. I suppose that’s when teams with leads start slowing down the game. There are also different rules for clock stoppage in the latter quarter of each half.

How about breaking it down by play location? Since 2006 the official NFL play-by-play breaks out seven run locations from left end to right end. It also divides the field into short and deep left, right, and middle for pass plays. Deep is defined as a pass thrown greater than 15 yards in the air from the line of scrimmage.

Location | % |WPA| |

short left | 16.3 |

short middle | 12.8 |

short right | 19.1 |

deep left | 6.6 |

deep middle | 5.5 |

deep right | 6.7 |

Location | % |WPA| |

left end | 4.7 |

left tackle | 4.1 |

left guard | 3.0 |

up the middle | 9.4 |

right guard | 3.4 |

right tackle | 4.3 |

right end | 4.1 |

About 81% of all pass attempts are short, so deep attempts are disproportionately determinative of game outcomes, by a factor of about 2 to 1. On a per play basis, the ‘middle’ location for both deep and short passes were both 10% more determinative than the left or right locations at the same depth.

“Up the middle” is far more common than the other run locations. On a per play basis, all the run locations are roughly similar in terms of their impact on winning.

So, unsurprisingly, the NFL is indeed a passing league. In fact, it’s a passing league to the tune of 55%, which is a little more than twice as crucial as running. Penalties don’t control game outcomes as much as I had worried.

So assuming offense and defense are equal, and special teams is around 11%, then we can assume that offense and defense are about 44% each, or 4x as important. Seems about right.

It would be interesting to see this on a per play basis.

So interceptions are really no worse than sacks? And turnovers don't seem to be nearly as important as the mainstream media seems to think.

"So interceptions are really no worse than sacks? And turnovers don't seem to be nearly as important as the mainstream media seems to think."

When the game is still close, I would think interceptions are significantly worse than sacks. However, a large percentage of INT's are thrown by the trailing team, often when the game is essentially out of reach, therefore lowering the WPA effect of those plays.

Just so no one is confused, the percentages in the article are not saying that any single sack has the same impact as any single interception. It's that the impact of

all sacksis as big as the impact ofall interceptions. It's just that sacks are more common.I would expect that sacks are also more common when a team is trailing and trying to catch up.

Yeah, sacks are roughly twice as common as ints. Still, you often hear about winning the turnover battle as a key to the game but rarely the sack battle. And saying that turnovers account for only 8.4% of the game's outcome certainly doesn't match the conventional wisdom.

I wonder whether it's worth comparing the difference between running and passing by quarters. I would expect to see running above passing in the 3rd quarter, as leading teams start to try to control the clock and trailing teams start passing more to chase the game.

At first it appears there is something odd with the third quarter, accounting for an unexpectedly low proportion of game outcomes. But it’s actually in line with the other quarters. It's actually the 1st quarter that has a higher |WPA| per play than the 2nd or 3rd quarters. It’s just that teams tend to play at a relatively slow pace in the 3rd quarter.There's a pattern in the NFL of low scoring in the 1st Q, high in the 2nd, low 3rd, high 4th. I presume this is because after a half-opening kickoff it usually takes a while for somebody to score, while in the 3rd and 4th Qs scoring can happen from the first play. Would this pattern affect WP by Q? If so it might help explain the low number for the third, but leaves the higher one for the first a mystery.

It’s understandable that the 4th quarter is more determinative than the other three.I would think so, but thinking more about this it becomes less obvious. A question was asked earlier: why is a TD worth less early in the game than late, when it is worth 7 points all the time? Well, it's obvious that a TD in the last moment can determine a game with a huge WP effect while one in the first minute never will. But only a few TDs late in the game have a such a big effect, many more occur in "junk time" with the game already determined and so have negligible WP effect, while all TDs early in the game will have some positive WP effect. So possibly on average TDs early and late do have the same WP value. Applying the same logic to all other plays, it is not self-evident to me that total WP should be higher late in the game than earlier, as intuitive as it seems.

I wonder if other factors, such as coaching strategies, are affecting this number. E.g, the 4th Q is one of the two high-scoring quarters to begin with, and by saving time outs until then and going "hurry up" in close games coaches may squeeze in more plays, boosting their WP total.

Saying that passing is twice as crucial as running could be somewhat deceptive...

Basically your results show that completions, incompletions, sacks, and interceptions are more influencial on winning/losing games than running the ball. Was there any doubt?

I disagree that "[the NFL is] a passing league to the tune of 55%". An analogy would be a conservative investment fund, with 90% of its assets in gov't bonds and cash, and 10% in some relatively aggressive stock. Perhaps 55% of the money made or lost (WPA) will come from the risky stock, but I certainly wouldn't call the conservative fund a "aggressive stock fund to the tune of 55%"... is that a decent analogy?

I'm not hating, it's good to see the actual numbers... and I love the site. Keep it up!!!

In 2009, there were approx 18,700 passes and 12,900 runs in about 43,000 total plays. If that year is typical, then with all other things being equal we'd expect passing to account for about 43% of the WP change and running to account for about 30%, and the two together to be about 73%. But all other things are not equal. The likelihood of a turnover is much higher during a passing play (as is a sack, obviously). As a matter of fact, just eyeballing it, it looks like most of the extra WP for passing comes from turnovers and sacks.

Fascinating stuff. You have developed a 'big play' index. But this distribution fairly closely follows play selection. For example, nearly half of all plays are passing plays, so it's not too surprising that half of all big plays are passing plays. And it's not too surprising that running plays are under-represented because most running plays are pretty boring. And turnovers are a small percentage of big plays because they don't happen very often.

It would be really interesting to look at what percentage of each type of play generates a big play - what percentage of passes has a high WPA, what percentage of runs has a high WPA. Then I think you'd find that a high percentage of turnovers and special teams plays have a high WPA.