Despite Grossman’s two interceptions, his impact on the game was a net positive. He averaged 9 yards per attempt, and ended the day with +7.8 Expected Points Added and +0.26 WPA. Most of that came on a single play, the 50-yard bomb to Anthony Armstrong to take the lead on a 3rd down and 19 from midfield. That one play took the Redskins from a 27% chance of winning the game to a 68% chance (+0.41 Win Probability Added) according to league averages. It also put Grossman into positive WPA and EPA territory for the season.

A few weeks ago I wrote about ‘high-variance’ strategies, and a deep passing attack certainly qualifies. Most offenses, including the Redskins, are too reluctant to air out deep pass attempts, fearing the waste of a down, or worse, an interception.

A risk-averse mindset is echoed in an old football saying attributed to Texas coach Darrell Royal and still often repeated: “Three things can happen when you throw the football, and two of them are bad.” Coaches tend to classify outcomes as good things and bad things, and then count them up. I’m sure coaches obviously know that interceptions are a lot worse than an incompletion or a one-yard stuff. But they don’t accurately account for the payoff and frequency of each possible outcome, and this is reflected in how infrequently they call for deep bombs.

There are a few more than three things that can happen when you throw the football deep, and most of them are good. To the obvious three outcomes of complete, incomplete, and interception, we should add pass interference, defensive holding, and illegal contact. Coach Royal wasn’t coaching in the flag-happy NFL of 2011.

Since 2006, deep passes in the NFL (classified as deeper than 15 yards through the air) drew defensive pass interference calls at a 2.8 percent rate. Defensive holding was called on slightly under 1 percent and illegal contact was called on slightly over another 1 percent. One of the three penalties were called almost 5 percent of the time. The NFL doesn’t classify passes any deeper than 15 yds, but I would expect those rates to be even higher for very deep pass attempts.

On the other hand, there’s offensive pass interference. But that’s has been called on only 0.7 percent of deep attempts since 2006. Defensive pass interference is four times more common than its offensive counterpart, and defensive passing penalties are seven times more common. The consequences are asymmetric, as well. Defensive interference is a spot foul, while offensive interference is a 10-yard penalty and a replay of the down.

A perfect example of the effects of pass interference was Josh Wilson’s highly questionable pass interference call on a Tavaris Jackson attempt to Mike Williams. The (non-)play will never show up in conventional passing statistics, but it still resulted in a 44-yard gain and ultimately led to a go-ahead Seahawks field goal.

Deep attempts are naturally completed less frequently than shorter attempts. According to the NFL’s classification, deep passes are only completed 45 percent of the time, and the type of very deep attempts I’m discussing are completed even less frequently. It’s understandable for offensive coordinators to shy away from low-probability calls, but what are the other options? A typical 2-yard run into a brick wall, setting up a 2nd and 8 or 3rd and 4?

The interception rate is high for deep attempts—7 percent. But that’s misleading because so many deep attempts are the product of end-game desperation or end-of-half Hail Mary-style plays. Interceptions on 40-yard pass attempts aren’t as costly as they seem. More often than not, the ultimate effect is not much different than a fairly good punt. Third down interceptions are not as costly as first down interceptions because a punt is that much more likely. Thanks in part to an alert tackle by Santana Moss, Grossman’s interception in the third quarter was really no worse than a 3-and-out followed by a 37-yard punt. Not good, but not exactly cataclysmic either.

All things considered, with one exception I’ll discuss below, deep pass attempts net just over 0.4 EPA per attempt, while short attempts are only slightly better than break-even at 0.01 EPA. I realize those numbers are abstract, but they mean that deep attempts are well-worth the risks.

The one consideration I haven’t mentioned is sacks. Sacks are obviously more likely on deep attempts, but it’s impossible to measure how much more likely. When a pass play ends in a sack, we never know what the primary route or read was supposed to be. However, even if we assumed all sacks were the product of a deep pass play, the numbers still favor airing it out.

An additional benefit of deep passing is that defenses are forced to adjust, which opens up intermediate passing routes and the running game too. Force the safeties to defend 80 yards of turf rather than just 20.

A vertical passing game is easier said than done. The quarterback needs extra time and a receiver needs to generate some separation. Pass protection for dedicated deep attempts can be schemed with play-action roll-outs or extra blockers. And receiver separation may not be so important, as the pass interference numbers suggest. If the coverage isn’t what the quarterback expected, there’s very little to be lost by putting a little extra air on the ball and angling the pass out of bounds. Plus, there’s something to be said for putting the ball and letting your better players make plays. And if it’s third down, let it fly and don’t fear the interception.

On the go-ahead touchdown pass, Anthony Armstrong had very little separation when Grossman released the ball, and even less when he caught it. There was safety help in coverage as well, but Armstrong made the play. And to put a bow on things, defensive pass interference was called on Seahawks cornerback Brandon Browner.

Offenses have to take risks to score points, and Sunday’s win showed why. When it comes to defending deep, vertical passes, the deck of cards is stacked against defenses. The rewards of deep vertical passes exceed the risks and can be critically important.