The numbers behind John Harbaugh’s aggression

Baltimore went into a 2-0 showdown with the Kansas City Chiefs and turned what we have known of the NFL aggression knob up to 11. They went for fourth-down conversions twice in their opening three possessions. They missed three separate two-point conversions. 

This is certainly an odd shift for a team that cut its Super Bowl teeth as a defense-first, conservative team. I understand if you’re feeling awkward about watching the Ravens dive into the fourth-down and two-point conversion chase. It is often jarring to see convention be broken, and especially when the results are not instantaneously good. 

@SharpFootball
Far too many coaches play to the current score. They ignore the fact that their opponent may score again. It’s a vital factor when facing Mahomes in KC. Ravens fans: get accustomed to this. This is the right way to make decisions. NFL teams not yet on this level: get on it.

Of all the quotes that John Harbaugh gave in his press conference on the matter, the one that got quoted most was the idea that it was mathematically calculated. I want to focus on two other quotes he gave. 

It wasn’t a field position game, it was a possession game.

This is the quote that strikes me as the reason for most of Baltimore’s aggression. The Kansas City Chiefs employ Patrick Mahomes and Andy Reid. Reid is one of the five best NFL offensive minds — and I’m being conservative with that ranking. Mahomes has been so good in his first 20 starts that even if you regress his stats heavily to the mean, it is very obvious that his true talent shines. He’s perhaps the most valuable player in the NFL. Deep math on the subject will lead you to Danny Tuccitto’s idea of “true” stat rankings. Mahomes’ rank among all quarterbacks in adjusted net yards per attempt, yards per attempt, and touchdown percentage based on filling his catalog with regression, are first, first, and second. 

(You can discuss this on the BSL Board here.)

The old-fashioned way of keeping a dominant offense off the field is to control the clock. There are several reasons why controlling time of possession is beneficial in theory. But one thing controlling possession doesn’t do is deal with the fact that the opposing team can still score really fast if they’re that dangerous. I like to point time of possession aficionados to the classic case of Miami versus Prime Peyton Manning. The Dolphins controlled the clock for 45:07 and lost. 

What Harbaugh’s words get at is that the truth of the matter isn’t winning time of possession, but the total number of possessions period. In other words, you recognize that Mahomes is not going to fail often. So what do you want to do? Score on every possible possession. By going for two, you also make each possession a yes-or-no verdict. If you can get the Chiefs “behind” by a point on each of their successful possessions, then you stand a better chance of winning the game. 

If you want to look at the pure math of the exercise, I spend a lot of time with the company that now owns Football Outsiders, EdjSports. They have a metric called game-winning chance or GWC that essentially is an expectation of the odds that a team will win a game from any given score-down-yards to go-time left-timeouts template. One thing I’ve learned in my time getting comfortable with GWC this year is that it really likes going for it on fourth-and-short even when you’re backed up. 

GWC is calculated both by the post-play result and on a decision-making level. So when the Ravens went for it on fourth-and-1 at their own 34, the numbers considered the chances that they’d succeed and decided the Ravens had a 39.6% chance of winning the game if they went for it, against a 35.5% chance of winning the game if they attempted a punt. When they actually got the first down, creating a known outcome, their odds of winning went up to 42.4%, which is the highest they’d ever be for the rest of the game. 

When the Ravens went for two after their first score, the decision was fairly close: the two-point attempt was favored by a narrow amount: 54.9% GWC versus 54.8% GWC for the regular PAT. However, once they missed the two-point attempt, their GWC fell to 52.4%.

They’ll tell you I go by my gut way more than I go by the analytics, and I do. I’ll go against the analytics a lot more than I go with it, in terms of 50-50 close calls.

So with that in mind, let’s point out that the Ravens actually missed a couple of chances to boost their GWC. They punted in the second and third quarters between the 40s, once on fourth-and-4, and once on fourth-and-5. On a pure decision-making level, they would have watched their GWC go up by 4.4% if they’d gone for it on those fourth downs as well. 

I think Harbaugh legitimately felt uncomfortable with this game plan, and I commend him for going as deep with it as he did. The Ravens are the talk of the analytics community right now and that’s despite the fact that they didn’t go for it on every numerically correct opportunity. As he noted, they’re not in the game without some of those fourth-down calls. 

Sunday’s game against the Chiefs taught us a lot about the Ravens. It showed us that the offense wasn’t infallible. It showed us that Lamar Jackson needs to work on his big blitz reaction skills, and that he’s still a young quarterback who is going to be flummoxed on occasion. The defense made a few coverage busts and Mahomes is … well, Mahomes is going to be Mahomes. Nobody is doing anything about that.

But it also showed a lot of promise for the Ravens in that most of what could go wrong did, and they still kept it to a five-point game, on the road, against an offense that good. With a little more well-considered play design on some of those fourth-down playcalls and two-point conversions, the Chiefs might have been in a real endgame pickle despite averaging 7.9 yards per play. 

I think if you’re a Ravens fan you have to leave this game feeling optimistic about their spot in the pecking order in the AFC. The rest of their division is falling apart. They played to win and almost did despite not playing all that well. This is as promising a start as they could have hoped for.