John Harbaugh is a Super Bowl-winning head coach. He might represent the new archetype for owners when it comes to hiring a head coach. He outcoached his brother in Super Bowl XLVII. But that doesn’t mean his fourth down decisions on Thursday Night were above criticism.

1) Punting is not the way to beat Manning

Facing 4th and 5 from the Broncos 40-yard line, Harbaugh elected to punt up 14-7 with 8 minutes left in the second quarter. Last year, I highlighted one of the most difficult fourth down decisions coaches have to make: 4th-and-7 from between the 34- and 38-yard lines. In the thin air of Denver and with strong-legged Justin Tucker, we can safely include this scenario in that definition of No Man’s Land. Facing 4th-and-5 is a lot easier than 4th-and-7, so going for it would have been my preferred choice. The Ravens elected to punt, but let’s consider the other two options.

According to Advanced NFL Stats, going for it would have been the right call if the Ravens had a 30% chance of succeeding ^[1]I’m splitting the difference here, between the 25% minimum threshold needed to make going for it the correct call according to maximizing the team’s win probability and a 34% threshold … Continue reading, and Baltimore needed a ~45% chance of making the field goal to make kicking superior to punting. Even in Denver, a 57-yarder is a tough proposition ^[2]For what it’s worth, since 2000, kickers are three for five when attempting field goals at the line of scrimmage is the 39, 40, or 41-yard line., so I would have advised Harbaugh to keep his offense on the field. That’s because on average, teams convert 49% ^[3]That’s according to Burke. For what it’s worth, teams converted 42% of all 3rd-and-5 attempts last year and 46% in 2011. of all 4th-and-5 attempts, and playing aggressively is even more appropriate when you are the underdog. Of course, Wes Welker muffed the punt and the Ravens scored a touchdown, but let’s not substitute process for outcome. While we’re on the topic, John Fox made an even less defensible decision in the first quarter, punting on 4th and 4 from the Ravens 39-yard line. Whole articles could be written about poor Fox decisions, so let’s just say the bar is set higher for Harbaugh.

2) Meek as can be: 4th and 1

Failing to go for it in No Man’s Land is a forgivable error, and one not worth bagging on Harbaugh since so many other coaches would do the same. But one of his decisions in the third quarter was too embarrassing to cut him any slack. The brothers Harbaugh are famous for their competitive nature, but Harbaugh coached as conservatively as possible here. With 20 minutes left in the game, Baltimore trailed 35-17. The Ravens faced 4th and 1 from their own 29-yard line. If the Ravens punt, they have almost no chance of winning the game. But Harbaugh sent out Sam Koch, and Baltimore picked up 45 yards of field position.

That means the Broncos faced 1st-and-10 at their own 26-yard line after the punt. Do you know what a team’s odds of winning the game are when they have the ball on 1st-and-10 from their own 26, up 17, with 20 minutes left in the game? Ninety-seven percent is what Brian Burke’s data tell us. According to Pro-Football-Reference’s new Win Probability Charts, the Broncos had a 99.7% chance of winning the game after Baltimore punted. Why the difference? PFR’s graphs incorporate the point spread, which shows why Denver, as an 8-point favorite, had a 72% chance of winning before the start of the game. Take a look at PFR’s win probability chart, which shows Denver’s odds of winning the game at every second of the game:

If you punt facing 4th-and-1, down by 18 points, in a game where you’re an 8-point underdog, you might as well pull your starters. According to Burke’s model, if you punt, you have a 3% chance of winning. If you go for it and fail, your odds drop all the way to… two percent. Now, if Baltimore went for it and converted, the Ravens’ odds would have jump to five percent. Now consider that teams convert 74% of the time on 4th-and-1 (also from Burke), and that when down 18 you want to take as many risks as possible. There is simply no justification for punting, and Harbaugh should have been ashamed to have coached so meekly in that situation. ^[4]There’s maybe one justification: if you have decided that the game is lost, and you don’t want your team to get completely blown out because you think the emotional baggage would be so … Continue reading

3) Kicking a field goal down 18

Let’s move to the final fourth down error by Harbaugh. With 5:33 remaining and two timeouts, Baltimore faced a 4th-and-4 from the Denver 12-yard line, trailing by 18 points. Theoretically, the Ravens could kick a field goal, and hope to score two more touchdowns, convert on a two-point attempt, and then win in overtime. But keep in mind both how difficult that would be and the hidden coin-flip scenarios that quarter their likelihood of winning. The odds are low that Baltimore would (1) make the field goal and then, in the course of five minutes, (2) stop Denver from scoring and score a touchdown, and (3) stop Denver again from scoring and score another touchdown. But whatever the odds of that happening are, you need to divide that number by four to represent Baltimore’s probability of winning the game the moment Harbaugh sent out Tucker. That’s because even after doing 1, 2, and 3, Baltimore still needs to convert a two-point conversion and win in overtime, and there’s only a 25% chance the Ravens would do both of those things.

The smart decision would have been to go for the touchdown. We can’t even use Burke’s model here, because the odds of the Ravens winning the game down by 18 with 5:33 to go are simply too low. But this is the same issue that a team down by 11 with 5:33 would have to consider — except in that case, it makes slightly more sense to kick the field goal.

For Burke, the break-even success rate to make going for it the correct decision when down 11 with 5:33 left at your opponent’s 12-yard line is 29%. That’s because a field goal gives you a 7% win probability, a successful conversion gives you a 20% win probability, and a miss gives you a 1% win probability. Since the success rate on 4th-and-4 is 48%, going for it is the easy and obvious call. The win probabilities are all adjusted downwards when trailing by 18 instead of 11, but the same principles apply (except logic would even more strongly favor going for it when down by 18 instead of 11). You simply are asking your team to do too many things in too short of a time period by attempting a field goal. Joe Flacco isn’t Peyton Manning, but if you don’t have faith in him to convert a 4th-and-4, (1) why give him $120M, and (2) how much faith do you have in him leading two touchdown drives in about three minutes of game time (at best)? Harbaugh coached as if he didn’t want the game to end instead of trying to maximize his team’s odds of winning.

The way Manning and the Broncos offense was playing, no team would have beaten them on Thursday Night, and Harbaugh’s decisions certainly didn’t cost Baltimore the game. But punting on 4th-and-1 when trailing by 18 is never acceptable, and coaching to keep hope alive should never be more important than coaching to maximize your team’s odds of winning.