Harbaugh's 4th-down decision was sound, even if it didn't work out

It was an all-in move and it didn't work.

The decision from Sunday's action that most rankled fans and studio pundits - and even Ravens cornerback Marcus Peters - was Baltimore head coach John Harbaugh's choice to go for it on fourth-and-2 at Buffalo's 2-yard line with 4:15 left in a tie game. Baltimore was at the end of a 13-play drive that took nine minutes off the clock.

A field goal would have given the Ravens the lead but would also have given Buffalo the ball back with more than four minutes left. The Bills fell behind 20-3 with 3:39 left in the first half but scored the last 17 points in the game.

Fourth-and-2 is not a tap-in. Data from Ben Baldwin's fourth-down Twitter bot reported the chances of going for it and succeeding was 47%. A field goal from there, closer than an extra point, had a 99% success probability.

The choice, on the face of it, ought to be clear: Take the three-point lead and let the Ravens defense try to defend it. That's where the second set of probabilities comes in.

Both Baldwin's Twitter bot and the NFL's own NextGenStats Twitter account said the preferred choice was to go for it. Both data models said the improvement to be gained in the Ravens' probability of winning was between 1.7 (NextGen) and 2.1 percentage points (Baldwin rounded to two points in the tweet).

But that probability improvement splits the difference between succeeding and failing based on the relative success rate. The real place to look is in comparing the actual probability if the choice succeeds and if it fails.

According to Pro Football Reference's win probability calculator, the Ravens' probability of winning the game stood at 71.3% after third down.

Here are the options Harbaugh was weighing. Even if he didn't have the exact numbers in front of him, he likely knew the general probabilities:

• A field goal gives you the lead but changes your probability of winning to 63%.

• A touchdown gives you an 83% chance of winning.

• Failing to score the touchdown decreases your probability to 49%.

Poker players and NFL head coaches who think about the odds would choose to go for it. They would want their money in the pot with an 83% chance of winning. That would force Buffalo to go 75 yards (assuming a touchback) just to tie - which is what Harbaugh said after the game:

Even if things go against them and the play fails, it's still a coin flip to win.

As we know, Harbaugh lost that coin flip.

Many would choose the field goal and the 63% probability. Bill Cowher and Nate Burleson said as much at halftime of the CBS late game, which is fine, although it would be nice if they could express their decision in terms that don't seem backward and out of touch with modern decision-making.

That 63% probability comes from analyzing similar situations from across hundreds of games - not all of which had Josh Allen at quarterback. Sunday was the 12th game-winning drive in his 65 games.

Harbaugh wanted the 83% probability. He thought he had a good hand with Lamar Jackson, the other offensive playmakers, and two yards to go. He bet that the Ravens' success rate in that situation was actually better than the 47% of an average team. Jackson had only thrown two red-zone interceptions in his career.

Harbaugh has shown a propensity for making tough fourth-down decisions, but it's not like he's a freewheeling gambler. In 2021, he went for it on fourth down 27 times, which ranked 13th. The Ravens were successful 18 times, a 66.7% success rate that was tied for first with the Chiefs. This season, Harbaugh has gone for it only five times in four games, although the Ravens have only been successful twice.

If you can fault Harbaugh, perhaps it's for the fact he failed to account for the reality of Baltimore's defense, which has allowed 1,700 yards in four games (425 per game, ranked 30th in the league). That's better than only Seattle and Detroit, who collected more than 500 yards against each other Sunday.

But the Ravens defense had held Allen to a 68.4 QB rating for the day on a season-low 213 passing yards. Allen also had to be the Bills' leading rusher, gaining 70 yards on 11 carries.

Sometimes you go all-in with what you believe is the best hand and you still lose.

Guy Spurrier is theScore's features editor.