The Ravens’ Fascinating Choice – No Not That One

Lamar Jackson, Baltimore Ravens
Lamar Jackson, Baltimore Ravens
Photo: USA Today Sports Images

NFL Week 13 - The most talked-about play of the weekend occurred at the end of a tight matchup between the Ravens and Steelers. Trailing by one point, John Harbaugh decided to put it all on the line and attempt a two-point conversion for the win. This was certainly an unconventional approach that was sure to elicit criticism if it failed, and it did.

Before we get into the criteria around the PAT strategy, it is worth examining several impactful moments that preceded the Ravens’ final touchdown drive. With 2:45 remaining in the game, the Steelers were trailing 13-12 with a first-and-10 at the Ravens’ 21-yard line. The Ravens still had all three timeouts but made the questionable decision not to call their first one after an 8-yard gain by Benny Snell, Jr. The Steelers happily responded by allowing the clock to run all the way down to the two-minute warning. The Ravens effectively traded 40 seconds of game clock for one time out, which resulted in a -5.2% reduction in Game-Winning Chance (GWC). It is nice to retain timeouts for a potential game-winning drive, but there was a clear and present value to be extracted by stopping the clock at 2:40.

Coming out of the two-minute-warning, the Steelers were about to snap the ball with second-and-2 at the 13-yard line when Ravens safety Chuck Clark charged across the line for an obvious, and somewhat suspicious, offside penalty. Assuming it was intentional, this was a fascinating and brilliant decision by the Ravens. Our analysis reveals that by proactively altering the game state from second-and-2 on the 13-yard line to first-and-goal on the 8-yard line, the Ravens gained 5.0% in GWC. Removing the first down potential served to reduce the amount of time the Steelers could burn before scoring. The Ravens would be indifferent on an immediate touchdown but would protect significant seconds if they could hold the Steelers to a field goal by using all their remaining timeouts.

Despite this clever strategy by the Ravens, the Steelers managed to score a touchdown along with a two-point conversion to take a 7-point lead. On the ensuing kickoff, the ball took a dead bounce that nearly pinned the Ravens inside their 5-yard line but eventually found its way out of bounds. The resulting penalty allowed the Ravens to begin their final drive on the 40-yard line rather than the 25-yard line, which boosted their GWC by 4.2%. With the benefit of the advanced field position, Lamar Jackson managed to lead the Ravens down field for a touchdown with just 12 seconds left in the game. This is when Harbaugh surprised everyone by going for the win with a two-point attempt. Let’s look at the major considerations that might have factored into this decision.

  • A successful two-point conversion effectively ends the game
  • The NFL average conversion rate since 2000 is 48%.
  • Since 2018, there has been an uptick in success rate to 50.1% although lacking in enough data to draw conclusions on statistical significance.
  • The Ravens were a pre-game favorite on the road and according to a custom simulation would be slightly favored in OT by a slim 52% margin.
  • Harbaugh was concerned about the absence of Marlon Humphrey at cornerback and how this might affect their pass defense in overtime.
  • Justin Tucker’s kicking prowess could be an asset in overtime. He also provided the Ravens with a non-zero probability of winning the game after a failed two-point conversion with 12 seconds for an on-side kick recovery and long field goal attempt.

Everyone is looking for a definitive answer to whether the Ravens made a mistake, and it comes down to a simple comparison of their overtime GWC vs. their two-point conversion success rate. The EdjSports model leans toward kicking the PAT and attempting to win in overtime, but it is very dependent on the underlying assumptions. If we simply look at NFL historical averages for two-point success rates and assume the game would be close to pick-em in overtime, it is a very close decision but still leans toward kicking the PAT. It was an exciting decision for sure, but it was far from being the most innovative. That happened at the two-minute mark.


26 comments, Last at 10 Dec 2021, 3:07am

#1 by jheidelberg // Dec 06, 2021 - 10:05pm

Simply decline the offside penalty, it was so obvious what the Ravens were doing and the decline was obvious as well.

In the article you mention assuming this was intentional, this was clearly intentional.

Points: 0

#2 by Bobman // Dec 07, 2021 - 2:52am

I am pretty sure there are some penalties you can't decline, and I suspect this falls into that category.  If Pitt could decline, then the Ravens could charge over the line every time the Steelers try to line up, and the Steelers would keep declining, and so on for the next 20 minutes until a ref goes postal, making a mockery of the game.  A mockery! I say.  And a bloodbath.

EDIT: My snarky tone and reference to a murderous ref turned out to be anticipated by the NFL. I did a little research.  A team trying to infinite-loop penalties like that would be assessed an unsportsmanlike conduct penalty (and probably fined), with a potential forfeit warning from the refs if it continued. I thought I recalled a non-declinable penalty, but that seems to be mistaken.  (One instance you can't decline is when there are offsetting penalties or one by each team--one team can't decline apparently.) 

Points: 0

#5 by jheidelberg // Dec 07, 2021 - 7:29am

Without the forfeit it would be an endless loop penalty because the Steelers should decline the unsportsmanlike conduct penalty as well.

Points: 0

#25 by OmahaChiefs13 // Dec 08, 2021 - 1:06pm

Every penalty can be declined unless specifically stated otherwise, and neither encroachment nor offsides have that note. I thought it was possible that encroachment, as a pre-play, dead-ball penalty fell into a non-declinable bucket, but it appears it does not:

Mechanically, pre-snap penalties are usually enforced without consulting the coach for an accept/decline, but Tomlin should have been able to get the ref's attention and decline it if he'd wanted to.

As to whether he should have or not, there's a case to be made that taking the penalty and gaining a new set of downs was advantageous. The Ravens could have stuffed Pittsburg twice and forced the FG attempt after only 2 plays, both of which would have been covered by Ravens timeouts. By picking up the first down, they gave themselves the opportunity to use up both timeouts AND run the clock on the next 3rd down.

That's dicey...given the distance, I'm pretty sure I'd treat the first down as an eventual given, and any gain provided by the new set of downs was immediately given up by scoring on 1st and hindsight, it's 100% the wrong decision, and in the moment, probably 85% the wrong decision. But you can kind of see the thought process if you squint.

Points: 0

#3 by DraftMan // Dec 07, 2021 - 6:38am

The Packers essentially did the same thing in the playoffs last year (the "Who wanted to kick that field goal?" game). After deciding that they'd rather try to stop Tom Brady and get a drive for the win, instead of going for the touchdown to tie right then, the first play of the Bucs' clock-killing drive gained 9 yards into the 2-minute warning, which was followed by intentional offsides to get a reset to 1st and 10 without consuming any clock. Of course, they couldn't stop the offense from picking up 10 more yards anyway, but letting them pick up 2nd and 1 normally would have fared even worse.

Points: 0

#4 by Raiderfan // Dec 07, 2021 - 6:39am

Apparently your model doesn’t take into account factors such as health.  Post game, Harbaugh stated they went for it because he was “out of cornerbacks” due to injuries.

Points: 0

#9 by takeleavebelieve // Dec 07, 2021 - 9:19am

In reply to by Raiderfan

If you have reason to believe that your team is significantly different in some way, then you can adjust the model. I’m sure the Ravens have in-house predictive models tweaked to their specific roster construction, whereas a writer than covers the entire NFL would find no real value in a Ravens-specific prediction when writing about the other 31 teams. 

Points: 0

#10 by RickD // Dec 07, 2021 - 10:23am

I doubt the Ravens have an available model for every combination of available players.  There are way too many combinations for such a task to be feasible. 

And I doubt they need a model to tell them "going on defense without CBs is a bad idea."  

Points: 0

#13 by Eddo // Dec 07, 2021 - 10:51am

I doubt they have an EdjSports style model, but that doesn't mean they don't have something.  Their models will be simpler, maybe even just heuristics, e.g. "When a starting CB is hurt, play more aggressively on offense."

Points: 0

#14 by takeleavebelieve // Dec 07, 2021 - 10:55am

I’m sure they understand the per-play value of cornerbacks in their scheme, which can then be adjusted for personnel and matchup. 

Points: 0

#6 by JMM // Dec 07, 2021 - 8:46am

When the Steelers got their go-ahead TD in the 4th, they also went for 2. They threw a pass to their tight end in the flat. Their decision to go for 2 was more straight forward but demonstrates the reflection these two teams are of each other in many respects.

Points: 0

#7 by DIVISION // Dec 07, 2021 - 8:54am

I understand his rationale and Harbaugh knows his team better than I do.

As a bettor, I would have liked the Ravens' chances in OT due to having the equalizer in Tucker.

Going for 2 made it easier for Pittsburgh in a sense because the Ravens were neutralized on offense all day.

I also don't trust and wouldn't trust Lamar Jackson with a game winning throw.

He threw a fastball when he needed a soft lob.

That's essentially why the Ravens won't win anything with Jackson at QB. 


Points: 0

#8 by Lost Ti-Cats Fan // Dec 07, 2021 - 9:00am

In addition to disliking their odds in OT (the no DBs left comment), another factor no doubt is that the Ravens likely thought they had a 2-point play that would succeed more than 50% of the time.

And I daresay they were right.  Looking at how that play unfolded, although it didn't succeed this time, it likely did succeed more than half the time across the multiverse.

Points: 0

#11 by RickD // Dec 07, 2021 - 10:25am

Given how the play unfolded, I would not agree that it was a great play call.  They let TJ Watt come in unblocked and relied on Lamar's ability to throw a pass by him. 

I think that's a good play call almost all of the time.  But it wasn't good in this specific situation because the rusher was TJ Watt.

Points: 0

#12 by Eddo // Dec 07, 2021 - 10:50am

I would also add that it's a good play call in general because it has the opportunity to go for a big gain (depending on the read, like in this case, when the DE rushes the passer all-out), but less so when you're capped at a two-yard gain.

Points: 0

#15 by Pat // Dec 07, 2021 - 11:25am

The main reason I don't like the 2-point (except for the fact that he was worried about injuries, which is totally fine) is that it removes the very real possibility of a tie, and a tie for the Ravens is more valuable than half a win, in terms of playoff seeding - it keeps them out of a tiebreaker situation (which they could lose to the Bengals and Steelers on now) since no one else will have 1 tie.

At 8-3, the Ravens were in the #1 slot in the playoffs (72.7% vs New England's best-possible value of 69.2%). When you're in the #1 slot, losses are more than twice as bad as a tie, because you fall back closer to more teams. A tie would've kept them at the #1 spot - a loss puts them in a 3-way tie for the #2 spot.

Ties are way underrated.

edit: I should also point out that since the loss took them out of the #1 spot, and that #1 seed is vastly more valuable than the others in terms of Super Bowl percentage (due to the bye) in terms of the Super Bowl win percentage the tie was almost certainly way more than half a win.

Points: 0

#16 by Meaningful gam… // Dec 07, 2021 - 11:54am

1. A tie is a very real possibility, with (as you pointed out) more than half the value of a win.

2. Your kicker is significantly better than their kicker.

3. Pittsburgh plays on Thursday. Putting in an extra 10 minutes is more deleterious to your division rival than to you, who has a standard week.

4. Pittsburgh being involved in the only 2 tie games of the year would be hilarious.

Points: 0

#17 by Pat // Dec 07, 2021 - 12:00pm

Plus giving a team 2 ties actually removes a lot of the value of a tie. Ties are often valuable because normally if you lose, in playoff seedings there are more teams below you than in front of you, so a loss often drops you down into a race for a spot. A tie drops you down, sure, but you're still ahead of all of the other teams one loss down. Handing a team another tie just converts into a loss, removing all of that benefit.

But I totally agree, Pittsburgh being the first team with 2 ties since the NFL introduced sudden death would be hilarious.

Points: 0

#20 by DIVISION // Dec 07, 2021 - 3:10pm

In a game like that, OT favors the team with the better kicker.

I think Baltimore likely either wins by FG or loses by FG in that scenario.

No tie.

Points: 0

#21 by Pat // Dec 07, 2021 - 3:14pm

If the Ravens win the toss, they can't win by a FG straight off. And I could easily see the two teams trading field goals at that point. And then with only 10 minutes total, the teams could easily run out of time. Both the teams have a playstyle that bleeds time like crazy.

Points: 0

#24 by Frank Frigo // Dec 08, 2021 - 11:35am

It is very interesting to think about the value of ties, especially as we get into the later phases of the season.   Our pre-game simulations  had this game going to OT about 6% of the time and about 7% of those games ending in a tie.  

Points: 0

#18 by Cythammer // Dec 07, 2021 - 1:55pm

"A successful two-point conversion effectively ends the game."


I don't think this is true. The TD was scored with 12 seconds left. However, the Steelers still had a timeout left. With the amount of time left Pittsburgh would have a chance at hitting a thirty or forty yard completion and instantly calling a timeout. Not a great chance, but a chance.

However, once Tomlin used his last timeout prior to the two-point attempt, the situation changed.

Points: 0

#19 by JacqueShellacque // Dec 07, 2021 - 2:51pm

Everyone is looking for a definitive answer to whether the Ravens made a mistake, and it comes down to a simple comparison of their overtime GWC vs. their two-point conversion success rate. The EdjSports model leans toward kicking the PAT and attempting to win in overtime

Everyone? By the standards previously applied in this column, only the numbers matter.

Points: 0

#22 by young curmudgeon // Dec 07, 2021 - 3:33pm

Much of this discussion is being based on the advantage Baltimore has in field goal kicking.  Meaningful gam...states that "your kicker is significantly better than their kicker."  Division says "OT favors the team with the better kicker" and implies that Baltimore is likely to win.  Stipulating that Justin Tucker is probably the best field goal kicker ever, I'd still like to point out that the advantage is not as massive as some would assume.  Tucker is 25 of 27 on FG this season.  Boswell is 24 of 25.  Tucker has not missed an extra point, Boswell has missed two.  One of those took place in this very game, but does that miss predict a trend or (unscientifically) "use up" his one mistake and make it more likely that he won't miss again in the same game?   Boswell would be kicking on his home field, a historically difficult place in which to kick.  If the game comes down to a field goal and Tucker is kicking, Baltimore will win.  If the game comes down to a field goal and Boswell is kicking, Pittsburgh will win.  Harbaugh evidently felt like the injuries to his secondary and the way his team played defense in the fourth quarter outweighed any slight advantage Tucker gives him.

Points: 0

#23 by ImNewAroundThe… // Dec 07, 2021 - 5:35pm

said it in another thread but in a scenario like this before...Tucker missed. 

Also people say that OT favors Baltimore this season because they are 2-1. But the wins were at home and the loss...was @ LV. And the Ravens were on the road again.

I respect Harbaugh for having the cojones to try and win there instead wasting my time with the NFLs awful OT.

Points: 0

#26 by Subrata Sircar // Dec 10, 2021 - 3:07am

... you think your chances of converting (plus a generous 1% for the onside recovery + FG), are lower than your chances of winning in OT.  Thus, usually the underdog is the one going for two here.

That said, I do not believe there's more than a percent either way, or more than enough to both be justified either way by slightly differing assumptions, and be totally screwed over when someone's grip slips off the ball at the wrong point.


Points: 0

Save 10%
& Support Frank
Support Football Outsiders' independent media and . Use promo code WRITERS to save 10% on any FO+ membership and give half the cost of your membership to tip Frank.