A Curious Decision in Baltimore
NFL Week 15 - Trailing by two touchdowns late in the game against the Packers, the Ravens managed to orchestrate an impressive comeback under the direction of backup quarterback Tyler Huntley. With 4:47 remaining in the game, the Ravens had just scored a touchdown to reduce the deficit to eight points and were facing a PAT decision. As we analyzed extensively in our Week 6 Risky Business column, it is clearly correct to attempt a two-point conversion in this situation, with few exceptions. Somewhat surprisingly, Harbaugh chose to deploy the services of Justin Tucker to pull the Ravens within seven points of the Packers.
Harbaugh is one of the most analytically sound coaches in the NFL and certainly was aware of the mathematical merits of going for two in this situation. At that moment, we speculated that the presence of Huntley may have been a key factor in his conservative decision. It now appeared Harbaugh would be playing for overtime if the Ravens were fortunate enough to stop the Packers on the ensuing possession and to score another touchdown. Well, apparently that wasn’t the case.
Inexplicably, the Ravens (unsuccessfully) went for the win after scoring the second touchdown, after shunning that option moments earlier. If you know you are going for the win in regulation, why wouldn’t you attempt the two-point conversion at the first opportunity? With a success you can simply kick the PAT on the second touchdown for the win, and with a failure, you still can go for two and possibly get into overtime. It is also worth noting that the Packers still had 42 seconds and one timeout for Rodgers to move his team into field goal range. However, this scenario exists whether the Ravens tie or take the lead. There could also be an argument that the Packers will be more aggressive if trailing by one than if tied. For instance, they would attempt a long fourth down in their own territory if trailing but run out the clock if tied.
There are three separate decisions to examine:
1. Is it still correct to go for two after the first touchdown with a backup quarterback?
As detailed in our prior analysis, even if the Ravens have only a 35% chance on the two-point conversion and are 45% to win in overtime (as our simulations suggest), it is still clearly correct to go for two.
2. If you know you are going for the win in regulation, is there any reason not to go for two after the first touchdown?
This doesn’t require any math. Unless there was something Harbaugh saw on the second opportunity that wasn’t present on the first one, it simply defies logic.
3. What about the independent decision to go for the win with the two-point conversion after the second touchdown?
EdjSports' custom simulations used a 13% reduction in team passing DVOA and an 8% reduction in team rushing DVOA for Huntley in place of Jackson. With these assumptions, we show it to be a close decision with a slight lean (~2% GWC) toward kicking the PAT.
Comments
52 comments, Last at 22 Dec 2021, 12:00pm
#1 by rpwong // Dec 20, 2021 - 2:41pm
2. If you know you are going for the win in regulation, is there any reason not to go for two after the first touchdown?
I don't think you can assume that Harbaugh knew he was going for the win in regulation at this point. Kicking the PAT pushed the decision down the line so that he could see how things played out and wait on new information (such as the health/stamina of his players).
It may not have been the correct decision by the numbers, but I can understand him wanting to delay it due to the state of his depleted roster.
#2 by Mike Y // Dec 20, 2021 - 3:19pm
The way I read that was that Harbaugh may have decided that he didn't want to go to OT at all. If you miss the 2pt attempt after the first TD, then your only choice is to go for 2 after the 2nd TD to force OT. By kicking the XP after the first touchdown, it still allowed him to go for the win after the 2nd touchdown. So, if he decided that he did not want to play OT at all, then this was the proper choice.
However, Huntley was playing well, so I don't understand his aberration to OT.
#3 by ImNewAroundThe… // Dec 20, 2021 - 3:42pm
He was playing well but Rodgers was playing well...er. And his team was decimated, prolonging the game against a team like GB...not exactly something you want to do with a bunch of backups with a noon kickoff next week.
#9 by jds // Dec 20, 2021 - 4:35pm
This is interesting. Perhaps Harbaugh knew he was going for the win BUT, didn't want to tell LaFleur he was doing this. Wanted him to think he was playing for OT, to influence LaFleur playcalling. And that is the interesting part. I can see GB changing their playcalling if the Ravens got the first 2 pointer.
#16 by ImNewAroundThe… // Dec 20, 2021 - 7:04pm
They know they probably can't make any noise in the playoffs. No point in putting more players at risk. They were perfectly fine with the result whether they lost or won.
Bless him for not wasting our time with another coin flip 🙏
#12 by Wifan6562 // Dec 20, 2021 - 6:34pm
Are you really suggesting that a good NFL head coach, in the playoff hunt, intentionally reduced his chance to win or tie in exchange for being able to avoid the chance of playing an extra (up to) 10 minutes of football? That doesn’t make any sense to me.
I think he just wasn’t prepared enough for this potential decision and he defaulted to “old school thinking”. If you haven’t thought through these decisions, it’s impossible to do so in the moment, no matter how informed you are on the topic. It was pretty surprising though, given his history.
#15 by BJR // Dec 20, 2021 - 6:53pm
Yes, on reflection this was my reading too. Harbaugh had no intention of playing overtime with his beaten up roster, and leaving the 2 point try until the end therefore ruled it out.
I think what he (and perhaps the Ravens analysts) missed was then leaving enough time on the clock for a Green Bay response, and how being down - rather than tied - might influence their approach.
#38 by Wifan6562 // Dec 21, 2021 - 1:08pm
All this talk about intentionally reducing win probability to avoid overtime makes me think a re-watching of Herm Edwards is in order:
”You play to win the game. Hello! You play to win the game. You don’t play to just play it. That’s the great thing about sports. You play to win…. When you start telling me it doesn’t matter, then retire. Get out.”
I find it unfathomable that Harbaugh would actually intentionally do that. Hello! They’re in a playoff race! They need to win!
#13 by RickD // Dec 20, 2021 - 6:39pm
and it's wrong. Whether the 2-point conversion would win the game or not is exactly the information Harbaugh needs, and he needs it as early as possible. Had he gone for it on the first TD, one of two things would happen:
a) he'd fail, in which case he'd know that a 1-point/2-point strategy was unsound. But he'd still have another shot at the 2-point try after the 2nd TD, to tie the game
b) he'd make it, and then he'd be in the far more secure position of being down only 6 points.
Avoiding the early 2-point conversion is like avoiding a doctor visit because you don't want to find out if you're sick.
#4 by JacqueShellacque // Dec 20, 2021 - 3:43pm
...until a scenario like this (roughly) comes up in a playoff game, maybe even a CCG or SB: play for OT or take your best shot with an opportunity you might not get in OT, from 2 yards away, with the ball, and your most skilled players (for almost all teams) on the field, right now. Regardless of how it comes about and what happens, it'll be one of the great moments in NFL history, up there with the Ice Bowl 4th and 1. Would I want my team to do that? Even though my gut tells me they should (I got #17 JA, all 6'5" 240 lbs of him), there's still a part of me that fears the buyer's remorse of an unsuccessful attempt. Unlike most people who call themselves analytics types, I'm not sure it's entirely irrational to feel that way. I just don't think you can say 'the numbers supported it', I think you need an 'I trust my guys' story like Sean McDermott did after not kicking an easy FG for OT against TEN earlier this year, and failing on a 4th and short conversion. No one really blamed him. Interestingly though he faced a similar choice (albeit from further away) in Tampa last week and chose not to go for the win. I wonder if what happened earlier in the season crossed his mind? In fact I'd be stunned if it didn't. Homo sapiens can't live by numbers alone. Avoiding outcome and hindsight biases is good, but that's not the same as living peacefully with the outcome, those are different things. I'm a Taleb groupie, I think humans have an innate sense of sample paths and a suspicion that one needs to avoid certain loss even if it comes across as a suboptimal choice. To a lot of fans, and maybe to frontoffice people and execs, Harbaugh has 'cost' his team 2 probabilistic wins this year. That could set things back.
#7 by BigRichie // Dec 20, 2021 - 4:18pm
Ice Bowl was 3rd and 1. But with 16 seconds and no time outs left, the QB sneak was a do-or-die play for the Mighty Pack.
Given the clock, Cowboy management (don't know about Coach Landry; loved that hat!) expected the Packers to pass on 3rd down.
#8 by ImNewAroundThe… // Dec 20, 2021 - 4:27pm
Uh, you must stayed off the internet but there were a ton of people that blamed Sean.
And yeah people like to hold onto hope longer. Doesn't make it the right decision. Shouldn't set back Harbaugh, should be praising...the future HOF HC that has his team multiple games >.500 despite an absolute tear of injuries.
#10 by jheidelberg // Dec 20, 2021 - 4:43pm
The front office and execs buy in to the analytics so Harbaugh has nothing to worry about.
As for the fans, many refuse to embrace analytics, but they are irrelevant.
I am waiting for a SB or playoff game to be lost by “taking the tie” and in reality taking the loss as the kicker misses the extra point. Even Tucker lost a game to the Saints a few years back in this exact situation.
i have no clue why coaches use their 92 percent kicker who is not Tucker and go for the tie. Also, when scoring early and up 6-0 go for 2 with the bottom 2/3 of kickers in the league. That day will come, I know analytics will win in the end, I have watched enough baseball to see what happens in a sport that is 1-2 decades ahead in analytics.
#18 by Lost Ti-Cats Fan // Dec 20, 2021 - 7:46pm
Hopefully the coach calls time out first to think about it, so all the talking heads on TV can demonstrate their "understanding" of analytics.
Then - more importantly - hopefully the coach decides to go for the 1-point conversion, which the kicker misses, so that none of the TV heads can admit that their prior opinions took into account that a 1-point kick does not have a 100% chance at success.
#25 by JacqueShellacque // Dec 21, 2021 - 8:36am
when Aikman used to be against anything other than FG attempts or punts on 4th down, no matter the situation. Now even he, dullest sports commentator who ever lived, almost seems to see a more rational approach to 4th downs as normal.
#11 by Steve in WI // Dec 20, 2021 - 5:30pm
Regarding the final play: assuming the math (after factoring in the specific context of this game) says that it's close to a tossup whether to go for 2 or take the XP, I don't understand why I see people suggesting that going for 2 to end the game then was clearly the right choice given how weak the Ravens defense was. Based on how easily the Ravens scored both TDs, you can easily argue that the Packers defense was the weaker of the two at that point.
The bigger issue I have is that with 42 seconds and a timeout left for Green Bay, taking a 1 point lead by no means ends the game. And I definitely believe that the Packers offense is much more likely to score if they have to as opposed to if the score is tied. I would argue that it makes going for 2 worth it only if you really think you have no chance if the game goes to overtime.
I definitely agree that he should have gone for 2 down 8.
#14 by RickD // Dec 20, 2021 - 6:41pm
Had they made the 2-point try, that would have forced the Packers to try to get a FG in the time remaining. A PAT to tie the game might have lulled the Packers into playing for overtime.
It's yet another factor to include in the computations.
#17 by Chuckc // Dec 20, 2021 - 7:46pm
I think that's a terrible line of reasoning. You can't let something outside of your control (Packers' level of effort to score) affect the decision you make first. If you go for the tie and the Packers don't try to play for overtime, you've shot yourself in the foot in the hopes that the other team does the same.
#39 by Steve in WI // Dec 21, 2021 - 1:14pm
Yeah, I think the fact that you know they'll go for it on 4th down is alone a disadvantage to being up 1 vs tied (although maybe not as much as it would be if there was 1:30 or 2:00 left on the clock rather than under a minute). It's not that I believe the Packers wouldn't try to score if the game was tied, it's just that if they're down 1 you know with absolute certainty that they are going to try to score until the final second, whereas if it's tied and you can make one or two good defensive plays there's a chance they'll give up and play for OT.
#41 by jheidelberg // Dec 21, 2021 - 1:34pm
Although forcing a team to go for 4th down at the end of a game that you are up 1 point is a disadvantage it is more than outweighed by actually having a lead. Otherwise, analytics would say that when you are down 6, score a TD late to tie, you should take a knee on the extra point. No such analytic argument has ever been made.
#24 by ammek // Dec 21, 2021 - 8:26am
If it starts to happen regularly that teams will go for 2 to win the game rather than kick the PAT to force overtime, I wonder if opposing head coaches will react accordingly (by about the year 2040, I mean).
Let's assume the Packers knew in advance that Baltimore would, if they had the chance, go for 2 if they scored at the end of the game. Would that change the Packers' defensive calls? Would it have altered their approach to their final offensive possession, knowing that there would be no overtime and if Rodgers touched the ball again it would be with only seconds to play?
There are many knock-on effects of a change in approach. Harbaugh cannot control them all, but he ought to be aware of the possibilities.
#31 by ImNewAroundThe… // Dec 21, 2021 - 11:33am
from others, apparently, take this as him having no idea what he's doing.
Clearly these people have worked in some toxic environments if their boss asking them what they should do creates such a response.
It's not like he's asking such questions on obvious play calls, it's a toss up! This is the unquantifiable stuff I'm ok with (most fan use it as a crutch though but Harbaugh uses it correctly and practices it fantastically).
#22 by jheidelberg // Dec 21, 2021 - 12:02am
Leaving the stadium on Sunday fans were cursing him saying they have had enough. The average fan is a fool. Its why I enjoy discussing things with all of you and the FO staff, intelligence in football discussion. Harbaugh started in 2008, here are records since 2010, the chart I can find online, if you go back to 2008 Harbaugh went to the AFC Championship game and won the wild card round in 2009:
Rank Team RecordWin Percentage Made Playoffs Playoff Wins Titles
1New England Patriots141-49 74.2% 10 16 3
2Green Bay Packers 126-62-2 66.8% 9 11 1
3Pittsburgh Steelers 121-67-2 64.2% 7 5 0
4Kansas City Chiefs 120-70 63.2% 8 7 1
5New Orleans Saints 119-71 62.6% 7 5 0
6Seattle Seahawks 117-71-1 62.2% 9 10 1
7Baltimore Ravens 117-73 61.6% 7 8 1
The only two teams that have been substantially better than Baltimore are NE and GB. Now lets examine the entire chart, as the Ravens are competing well with all. Lets see what makes Baltimore so different.
NE: Brady Hall of Famer
GB: Rodgers Hall of Famer
PIT: Big Ben Hall of Famer (or mighty close at worst)
KC: Cassell/Alex Smith/Mahomes Mahomes best in the game during his tenure
NO: Drew Brees Hall of Famer
SEA: Russell Wilson Hall of Famer
BAL: Joe Flacco Hall of Famer HAHAHAHAHAHAHAHAHAHA
Basically Harbaugh has accomplished with Joe Flacco what others have done with RW, Brees, Cassell/Smith/Mahomes, Roethlisberger, Rodgers, and no one can touch Belichick Brady.
163 Games of Flacco, +6 games of Flacco replacement when injured in 2015 that is 76 percent of his career games coached and 9 of the 13 playoff games that he has coached. Jackson started 49 games, 22% of games that he has coached.
Please give me another decade or two of Harbaugh in Baltimore
#27 by JMM // Dec 21, 2021 - 9:51am
Two of the coaches listed have a division competitor on the list: Harbaugh and Tomlin. So not only has Harbaugh accomplished his record with Flacco, he has done it in a division with another top 7 (record) coach in his division.
I love the Balt-Pitt rivalry games and the Harbaugh-Tomlin competition is one for the ages. I also wonder about those who call for either to be replaced and am looking forward to more of this great match-up.
#36 by jheidelberg // Dec 21, 2021 - 12:49pm
Took the 4 hour drive to Pittsburgh for the 7th or 8th time two weeks ago. As always a great game. Steelers fans are great, I am treated great always can converse wearing Ravens jersey and have fun.
If you ever want to go to a football game not involving your team, a Ravens Steelers game is a must, I recommend going in Pittsburgh, the fans are more knowledgeable and kinder in my opinion.
As for Tomlin, despite shunning analytics (he was last for 2021 on EDJ sports coaching decisions last time I looked), the man is a magician as he can take any team and get at least .500. This goes for 2019 with replacement QB’s , a washed up BIG BEN and a compilation of games in which Ben was injured including using Dixon, Leftwich, Rudolph, a washed up Vick and more.
#42 by JMM // Dec 21, 2021 - 2:34pm
I moved out of the Pitts area in the '70's. (Whoa, that was a while ago!) I did get a chance to go back for a game once. Got to see Rod Woodson and Carnell Lake play. Of course Pitt lost that game to the Browns 51 to something. It was ugly. I was there with customers from Ohio. All they did was smile. After that, I have stayed away. Not superstitious or anything, knock on wood.
Don't assume Tomlin is shunning analytics. Like all the coaches on that list, he certainly knows the analytics but isn't bound to them. He is process oriented and analytical in his approach to much of the game. Unlike some coaches, his press conferences are instructive on how he thinks. He doesn't explicitly cite analytics of play calling, but has acknowledged them after some games, explaining why he didn't do as analytics would indicate.
#30 by ImNewAroundThe… // Dec 21, 2021 - 11:28am
He's a HOFr.
I thought back in 2018, both he and McCarthy started on the hot seat, and I was thinking that, with both one year left on their contracts, that we should do a coach swap if things didn't go as planned (sadly we'll never see such creativity from the NFL).
Turns out GB would lose to the Rosen Cardinals at home and Harbaugh would help Lamar become the first unanimous MVP. Went completely different directions.
#37 by jheidelberg // Dec 21, 2021 - 12:55pm
Harbaugh would possibly have been fired had they not won the last game of the season that you referenced to make playoffs in 2018, and if so I would have gone nuts and know that the end of a great era had occurred. I would never trade Harbaugh for McCarthy, but I ll give you Jackson right now for Rodgers as long as he signs an extension with the Ravens Are you in as long as Jackson signs extension with the Packers?
#46 by ImNewAroundThe… // Dec 21, 2021 - 5:41pm
It was rookie Lamar who saved his job.
2 games below .500 the previous 3 years and no playoff appearances. One appearance (divisional) in the previous 5 years. That's why he was on the hot seat. Then opened 2018 4-5 to boot, before Lamar swung them into the playoffs. If things continued south why not swap SB winning HCs for a year to see if a change of scenery helps both? No long term commitment.
#47 by BJR // Dec 21, 2021 - 6:59pm
That's not really true; Lamar did not play well as a rookie. His passing DVOA was significantly worse than Flacco's and his rushing efficiency numbers were also dreadful (due in no small part to 10(!) fumbles).
What happened that year was Baltimore played a very soft schedule in the second half of that season, allowing them to sneak into the playoffs. But even then Lamar was bad in that playoff loss to the Chargers - I even remember it being a debate as to whether he should be benched for Flacco at some points in that game (though Harbaugh was, of course, too smart to change course like that).
All of which then made Lamar's improvement to MVP in 2019 particularly impressive. That was a superb coaching effort.
#48 by jheidelberg // Dec 21, 2021 - 9:50pm
Lamar's DYAR and DVOA stats do not tell the story. I have been asking for a long time and a response to a prior post earlier this year was this article by Vincent Verhei.
https://www.footballoutsiders.com/quick-reads/2019/quick-reads-2018-review
#50 by BJR // Dec 22, 2021 - 9:24am
Yes, to backtrack - I wasn't trying to shit on Lamar; he was perfectly fine for a rookie in 2018 and showed obvious promise/upside. Fumbles were a big issue, but something that could probably be fixed. And I agree that rushing DYAR/DVOA for QBs is quite broken, and should not be taken as read. He clearly breathed life into the running game.
That said, he was still significantly worse than Flacco as a passer by DVOA, so it's a huge stretch to say he saved Harbaugh's job that year.
#51 by BJR // Dec 22, 2021 - 9:30am
Yes he had higher ANY/A - against a much weaker schedule.
We've been here before. DVOA/DYAR aren't to be taken as gospel, but they are a very appropriate tool for comparing two QBs on the same team. There's not much point contributing on this site if you are going to look straight past them.
#45 by DisplacedPackerFan // Dec 21, 2021 - 5:03pm
I was poking at records for teams/coaches/GM tenures and found statmuse quite helpful here's the data since 2008 if you like. I was actually using it in relation to how bad GB ST have been since Thompson took over in 2005 vs how good NE ST have been. The Ravens move up to 5th behind NE, GB, PIT, NO. (Seahawks fall to 8th, Chiefs to 6th). Doesn't really change the argument at all. In fact it helps when you consider that both SEA and KC have been better recently with the better QBs, KC especially. You also clearly see the BAL-PIT rivalry being amazing when no other division comes close (Saints and Falcons in the NFC South would be the next best rivalry since 08). But statmuse lets you change the year and see quickly. So if you go with 2015 you finally dethrone the Pats and the Chiefs take over. If you go with the LaFleur tenure and start in 2019 then it's the Packers. But you also always still see PIT and BAL near the top.
#26 by TyWill40 // Dec 21, 2021 - 9:48am
The decision I was hoping "Risky Business" would also analyze was Harbaugh's decision to go for it on Fourth and five from his own 30 yard line trailing by 11 with 12:00 minutes to go in the game. I can't imagine the risk of giving more easy points to the Packers was outweighed by the reward of the Ravens keeping possession, especially given that it was a five yard proposition, not fourth and inches. Could you evaluate the math on that one? Thanks
#29 by ImNewAroundThe… // Dec 21, 2021 - 11:19am
#28 by nat // Dec 21, 2021 - 9:57am
It looks like since 2011, teams have gone for 2 when down 8 late in the game about 10% of the time. The percentage didn’t change much when I looked at 2016-21, so there may not even be a trend towards doing this.
10% is not zero. So there are circumstances when teams will do this. But it is pretty low.
This is a situation that comes up about 8 times a year, or once every four years for a given coach. So I wouldn’t expect them to spend much time thinking about it.
Still, it’s time to update the “Go For Two” charts.
#43 by Johnny Ocean // Dec 21, 2021 - 3:10pm
Typically games are played with your starting line-up for all but a few positions. The Ravens, due to injuries and COVID-19 were playing with an extraordinarily high number of replacement players against the Packers. The decision to kick the extra point was probably influenced by the coach's concern for the potential momentum swing to missing the 2 point conversion on his players. Young players playing their first NFL game do not have the perspective and maturity to handle big in-game negative events.
Analytics is great but football is more than just probability and a coach's decisions across a season will sometimes look pretty conventional from an analytic perspective but at other times the intangible aspects of the decision will come to the fore as they did in the Packer game.