Overtime: is going 2nd really better?
It's conventional wisdom that the team that wins the coin toss should always choose to take the ball second, so that they know exactly what they need to win or continue the game. I've accepted this argument all along for years without question, but recently I've started rethinking this.
Were we actually disavantaged in OT by losing the coin toss?
We got to choose our end of the stadium to play the OT session on, giving us de facto crowd advantage. When we took the ball, the score was tied. We didn't know what Bama would go on to do in its possession, but that's arguably not a bad thing. You play knowing that at least your defense will get its chance to get a stop in the next half.
By the time Bama got it, they were down seven points. Yes, they knew exactly what they needed, but knowing that you must score a TD brings is a pressure that we didn't have to face in our own possession.
If you go second, and manage to hold your opponent to a FG or less, then you certainly have an advantage going into your possession. But you don't know that going in.
I think it may actually be a close call about which is the best course of action.
January 4th, 2024 at 2:22 PM ^
Not when you win anyway.
January 4th, 2024 at 10:01 PM ^
cut to SheaPattersonVsArmydotgif
January 4th, 2024 at 2:23 PM ^
Going second reduces number of decisions, which one could argue is a coach execution advantage.
January 4th, 2024 at 2:34 PM ^
That also could be a DC’s advantage too
January 4th, 2024 at 6:19 PM ^
Going 2nd has also been advanced on the theory that should the game still be tied after the first OT, the team-going-first's defense will need to stay on the field on consecutive series, thereby rendering them more tired. Meanwhile, the team-going-2nd's defense would've had an extended rest. Theoretically, then, this should tip the scales to the team-going-2nd in a 2nd OT.
The fallacy in this thinking is if the game is a Michigan - Alabama Rose Bowl, Milroe gets stuffed on 4th down and Michigan wins in the first OT.
January 4th, 2024 at 3:03 PM ^
There is a time value to the information for turn based play. The value of new information increases once the turn has concluded since it reduces the array of potential outcomes that are unknown before the turn has started.
January 4th, 2024 at 3:41 PM ^
What really matters is the data!
Apparently, analysis of overtime games shows there is no advantage to choosing defense first (or to choosing offense first). 51% of teams that choose defense first win, which is not statistically significant:
https://www.frontiersin.org/articles/10.3389/frai.2020.00061/full
January 4th, 2024 at 4:13 PM ^
It would be interesting to know the breakdown based on who the "favorite" was to start the game.
What is the winning percentage of favorites having the ball first vs second in the deciding OT (be it first OT, second OT, etc).
January 4th, 2024 at 5:08 PM ^
They included the point spread (and home team status) in the regression model. Unsurprisingly, it predicts who will win, but they did not examine or report an interaction between point spread and OT order.
Still, since it was included in the model, we can say that OT order has no effect, while controlling for who was favored to win the game.
January 4th, 2024 at 4:44 PM ^
That may make it seem easier for a coach, but from the data it doesn't seem to actually help teams win more (to any statistically significant degree).
In our case, I liked getting the ball first because we'd just had a long drive to score. I'm curious to know what the record is of teams that scored last in regulation - if there is any "momentum effect."
January 4th, 2024 at 8:06 PM ^
Only people that were in the Rose Bowl will understand the importance of making Alabama drive into the north endzone. If I was Bama I wouldve took any scenario that had me in the south side.
The Michigan fans in that corner were frothing at the mouth.
January 4th, 2024 at 9:19 PM ^
The Michigan fans in the whole stadium were frothing at the mouth, but some of us had to listen to the Alabama band play over ours doing Temptation on 3rd down stuffs.
January 4th, 2024 at 2:24 PM ^
I mean, if you score a TD immediately, it of course changes the way you feel there.
That said, in your one OT possession, if you go first and kick a FG not knowing that the opponent will score 7 next. You probably would have used that FG down to get the first down. So, yes, it is still advantageous to go second. The crowd helps, yes, but being able to effectively determine how to use your four downs optimally is a key advantage.
January 4th, 2024 at 3:28 PM ^
That said, in your one OT possession, if you go first and kick a FG not knowing that the opponent will score 7 next.
Technically, the opponent would score 6 since there's no need to kick an XP on a TD.
January 4th, 2024 at 2:25 PM ^
Yes, as opposed to going first. Definitely better for knowing what to do on offense.
But, maybe not outweighing end zone factors.
If you trust your team's focus, however, the end zone factor may not weigh heavily with you.
January 4th, 2024 at 2:26 PM ^
I think knowing what you need to get is a definite advantage. It can be reduced by the team in defense picking an advantageous end of the stadium. I’d say sometimes teams give away that advantage by getting a bit conservative if the other team gets a FG first, but that could be all in my head. Michigan does have a pretty amazing record in OT given it should be close to a coin flip.
Upon further thinking- going second also allows you the option of going for 2 and winning the game if the other team scored and kicked the XP. Theoretically the team that goes first could also go for 2, but I honestly have never heard of that happening.
January 4th, 2024 at 2:26 PM ^
Okay, whatever...
January 4th, 2024 at 2:26 PM ^
By going second, Michigan got to choose which endzone they thought would be the loudest for them.
However, I like going second. Bama knew they had tobuse all 4 downs and that is huge. Had Bama went first and had a 4th and goal from the 3 they are likely kicking a FG.
January 4th, 2024 at 4:00 PM ^
agree with this - Michigan got to pick the endzone.
Once we scored a TD then I think it worked to Bama's disadvantage since they HAD to match and each mediocre outcome play added stress. Had we only kicked a FG I think it would still be to their advantage.
January 4th, 2024 at 2:26 PM ^
Division 1-A college football adopted overtime rules in 1996. There have been 328 overtime games since, and only four times have coaches opted to go on offense first upon winning the coin toss. Thus, there is an accepted belief that starting on defense first is advantageous, as validated by surveying college coaches. This study examines past game data to analyze whether there is truly an advantage to being on defense first. Results show some support for this idea, but that there may also be situations where not following the defense first strategy is worthwhile. Implications to coaches are discussed.
January 4th, 2024 at 3:20 PM ^
From the paper: "It was shown that a team who starts on offense and scores a touchdown wins over 70% of the time."
"There may be times when a coach who wins the coin toss should consider eschewing conventional wisdom - for example, with a high-powered offense or with a nothing-to-lose attitude, the psychological impact of scoring first may outweigh the perfect information a team gets from playing offense second."
This paper was published in 2007, so there's at least 16 years additional data available to revisit this if there are any enterprising grad students with a stat class requirement. They could factor in red-zone effectiveness if that data is more readily available now.
January 4th, 2024 at 2:27 PM ^
It's a good question. Not knowing what will happen heading into OT, I have to think that going second is generally best for the standard reason that you'll know what you need to do. But I suppose I could imagine an end of a stadium being so hostile that it would make a difference. And certainly you've put yourself in the driver's seat if you manage to score a TD on the first possession of OT. Your opponent now has to score a TD and then, assuming they successfully kick an extra point, immediately start over on another drive, after which you will know what you have to do.
Another thought: it's absolutely ridiculous that a game can become an exchange of two-point tries. The PSU-Illinois game that ended that way was absurd.
January 4th, 2024 at 2:29 PM ^
Second is the right play. I get the pressure argument if the first team scores a TD but that is not enough to give up the decision advantage. Especially if the first team fails to score at all. Then the second team just needs to set up proper FG placement.
Just an aside I wish they would start with the ball at the 35. Most college kickers can hit a 42 yard kick pretty consistently, but a 52 yard kick totally different story. I think having to get at least one first down to get a scoring opportunity is not unreasonable.
January 4th, 2024 at 2:37 PM ^
“Most college kickers can hit a 42 yard kick pretty consistently, but a 52 yard kick totally different story.“
Agree, great idea…that’s probably why the NCAA doesn’t do it.
January 4th, 2024 at 2:45 PM ^
A factor here is that they try to write the rules to apply to all 3 divisions (instant replay is still I think the only exception).
Especially 25 years ago, I think they figured 35-40 yards was right around the max range of an average D2 or D3 kicker, so teams could still have a shot at a FG even if they didn't pick up a first down.
Plus they realized that game length would quickly become a problem if they started much behind the 25. It ended up being so much of a problem that they just do 2-point conversions after the 2nd overtime now.
January 4th, 2024 at 2:49 PM ^
Just an aside I wish they would start with the ball at the 35. Most college kickers can hit a 42 yard kick pretty consistently, but a 52 yard kick totally different story. I think having to get at least one first down to get a scoring opportunity is not unreasonable.
Completely agree from a competitive/philosophy standpoint, but it would probably make OT games last longer, especially at sub-Power 5 levels.
January 4th, 2024 at 3:33 PM ^
But does the length of games really matter at FCS, DII and DIII levels? Even sub-P5 FBS? I would suspect those games would take considerably less time given the reduced TV presence. There aren't many nearly 4-hour Fox games at those levels.
January 4th, 2024 at 4:29 PM ^
I think the concern is more about the physical impact of extending games.
January 4th, 2024 at 2:53 PM ^
My thought is - star the first OT from the 25, the second OT from the 35, and the third OT (and any subsequent OTs) from the 45.
January 4th, 2024 at 3:36 PM ^
Shouldn't it be the other way around? The longer the game goes the easier you want the scoring to be.
January 4th, 2024 at 3:44 PM ^
No, I think the scoring should get tougher to reduce the number of OTs.
January 4th, 2024 at 4:46 PM ^
If scoring is tougher, that could also cause games to drag on.
I think starting at the 25 is a reasonable compromise.
January 4th, 2024 at 4:38 PM ^
"Most college kickers can hit a 42 yard kick pretty consistently, but a 52 yard kick totally different story."
I'm not too sure that College Kickers really do kick 42 yard field goals consistently.
For example, using the data here a 40-44 yard field goal is successful only 63% of the time. (I selected the larger range for more sample size since there is so much variance, but will call out it is 74% for 42 yards exactly)
Game winning pressure kicks are going to be an even lower percentage chance to boot. So I'm not sure that there really is a need to move things further back to make things harder on the kicker.
Also let's me honest... nobody wants to lose because their kicker missed a kick in overtime.
January 4th, 2024 at 2:30 PM ^
its always better to go 2nd
January 4th, 2024 at 2:30 PM ^
If Bama goes first they kick a FG and Michigan wins with the TD run. They know they had to go for it because the went 2nd. Going 2nd gives you a complete data set and you always want to play with a complete set of information.
January 4th, 2024 at 2:32 PM ^
It's better because you know before you run your first play what you have to do to either win or keep the game alive. The team going first doest have that conundrum so if faced with a fourth and one at the 16, do you go for it or kick the FG?
Just because we won doesnt mean it's the better choice.
January 4th, 2024 at 2:34 PM ^
Without looking up any data (as it pertains to how many overtimes games have on average)- it's probably just as beneficial to be going 2nd in the second overtime, if not more due to the 2 point rules. So if you happen to be tied after the first one and you went first, you get defense second on the side of the field you want (I think) with the knowledge of if they converted the 2 pointer or not. Same pressure caveats apply.
January 4th, 2024 at 2:35 PM ^
Older study shows there is a meanignful, but not overwhelming, advantage to going second.
https://www.degruyter.com/document/doi/10.2202/1559-0410.1049/html
January 4th, 2024 at 3:42 PM ^
One of the authors revisited this more than a decade later and found no evidence of any advantage either way. He used the NCAA's database of overtime games (which ran from 2013 to 2019) and found the results of the coin toss and decision in ESPN's box scores. There were 243 OT games in the sample; the team on defense first was 124-119.
This doesn't mean there's no advantage, of course. It's not large, but we wouldn't expect it to be.
January 4th, 2024 at 3:49 PM ^
I love how quick this blog turned this into a data analysis issue.
January 4th, 2024 at 4:04 PM ^
The funny thing is, I was trying to do the opposite. There isn't enough data so we have to understand another way.
January 4th, 2024 at 4:03 PM ^
Adding the two data sets together it comes to 304-267. Under the assumption that it's 50/50 we'd expect 285.5 +/- 12 wins.
In passing I'll note that the author is unwilling to assume that the advantage has been constant through the years; he proposes the possibility that strategies might change over time and alter the probability we're after (in particular, he thinks the data suggests the advantage was bigger in the early years of overtime). I wouldn't expect any change to be significant and the change he notes is consistent with random noise, but it's another example of why it's hard to demonstrate anything convincingly out of small samples.
January 4th, 2024 at 2:36 PM ^
I think the argument about the crowd is a good one - at a neutral site field especially. But knowing you need 7 is better than knowing your defense gets a chance to stop them. Especially when their kicker is that good. Had they gone first and the plays went the same way but on the Alabama EZ, they lose 26-23. Maybe there's no bad snap by the center on that side but also they're kicking it on 4th down.
January 4th, 2024 at 3:27 PM ^
Yeah, the neutral site factor makes it slightly more meaningful on which endzone you're playing in. But if M had won the toss and not chose to go on D first I would've considered it coaching malpractice on the scale of Marty Mornhingweg not taking the ball in OT against the Bears.
January 4th, 2024 at 2:38 PM ^
From 2012-2019, it was 54.7%-45.3%, in terms of who has the ball last in the last overtime winning the game.
That's statistically significant, that's not random noise. There's definitely an advantage.
It was 51.1%-48.9%, in terms of who has the ball last in the first overtime winning the game. It's still an advantage, but if the game moves to a 2nd OT, then you're working from a disadvantage.
January 4th, 2024 at 2:43 PM ^
In the first OT, if you get the ball second and the other teams scores 7, your offensive plan should be to score 8.
January 4th, 2024 at 2:48 PM ^
Maybe.
From a strict analytic POV, you'd want to go for 2 if and only if (1) you thought the likelihood of your 2-point conversion play working was higher than (2) your likelihood of winning the game in some other overtime.
That 1st thing is a bit hard to quantify, of course. So It often becomes more of a gut decision than an analytic one.
January 4th, 2024 at 3:13 PM ^
That would be a roll of the dice. 2-point tries are converted about 40-45% of the time. In game-deciding situations, the numbers may be different, but I'd be surprised if they were better.
January 4th, 2024 at 8:14 PM ^
You do this in games in the 40s not in the 20s. Both teams had full quarters where they couldnt move the ball you dont put that on one play.
January 4th, 2024 at 3:47 PM ^
I posted the link above; the 51%-49% could easily be random noise.
To be clear, I think there's an advantage, but seven years of data wasn't nearly enough to demonstrate it statistically and we don't know very much about how large it is.