It’s apparently “people complain about the NFL’s overtime system” week. I hate the overtime discourse, not because it’s silly (everything about football is silly), but because most of the time it misses the biggest, most glaring problem.
You could lock three people in a room and they’d come up with four different preferred overtime systems, and no matter which one they implemented it would result in complaints about fairness the first time it was tried because they all preserve the original sin of overtime– arbitrarily assigned possession.
Anyway, before I dive into my preference, I wanted to outline what in my opinion are the key features of a good overtime system.
- It should be fair.
- It should be quick.
- It should be legible.
- It should be football.
- It should be known in advance.
Let’s run through them one by one.
1. It Should Be Fair
Most overtime complaints claim to be about this, but they’re actually not about this. For instance, the old canard that “both teams should get a chance” is theoretically an appeal to fairness. But the college overtime format gives both teams a chance, and it’s way less fair than the NFL version.Want proof? NFL teams that win the coin toss usually elect to receive, but sometimes elect to kick, suggesting that receiving and kicking are somewhat balanced and certain factors will cause a team to favor one or the other.On the other hand, no college team that wins the coin toss ever elects to go first in overtime. Ever. Because going second (and therefore knowing exactly how many points you need on your possession) is a huge advantage, way bigger than anything granted in the NFL.A fair overtime system either (A) uses a starting state that is perfectly balanced such that both teams would be equally comfortable with either side of the deal (what the NFL attempts to achieve), or (B) determines starting state based on characteristics of the teams and the game itself. (We’ll get to this in a bit.)Here are two tests for how well a system achieves fairness. Test #1: does the team that wins the toss win the game 50% of the time? This is a good test of whether the starting state is perfectly balanced, as desired in option A.
The second (and my favorite) test for fairness would be: does the outcome of the coin toss impact the live betting odds for the game? This is a great test of whether overtime results are being determined by a coin flip or by features and characteristics of the teams and game itself.
2. It Should Be Quick
A lot of pie-in-the-sky overtime solutions ignore the basic time constraints the NFL is operating under, especially during the regular season. They have a limited broadcast window. The game needs to end during that window. Any proposed overtime system that could possibly extend the game by two hours is a non-starter. The playoffs are a different animal, although broadcast issues aren’t the only constraint. Football is an entertainment product and the league is keenly aware of attention span issues. They find that shorter games are more engaging and every offseason track average game time and make tweaks to ensure it’s falling under a certain threshold. People prefer quicker games.
3. It Should Be Legible
Again, football is an entertainment product. It shouldn’t take a degree in economics to come up with an overtime system, and it shouldn’t take 30 minutes to explain the system to a casual fan.
4. It Should Be Football
People watch football because they like football. Dunk contests in basketball are exciting, but settling tied games by staging impromptu dunk contests would be dumb because dunk contests aren’t basketball. Every distortion to the order of play or constraint on the decisions available to a team makes overtime less like football and should be used sparingly, if at all.(Arguably these last two could be bundled into a single point, though I think there are enough proposals that resemble football but aren’t legible to justify keeping them separate. But essentially everything that fails this point will also fail the previous point.)
5. It Should Be Known In Advance
However you’re going to assign possession in overtime, teams should know ahead of time so they can manage their end-of-game decisions accordingly. This is the least-necessary component, to be honest, and could really be considered a sub-category of “it should be fair”.I would say points #1 and #2 are absolutely mandatory of any proposed system, points #3 and #4 are strong aesthetic preferences, and point #5 is more of just a nice-to-have (that, again, could really serve as a sub-category of #1).With that framework in mind, let’s look at some common overtime systems.
The Old NFL System
(15 minutes, coin flip for possession, sudden death)
This system succeeds on points #2, 3, and 4, but gets a giant F for point #1 (and #5). Blatantly unfair, the team that wins the toss has a substantial starting advantage.
The New NFL System
(10-15 minutes, coin flip for possession, modified sudden death– FG doesn’t automatically win on the first drive)
This one is fairer (as in it gets closer to 50/50 chances to win), but the problem is that even if you perfectly balance for a given offensive environment in aggregate, that won’t be perfectly balanced for individual games (the TD requirement means a different thing when a game features two juggernaut offenses!), and also offensive environments change. Such a system will never be finished, you must constantly tweak it year after year to keep it fair.It also runs into problems with point #2 because giving an extra possession after a field goal by definition extends the game more. (And possibly a third possession after a second field goal, too!)
Both Teams Guaranteed One Possession
This fails point #2 and also if both teams score we’re right back where we started in the current system where the team that goes first has an unfair advantage.
Both Teams Guaranteed *MATCHED* Possessions
Meaning Team B is guaranteed to get as many attempts to score as Team A. The college format is a subcategory of this where possessions start on the 40 yard line, and I already noted how badly that fails test #1. (Example: imagine you have 4th-and-6 from the 27. Do you kick or go for it? The first team doesn’t know the right answer, while the second team often does– if the first team scored a TD, you go for it. If the first team failed to score, you kick.)In addition, it should be obvious that this is the worst proposal of all as far as point #2 is concerned. And it fails points #3, #4, and #5, to complete the sweep. This is the worst overtime in sports!
Both Teams Guaranteed Matched Possessions, But No Punting/Kicking
This solves the information advantage problem of the last format (since teams might know what they need, but they can’t act on that knowledge by managing their drives any differently). It keeps the time disadvantage, increases the “doesn’t resemble football” disadvantage (because special teams are part of football), and becomes less legible. Still better than the last system.
The Field Position Auction
I’ve seen two basic flavors to this proposal. In the first, both teams tell the referee “we would like to start with possession at the X yard line” and whoever names the lowest number gets the possession. So a great offense might say “we’re willing to start at the 14 yard-line” while a worse offense might say “we’re willing to start at the 32 yard-line” (knowing that this makes them less likely to win the bid and more likely to start on defense).In the second proposal, one team names a yard-line and the second team opts to start on offense or defense from that point. So in last night’s game between the Bills and the Chiefs, the Bills could have set the starting point at the 10-yard line and the Chiefs would have had to decide whether to start with the ball 90 yards from the end zone or let Buffalo go first.These proposals have a lot to offer. They’re fun strategic thought experiments. They’re almost perfectly fair (though not *completely* perfectly fair, since you still have to decide how to break ties in the first case or who offers and who accepts in the second).The problem? They’re fairly illegible (would take a lot of time to explain this to a super casual fan, though given a decade or so anything will start to seem pretty normal), and also they’re “not football”. (There’s no other situation in football where starting field position is determined in this way.) But overall there’s a lot of positives and I think it’d be an improvement.
Possession is Automatically Awarded to the Home Team
I actually like this one as a decent compromise because awarding possession based on characteristics of the team open up space for a system to be fair without being even. Yes, it gives the home team a big advantage. But, like, we’re pretty okay with home teams having a big advantage already. And it’s probably a better fan experience. Fans like seeing the home team win.This is our first system that solves point #5, though that was the least important point (and a lot of other systems could also solve point #5 by moving the overtime coinflip to the beginning of the game, which they should do because it’s a positive change IMO).Plus in the playoffs where home games are awarded based on season-long performance, this system rewards teams for good play, which is good. No perverse incentives or anything. Win 13 regular-season games instead of 11 and you get a built-in advantage in the playoffs. That seems fine.(I still think we could do better, though.)
Ties Are Ties
Obviously this system wouldn’t work in the playoffs (though since time is less of a constraint in the playoffs we could do something basic like just keep adding non-sudden-death 10-minute “quarters” to the end of the game and playing on until one team has the lead at the end of one of them).Otherwise, I love just letting ties be ties. It gives us better information about teams when it comes to seeding. (All else being equal, a team with 9 regulation wins and 7 losses is better than a team with 5 regulation wins, 7 losses, and 4 overtime wins.) It’s the quickest solution of all, it’s exquisitely fair, it aces every test. Even with my preferred overtime system, I’d rather that be a playoff-only thing and just let regulation ties be ties.But Americans hate ties, so it’d probably never happen.
My Proposal: Stop Awarding Possession
Again, at the top I pointed out that the solution to possession being awarded by a coin flip seems to be… just not awarding possession by a coin flip. At the time the game ends there’s a perfectly good possession already. Why are we throwing that away and creating a new one whole-cloth? That’s just creating the fairness problems and then trying to solve them.I’d propose: at the end of regulation, teams switch sides of the field and then play continues where it left off, except sudden death.This proposal isn’t *even*. One team will enter overtime with an advantage. But it’s *fair* because that advantage is determined by the play on the field to that point. Don’t want your opponent to start sudden death overtime with 1st and goal from the 6? Then don’t let your opponent drive down to 1st and goal from the 6.More than any other proposal out there, this system would ensure that live betting odds were perfectly stable from the end of regulation to the beginning of overtime, which as I mentioned is my preferred test of fairness. There’s no arbitrary event that meaningfully impacts who we think is most likely to win the game. That belief is only determined by the teams on the field and the play so far.This proposal is fairly quick since teams will often start overtime in much better scoring position. As a counterpoint, teams that would have once tried to win in regulation will sometimes try to win at the beginning of overtime instead. But I think the net result would be making normal games a hair longer while making the longest games significantly shorter. And since I think length is more a problem at the 95th-percentile than at the 60th percentile, I think that’d be an acceptable tradeoff (and a huge advantage over any matched possession format).It’s completely legible; we’ve seen teams make transitions like this twice a game every game for our entire football-watching lives. It’s exactly the same as the transition between the 1st and 2nd quarters and the transition between the 3rd and 4th quarters.It’s completely football. Teams play offense, defense, special teams. No special rules, no special constraints. In fact, it makes the end of tied games look more like football to boot. As things stand, if a team gets the ball with 27 seconds left in a tied game, they’ll often kneel out the clock and accept OT. If not, they might throw a few bombs and, if they don’t connect, kneel. “All Bombs And Kneels Offense” is exciting until you get to the kneels, but it’s not really “regular football”. Preserving possession would result in teams playing their regular offense on the final drive, only with the entire game on the line.Finally, the setup in overtime is known in advance. Every team knows at the end of regulation that what they do will impact where things stand, and they’ll manage the game accordingly.I believe this format is intuitive and solves all of the issues with overtime. I think most of the resistance to the idea stems from the fact that we’ve been doing things the other way for so long that we’ve become blind to how absurd doing things the other way really is. We start with the assumption that we need to come up with a way to arbitrarily assign a starting game state and then agonize over how to make that arbitrary starting state as palatable as possible. But this is begging the question. Why should we accept as given that we need to arbitrarily assign a starting state instead of just using the one we already have?Two quick arguments that I have received against this system:
Argument #1: It Makes Game Less Exciting
The idea here is it’s exciting when a team has to go 40 yards in 37 seconds to get a game-winning field goal. And sure, it is. But we’ll still get plenty of that from teams that are *trailing*, we’ll only lose it from teams that are *currently tied*.The two problems is that for teams that are currently tied, that game-winning walk-off field goal isn’t “do or die”, so they can easily switch to kneeling out the clock and accepting overtime if they ever find themselves in an unfavorable circumstance. Kneels are not exciting!Secondly, this only affects a very narrow range of circumstances. With more than say 1:20 left, a team has plenty of time to just run its normal offense anyway (except maybe skewing more towards the pass, which is maybe more exciting but definitely less like normal football). With less than 20 seconds left, teams aren’t going to go for a field goal, anyway. (Maybe they should, but we all know they mostly won’t.)So there’s just a narrow range of games that this would impact– games where the trailing team tied with somewhere between 30-90 seconds left. And again, the thing that replaces that “bombs and kneels” drive is still plenty exciting in its own right– a team running its best possible offense with the game on the line.I’m not at all convinced this proposal would result in a net loss of excitement. We’d lose some here, we’d gain some there, but I suspect it’d all come out in the wash.
Argument #2: There Are Actually *TWO* Abandoned Possessions
This is a very esoteric argument and I’ve only ever heard one person make it (hey Maurile), but in the interest of completeness, I’ll give a response anyway. Basically, the complaint is that there are abandoned drives at the end of both the 2nd and 4th quarters, so why should we make the 4th quarter drive meaningful but not the 2nd quarter drive?Maurile proposes a coin flip to determine which abandoned drive determines starting game state in overtime. Which would have the advantage of making the end of the first half a lot more exciting, too.But it’s less legible (try explaining to a casual watcher why the other team suddenly has 3rd-and-2 at their own 27 yard line). And it’s less football, because it’d be the only example of reverting back to a previous game state. At every other point except for the two opening kickoffs, state of play is determined entirely by the state of the game at the end of the previous play. (And each beginning-of-half kickoff is balanced by the existence of the other.)Additionally, it’s not known in advance and would result in a large luck component if e.g. one team had dominant position at the end of the 1st half and the other had dominant position at the end of the 2nd. (Maurile would probably argue that this is good, but it would dramatically fail the “do live betting odds move as a result of the coin flip” test of fairness.)Additionally, I don’t know how much fairer it really is. Receiving the ball at the beginning of a half is an advantage because it guarantees you either the same number of possessions as your opponent, or one more possession than your opponent. In expectation, you’re granted an extra half-possession over your opponent in that half.Maurile will acknowledge that an extra half-possession in the second half is more valuable than an extra half-possession in the first half. (Later possessions are more valuable because you have more information about what your team needs.)Preserving game state from the end of a half essentially mitigates that half-possession advantage. If you had 5 viable possessions (meaning you had a realistic chance to score, receiving the ball with 7 seconds left doesn’t count as a “viable possession”) in the 2nd half and your opponent had 4, then in overtime you’ll either continue your 5th possession (preserving the advantage) or else your opponent will begin their 5th possesions (erasing the advantage).Basically, overtime has a 50/50 shot to erase the “half-possession advantage” from one half of football. But since the half-possession advantage in the second half is greater than the half-possession advantage in the first half, consistently putting that advantage at risk results in better balance in expectation than sometimes putting that at risk and other times putting the other (smaller) advantage at risk.(I’m not especially confident about this last part and could be argued out of it, but the “legibility / football / fairness” complaints alone are enough for me to prefer simply preserving the state of the game at the end of the 4th quarter.)