3: Concession culture

2020.07.23

Herm Edwards’ fabled 2002 speech as New York Jets’ head coach was an inspiration to all. Faced with an organization imploding around him, he let the world know that no matter how dire the situation, you play to win the game.

But Herm doesn’t always feel that way. On October 19, 2019, Edwards’ Arizona State team faced a fourth-and-5 from its own 47 yard line, trailing Utah 21-3 with just under nine minutes left. The path to winning would be to pick up the first down, score, force a Utah punt, score again, recover an onside kick and score again.

An unlikely series of events to be sure, but not unprecedented. It would have been a part of Arizona State lore for decades to come. And it’s not like Edwards wasn’t directly involved in one of the most improbable finishes in football history. Besides, what was there to lose? A football game, sure, but you can only lose it once. Trying and failing on fourth down doesn’t add any more losses to the team’s record than punting.

Well, Herm Edwards chose to punt. To just give the ball to his opponent. Having seen his squad score a total of three points so far, he decided chasing a tiny chance of winning was less appealing than preventing a 21-3 score from getting worse. Herm Edwards plays to win the game – except when it increases the chance of an embarrassing score.

Greg Schiano is also a member of the win-at-all-costs hall of fame. As head coach of the Bucs, he asked his team to blitz the Giants’ victory formation at the end of a 2012 game that was already decided. Facing criticism after the game for risking injury to his opponents, Schiano was quoted as saying

I don’t know if that’s not something that’s not done in the National Football League, but what I do with our football team is we fight until they tell us game over.

Inspiring to be sure, except that is not always true for Greg Schiano. In a September 22, 2013 game against New England, Schiano’s team faced a fourth-and-9 from its own 21 with 7:45 remaining and trailing 23-3. Winning was a longshot here, too, and Schiano chose to punt instead of facing the prospect of not converting and losing 30-3 instead of 23-3. A weird decision from a guy who says his team is going to fight until the game is over.

Now consider we add a new rule to football. At any time, a team is allowed to quit. With such a protocol, there is no doubt in my mind that Edwards and Schiano would have tried to convert those fourth down situations. If they didn’t get it, so what? They could have given a friendly wave to their counterpart on the other sideline and walked off the field. Their dignity, and the current score, would have remained in tact.

Curling is the only sport I’m aware of where conceding before one is mathematically eliminated is an option. In fact, it’s not just an option – it’s expected in many cases. When a game is out of reach, it can be viewed as rude not to concede.

This feature is liberating when it comes to strategic decisions. Facing the curling equivalent of these football situations, you always go for it. And if there comes a point where the outcome of the game is obvious and play becomes meaningless, the contestants shake hands and get off the ice. This results in a loyalty to win probability that is unmatched in competitions where the winner is based on counting things.

The blank end predates concession culture in curling, which didn’t become ubiquitous until the 1970’s. But it is the most obvious example of how simply trying to score the most points can be a sub-optimal strategy. If one has the hammer, then scoring zero is superior to scoring one.

Let’s go graphing for the first time at the DoubleTakeout.com blog. Here’s the win probability for the team with last rock after scoring 0 or 1 in an end (with a tie score starting the end). This is from men’s and women’s events combined for the past two seasons.

This plot confirms the strategy that has governed the game for decades. Scoring zero is better than scoring one in every end. Well, except the second end for some reason. There are lots of cases here so it’s not a product of small sample size. Propose your theories on why the second end is an exception in the comments or through the feedback link above.

Regardless, blanks are better than single points (until the final end, obviously) because maintaining control of the hammer is not worth the trade of scoring a single point. This is the most obvious case where win probability rules, but curlers are making choices regarding win probability all game long. Not always accurately or even consciously, but given enough experience, curlers quickly understand you always play to win the game and not to avoid losing by a lot, as our friends in other sports sometimes do.

As in other sports, the strategies that lead to more points in curling expose one to giving up more points. Thus making up a deficit requires playing in a way that risks giving up even more points. But the ability to concede gives one the freedom to risk big deficits in the pursuit of victory.

Because of that, all decisions must be evaluated based on their contribution to win probability. Here’s an example: Over the past five seasons, with hammer, Jennifer Jones has outscored opponents by 1.02 points in the first end, while Rachel Homan has outscored opponents by 0.70 points in the same situation. Who has been more effective? It’s a trick question. There is not enough information to make a judgment. Because zero is better than one, and both are worse than two, you can only evaluate their respective first-end performance from the cumulative win probability of individual cases.

If it wasn’t for concession culture, I’m not sure things would be this way. Post-game interviews with curlers are notoriously bland. There’s no Herm Edwards or Greg Schiano quotes out there. Maybe that’s because it’s obvious you always play to win. At least until it’s time to quit.

User comments
  1. Ryan McGhee · 2020.07.27 · 7:45 am

    My 2nd end theory:
    The only way the game is tied in the 2nd is if the 1st is blanked. Most teams aren’t going to try to go blank-blank. They’re going to try to go blank-deuce and have hammer in evens. If the 2nd is blanked, it’s likely because the team that started with hammer made a mistake and had to bail or was out-played by the team without. So, to me, a blank in 2 after a blank in 1 is an indication the team with hammer might be a little off their game or that the team without is doing a better job of controlling the house.

    1. kenpom · 2020.07.28 · 1:16 am

      That was my thinking as well. We should be able to test this by seeing what teams normally average in the first scoring end after blanking the first vs. blanking the first and second.

Join the discussion!