The PEI vs Northwest Territories game at the Brier was one of those sneaky interesting games that can really only happen in the Brier. A playoff berth was on the line and making the playoffs in the Brier might have been the pinnacle for both teams curling lives.
Heading into the ninth end, PEI led by 2 and NWT had last rock. By the time the TSN coverage returned from a commercial, it was apparent that this was going to be an open end. Both teams preferred to blank. The broadcast team of Russ Howard and Joanne Courtney were in agreement that this was a mistake for NWT. You’re down two with two ends left, so you can’t waste an end. You can hear the commentary here (with possibly the help of a VPN for my Canadian friends).
I wanted to believe this was true, but given that the end was headed for a blank from the first shot, Russ and Joanne had a lot of time to talk about it, and the more they talked about it, the less convinced I was becoming. I was trying to simplify the decision in my own mind but couldn’t really get to a place where I was sure about it.
Eventually, Russ decided to play devil’s advocate and pointed out that if your score 3 in 9, you still have to do a bit of work to win the game, but if you score 3 in 10, you win the game. I think we pointed this out on the blog some time ago, but I’m too lazy to go find it. It’s actually a pretty tough point to refute. Joanne offered that it’s harder to score 3 in 10 than it would be in 9. Which actually seemed like a decent counterpoint.
Also, you can set up some corner guards in 9 and go for 3, and then bail out and maybe try to blank in certain cases if things aren’t going well. The broadcast team didn’t mention that, but that would seem to be a point in favor of not auto-blanking.
Anyway, we have a lot of data to help us with this, and has it turns out, one team did make a mistake here and it wasn’t NWT. Let’s break all of it down in 3 easy steps.
1. Being down 2 with hammer in the 10th is better than being down 2 with hammer in the 9th
You can check the true win probability table (which is due for an update this summer) and see that elite men’s teams have an 8.2% of winning when down two with hammer with 2 ends left, but if they blank that rises to 10.1%. I think these teams would admit that they aren’t among the top 25 in the world, but even using the 26-100 range we get the same relative answer. You have a 16.6% chance of winning down 2 with 2 ends left vs a 18.6% chance if you blank.
If we look at real-world data, we come to same conclusion. Using data from my shot database from the Slams, Curling Canada events (Brier, Canada Cup, Pointsbet Invitational, etc.), Olympics, Worlds, and Euros, we find that teams have gone 43-385 (10.0%) when down two with hammer with 2 ends left and 51-348 (12.8%) down two with hammer in the final end.
In itself, this isn’t proof that Koe’s approach was correct, but it’s something. All things being equal, the trailing team is fine with a blank.
2. Teams up by 2 that start with a draw lose more than teams that guard.
We can drill deeper into that shot data to help us out. PEI opened with a draw in this case which is what set the blank into motion. Teams up 2 without hammer with 2 ends left that opened with a guard have gone 137-10 (93.2%), while teams in that situation that opened with a draw have gone 249-33 (88.3%).
It’s close, but based on history, it appears it’s PEI that made the mistake in this end. The analysis isn’t quite this simple, of course. You might notice in that graphic that the blank rate is hardly different whether the team throwing first puts up a guard or comes in. That’s because the team with hammer has normally guarded regardless. NWT’s hit was just the 16th recorded in my collection of shot data, with this being the third victory in that sample (for an 18.8% win rate).
PEI kind of admitted to the mistake, because when NWT rolled out on their fourth shot of the end, PEI threw up a guard, willing to take a shot at a force.
It’s interesting that PEI’s choice escaped the criticism of the broadcast crew, because in modern times, elite teams have usually chosen to open with a guard in this spot. In the last three slams, teams have chosen to open with a guard in all 14 situations like this. The Brier is a little behind on this strategy, with teams opening with a guard in just 7 of 16 cases over the last two years. Though most of the draw cases were done by weaker teams.
That includes Jamie Koe himself, who opened with a draw in that spot against Quebec’s Julien Tremblay in their opening Brier game this year.
3. It is not harder to score 3 in the 10th end than the 9th end.
This seemed like a great point from Joanne. In the 10th, your opponent is going to do everything to prevent a 3. In the 9th, it’s possible this isn’t totally the case. Maybe there are times an opponent will accept a 3 to not give up 4, for instance. But this is not the case in practice. It’s been easier to get the 3 in the 10th, and it’s actually not all that close.
In 399 games, hammer has scored 3 or more in the final end a total of 27 times (6.8%). In 430 games, hammer has scored 3 or more in the next-to-last end a total of 19 times (4.4%). There have been 53 blanks in those cases, but even if we ignore blanks the chance of scoring 3 or more is just 5.1%.
The difference isn’t enough to conclusively say it’s easier to score 3 in the tenth, but there’s certainly nothing here to support the idea that it’s easier to score 3 in the ninth, either.
What can we learn from this? If you’re up 2 without hammer in the second-to-last end you really should play for the force. Which seems to be what most elite teams do under the five-rock rule, anyway. It was actually pretty shrewd of Koe to go for the blank when presented with the opportunity.
Same situation with Jennifer Jones and Rachel Homan in Scotties final. In that case, Rachel put her first rock in the rings. Jennifer called for a corner guard. Rachel then put another one in the rings. Not correct choices based on the data.