Now Available! EBook from CWM

Now Available! EBook from CWM
Order Now from Amazon

You can also get an epub copy


Monday, December 15, 2014

Giving Away Points at the Canadian Open

Eve Muirhead won her first Canadian Open and Brad Gushue won his second Grand slam of the year.  This year the Open added a women side to the bracket and also had a bracket.  Instead of a round robin format, qualifying came through a triple knockout format.  Unlike in a round robin where an 0-4 team still has to finish all their contests, this ensured no team would be playing without some chance of still winning the event.  By qualifying in the A side, teams were awarded hammer for the quarter finals.  Three qualifiers came from a B event (where teams had lost 1 game) and three more from the C side (teams which had lost two games).  This also meant half the field qualified for the championship rounds (16 of 32 teams).  A small wrinkle that makes the event somewhat different without really impacting the actual games.  

Another wrinkle that continues to introduce never-before-seen strategy decisions is the five rock free guard zone.  The semi-finals were all played during the same Saturday evening draw.  In the span of a few minutes, Steve Laycock and Rachel Homan both appeared to concede a stolen point in the 7th end.  Each of them were one up with hammer and rather than draw to go two up heading to the last end, chose difficult hits which were highly probable to give away points to their opponent.

In the case of Laycock, he had a possible draw to full four foot (blue line) but instead chose a hit and roll that, to score one, needed to land in an area of a few inches (green line).

Laycock is Red

Everyone in the booth was questioning this decision.  His opponent Brendan Bottcher was possibly pleased. Steve is from Saskatchewn and historically teams from that province tend to hit more often than draw (sadly, I could not find a Bob Pickering photo on the internet), but that was in the old days of curling.  These days, most any Grand Slam level skip would draw.  

Rachel Homan had a possible chance at a blank.  Facing a simple draw for one (blue line), she instead chose to try a double (green line) that could result in a blank to hold hammer one up heading to the 8th end, but also could be a likely steal.

Homan is Yellow

Team Homan was being coached by Richard Hart.  He came out and encouraged Rachel to make the call she wanted to make.  They discussed that whether she is down one or tied in the final end, she'd need to make the same draw.

Laycock and Homan ended up with victories and trips to the finals, where they both lost.  So what do I think of these calls?

Looking at 4 rock FGZ numbers (2003 to 2014), in the final end, mens teams win 76% when tied and 90% when up two without hammer.  Womens teams win 70% when tied with and 86% when up two without.  The 76% is deceiving because it includes games including weaker teams.  If we look at only Grand Slams, it's closer to 79%.  Still, not 90%.  Women's results don't show this type of difference for elite play.

So, the question is whether adding a 5th guard for a team down 2 is significant enough to overcome the gap (11% and 16%) in WE.  In the case of a tied game, the rule has no impact.  When up 2 without hammer, a team now needs to decide if they will try tick shots on corners or pile up rocks in the house or even put up a centre guard (common if they are up 1).

There could be some over thinking with both of these shots.  Both Steve and Rachel called shots which if executed would NOT result in a steal.  A made shot by either (however slim their chances) would benefit them. Their calls both removed any chance of a steal of two and in the case of Rachel, perhaps even still get a blank.

We do not have enough data yet to specifically state that the WE of a team 2 up without hammer in the final end in 5 rock FGZ is closer to the 79% or 70% WE of being tied with hammer.  We could start to look at odds to score deuces and threes in 5 rock, but that will take me too much time and I'd still rather wait until we have more data.

Let's look at the notion of tied with hammer.  The ability to make the tick shot has improved dramatically just in the past 2-3 years.  Let's look at recent data:

These are very small sample sizes (around 100 situations each) but it is clearly showing a trend for teams at the elite level.  Reason to be wary of small sample sizes: the data from 2010-2012, taken from the most elite events, is actually 10% BELOW that from Playdowns (78% from Provincial, Brier and Worlds), games which include many "B" level teams.  

If these trends hold up for larger samples, it's fair to say, for Laycock the decision was either a small mistake or fairly close to equal (with the 5 rock rule), and for Rachel given some chance at a blank, possibly close to an even decision. Given their proficiency with the tick shot, she likely expects her WE to be above 80% tied with hammer.

Top skips will almost always err on the side of having hammer.  Even when the numbers indicate otherwise, skips will make a decision that includes having the last shot.  I often argue, as in the case of "two-hammers-to-one" its importance is sometimes over emphasized.  Because of our validation of the numbers when 1 up without hammer in the final end (60% to 40%), many teams altered their strategies in the later ends. 

While we're here, let's look at recent results (again, small sample sizes) for these top events for 1 up without hammer.  We see not much has changed and results are in line with historical numbers.

Given a close decision, I understand why an elite skip wants to have the last shot: if they thought otherwise, they wouldn't be the best in the world.  In both of these situations, the odds were probably very close, we can't argue much with their preference to have hammer in the last end.  

As teams improve even further at the tick shot, I expect we may reach a point where elite level games are above 90% for tied with hammer.  There's been some discussion about potential rule changes to thwart this type of result. A "no-ticks-allowed" rule might be just around the corner. 

Of note, at some point during the mens final broadcast we were told Sundays draws were a sell out.  That is nice to hear but the chairs still appeared just over half full.  I have hope that one day a Grand Slam will be standing room only and fans will cheer loudly throughout.

Until next time...


  1. You're too polite, Kevin. At best a skip sacrifices five percent of win expectancy in exchange for the reassurance of having final stone. Maybe for most players a WE swing < 10% is seen as trivial, because it's single digits?

    Five to ten percent is enormous imo. Trading that for momentum or emotional security of hammer seems similar in reason to NFL coaches choosing to punt on 4th & 3 at the opponent's 40. Risk aversion.

    Anecdotally, it sure does seem like five rock will lead to an increased number of 2+ ends tho.

    Great work as always.,70%20&%2071.jpg

  2. It is possible I'm too polite, but there are many ifs in this analysis with less hard numbers than we would like. But in the same way stats folks felt bunting was originally a poor decision, later analysis then showed that it can be under certain circumstances. You've made me consider wirtting a part 2 that includes some more math to break these decisions down further. Stay tuned.

    1. Also, I found that photo of Bob Pickering, but I was looking for the shot og him lifting his rock on the backswing. Strange there's nothing online. Would make the modern no-lift player cringe.

    2. This comment has been removed by the author.

  3. Hello Kevin,

    Thank you for all your efforts on this blog, I have been a huge fan since day one as you know. I have used the information gained here as both a player and a coach.

    I am pretty sure Steve was "right" or at least not wrong. While watching the game the shot was called and instantly the announcers where all over him, " I like to be winning" and many other comments like this. Sounded a lot like "trust your defense"

    When I am making a choice that is potentially going to drastically change my WE I want to hear "Trust my stats guy"

    Hopefully in part 2 you can do the proper math, but I had a robo perfect drawing Steve needing to win in 8 tied with hammer 90% of the time to make hitting correct.
    @85% on the draw in 7 he would need to win tied in 8 84% of the time
    @ 75% on the draw in 7 he would need to win 78% of the time.

    All seems pretty reasonable to me.

    Couple that with the fact that like you mention 2 down without playing 8 is a spot that is changed the most by the 5 rock rule. (Currently 10% WE)

    In that scenario upping your chance to get three greatly changes the overall WE ( admittedly a change from 10% - 14% would be considered massive ) and most people believe that the 5 rock rule will increase 3's scored in these situations.

    When 2 is scored the fact that elite teams lose very few extra ends these days comes into play.

    I think this is a call that will be considered standard in 4 years, and makes for a great topic to speak about now.

    All the best,

    Jason Gunnlaugson