Statistically flawed: questioning how statistics are kept in curling

After reading one of my favorite books again it made me think about how statistics in curling need to be completely rethought. “Moneyball: The Art of Winning an Unfair Game” by Michael Lewis focusses on a statistical approach used by the Oakland A’s to build a competitive baseball team using modern indicators such as on-base percentage and slugging percentage instead of inefficient and old statistics such as batting average, runs batted in and earned run averages. General manager Billy Beane’s methods were contrary to conventional baseball wisdom and the subjective analysis of players used by executives and scouts.

The thoughts that came to my mind were not about how to build a winning curling team, but about how poor the method is that we use to statistically analyze the performance of curlers during games. The method that is currently used by scoring a shot out of a possible of four points is far too subjective and flawed. No matter how much training is given to volunteer statisticians, there is no possible way to have consistency amongst these people with this method. While watching every draw of the 2009 BC men’s curling championship, I was able to see first-hand how badly stats were kept. One particular statistician, who never curled at a competitive level, consistently had lower percentages for games she was keeping stats for than everyone else. These issues are similar to the problems with the method for keeping statistics in baseball and the old methods used since the 19th century for gauging a player’s performance.

A great example of a bad statistic in baseball is the batting average which is calculated by counting the number of hits a player makes divided by the number of at-bats. This may seem to make sense but it doesn’t take into account walks or errors. If a player walks it doesn’t count as an at-bat nor does it count if an error is made allowing the player to reach base. It’s like a player’s appearance at the plate didn’t happen. If a batter has the skill to not swing at pitches outside of the strike zone and draws a walk, shouldn’t he get credit for that? When it comes to errors that allow a player reach base a statistician must subjectively decide whether it was an error or not.

To remove subjectivity in the performance of a batter, statistics such as on-base-percentage are being used more often. It’s calculated by simply dividing the amount of times a player reaches base divided by plate appearances. There is no decision that has to be made by a statistician. The player either reaches base or he doesn’t.

Currently in curling, a statistician must decide within seconds what a shot is worth out of four. Other methods even use factors of 10 or, even more ridiculously, 100. There can’t possibly be any way to have consistency when somebody has to decide whether that draw behind a guard is a three or a four.

Waiting to decide what a shot should be worth depending upon what their opponent does is even more ridiculous. For example, a skip without last rock draws around a guard and it may or may not be accessible to the other skip who has last rock. The stats person decides to wait to score the shot depending upon what the skip with last rock does. So if the final shot of the end is made and the rock is removed, the shot by the previous skip may only be a two or three. But if it isn’t, it may be worth a four. This is incredibly flawed and far too subjective. If this method was used in baseball we would wait to decide until the bottom of the ninth to decide if a two run double by the visiting team in the top of the ninth should be scored differently because the home team was able to score three runs in the bottom of the ninth.

There is an even a more insane method of scoring a shot in curling depending upon the perceived difficulty or pressure of a shot. If a curler has to make a tough double on the last shot in the tenth end of the world championship final, that shot may be worth five or even six out of four. Huh? Am I the only one who sees the stupidity in this? There can’t possibly be any objectivity this way. If this logic was used in baseball, a player who hits a grand slam home run (four runs) in the bottom of the ninth of the seventh game of the world series with his team down by three runs with two out and two strikes should be scored as five or six runs instead of four.

We need a new method that removes the subjectivity and inconsistency of how stats are kept. Any method needs to start at a binary level: either the shot was made or it wasn’t. There should be no talk about what a shot should be scored out of, deciding after an opponent’s shot what the previous shot was worth and there definitely should not be bonus points. Even if the last shot of a game is a raise quadruple take-out and it’s made it should only be scored as a made shot – nothing more.