The Puck Stops Here

When I compared the top 10 defencemen of all time according to Daryl Shilling's Hockey Project Rating Method and Pnep's Hall of Fame Monitor, one very suprising selection was Didier Pitre. He was ranked as the fourth best defender of all time by Daryl Shilling. This is a suprising result, not so much because Pitre wasn't a good player, it is suprising because throughout most of his career, Pitre wasn't a defenceman. For example, the Legend's of Hockey biography on Pitre calls him a forward and does not mention his having played defence.

Didier Pitre's first major hockey league was in an early IHL playing with the American Soo Indians, who played in Sault Ste Marie. He joined the Soo team in 1904. By 1905/06, he was the already the top scorer in league scoing 41 goals in 22 games played. Pitre swas on the IHL first team all star that year in 1906 and again in 1907.

The next season, he left as a free agent and played with the Montreal Shamrocks in the Eastern Canada Amateur Hockey Association. He lasted there only one year before leaving to play with the Edmonton Eskimos . He stayed there only three years before jumping contract and coming back to eastern Canada where he played with the Renfrew Creamery Kings for the remainder of the year.

He joined the Montreal Canadiens the next year in 1909. He stayed for four years, before leaving for the west again. He spent a year playing with the Vancouver Millionaires. Pitre returned to Montreal the next year. In 1916, Pitre led the National Hockey Association in regular season assists and points. He scored 24 goals, 15 assists (assists in those days were one per goal and only if the official scorer thought it contributed to the goal being scored) for 39 points. He also helped lead the Canadiens to their first ever Stanley Cup. He led the playoffs in goals as well. He remained with Montreal through the formation of the NHL and into 1923. In the 1919 Stanley Cup playoffs, which were never completed due to the influenza epidemic, he led the playoffs in points. By 1921, the Canadiens had so much depth at forward, and an opening on defence due to the death of future Hall of Famer Joe Hall in the influenza epidemic, so they decided to try Pitre as a defenceman. At age 38, Pitre first regularly played as a defenceman. He played defence for his final two seasons before retiring.

Pitre played essentially his entire career as a forward. He played his final two seasons, probably his least effective seasons, as a defender. Shilling incorrectly rates his entire career as a defender. The seasons where he performed the best were seasons he played right wing. Pitre was a good scorer. If a defenceman twice led major hockey leagues in points, while playing a long career like Pitre did (~20 years) he would be one of the all time great defenders. If a player plays 18 years as a forward twice leading major leagues in points and then hangs on for two final years as a defender, he is a good player. Pitre was inducted into the Hockey Hall of Fame in 1962 to show this. However, rating him as a defender is wrong.

This is an example of the problem of calibration in hockey sabermetrics between different positions. This is a specific hockey sabermetrics problem. Exactly how does one compare players in different positions with one another? How does one deal with players who played multiple positions? I think Daryl Shilling does a better job of this then Pnep does, because Pnep has arbitrary differences between how he rates different positions. Shilling uses essentially the same formula, although it is not the same formula. For example, he credits defenders for being (for example) the top scoring defender in a season when there may be several forwards who outscore them. This is a big bonus for Pitre. As a forward, his scoring numbers get rated against defenceman. It is an unfair comparison.

Didier Pitre was a very good forward. His is a Hall of Fame forward. He held on at the end of his career as a defenceman. He is an example of the incorrect results one can get in hockey sabermetrics when rating players who do not play only one position (unless care is taken). Should Pitre get rated as a forward, he would be in the top 60 or 80 of all time probably. It is wrong to rate him as a defender. It is ever more incorrect to rate him as a defender for the time he played forward.

This shows a fundamental flaw in these sabermetric systems. A player's value should not increase if he is listed in a different position then he actually plays. If a forward plays like a defenceman, he should not gain or lose points because of the position at which he is listed. If a defender plays like a forward, he should not gain or lose points because of the position at which he is listed. He has the same value to his team despite his "official" position listing.