Wednesday, November 15, 2006

Alan Ryder on first and second assists

I just found out that hockey sabermetrician Alan Ryder, of, has been writing an online column for The Globe and Mail newspaper for several months now. I don't think he appears in the printed newspaper, but an archive of his columns can be found here.

This column, from October, is on assists. Ryder notes that the NHL data divides assists into "first assist" and "second assist." The first assist is the player who presumably passed the puck to the scorer, and the second is the player who passed the puck to the first assister. Ryder argues that the first assist is more important than the second, and is a better indicator of actual skill. That's because the first assist contributed a more crucial task – passing the puck to a sniper in position to score. The second assist, on the other hand, could have been a routine pass through the neutral zone, or some such.

Last season, assists leader Joe Thornton had 72% of his assists being of the "first" variety – of Thornton's 96 assists, 69 of them were first assists. That compares favorably to the league average of around 60%. (Actually, as Javageek and "The Puck Stops Here" wrote in comments
here, there were 1.73 assists per goal last year. That means first assists must be 1/1.73, or 57.8% of all assists.)

Of the top 30 assists leaders, the Stars' Brenden Morrow had the highest percentage with 83% first assists. Jason Spezza of the Senators was the lowest, with 52%.

In this subsquent column, Ryder creates something called "Goals Created." He assigns 30% of the goal to the first assister (if there is one), 20% to the second assiter (if there is one), and the remainder (56% on average) to the goal scorer. On that basis, Jaromir Jagr led the league last year with 48 goals created.

I'm not sure how you would go about validating the percentages Ryder assigns, by correlating them to team wins or such. But regardless, it's fun to look at his list of leaders.

It's too bad Ryder doesn't appear in the newspaper itself ... he'd have a large and interested audience (the Globe and Mail is Canada-wide) and his work is great.


At Thursday, November 16, 2006 1:14:00 AM, Anonymous Anonymous said...

"That means first assists must be 1/1.73, or 57.8% of all assists."

Not necessarily. Some goals have 0 assists awarded. There could be, say, 173 goals scored with 2 assists and 27 goals scored with 0 assists. 346 assists / 200 goals = 1.73 assists per goal, yet only 50% of assists are first assists.

Obviously, I chose an extreme example, but the percentage of first assists is between 50% and 57.8%, and can not be known without knowing the distribution.

At Thursday, November 16, 2006 4:25:00 AM, Blogger Phil Birnbaum said...


You're right, thanks very much. I've struck out the original text.

At Thursday, November 16, 2006 11:03:00 AM, Blogger JavaGeek said...

I calculate assists as:
95% of goals have an assist.
80% of goals have a second assist.
which works out to:
55% first assists.

At Tuesday, August 02, 2011 4:50:00 PM, Anonymous Anonymous said...

I like the 'goals created' stat, but i think that it might be better if it where 'goals + quality scoring chances created'. this wouldn't hurt a player as much who has inferior teammates that can't finish.


Post a Comment

<< Home