February 11, 2009

Team Depth Chart

L1 = top line points per game
L2 = second line points per game
L3+L4 = 3rd & 4th line points per game
PP = power play goals for while on the ice per game
EV = even strength goals for while on the ice per game
OVRPCT = L1/(L1+L2)

I created this a few weeks ago (updated it this week) to find out what was wrong with Ottawa:
Many have criticized Ottawa's top line for their terrible season, however the chart above tells a different story: Ottawa has no depth. There are two reasons the depth players might be doing worse than average: bad offensive defensemen, or lack of quality depth forwards. For example, Chris Neil has got 6 points in 44 games ($1.2M), that is simply unacceptable (that's with PP time). Either way there is only one person to blame for a poorly balanced team: the GM.

The chart helped highlight a number of other interesting facts:
- Toronto's second line is more productive than it's first unit (and that's with less ice time)
- Montreal could probably better utilize it's star players: Why does Plekanec get so much ice time?
- I'm sure others will see other things...


Sunny Mehta said...

Interesting idea.

First off, how did you determine which players constitute L1, L2, etc? Did you go by ice time? What if certain teams mix and match lines? How did you determine which player to put where? Also, what if different players play on the first PP unit than the first EV line, etc?

Secondly, I'm not sure points per game is necessarily the best measure. It seems like there would be a ton of redundancy/synergy (if a player scores a goal, of course his linemates are the most likely to record the assists - thereby exponentially increasing the results for lines that score). Also, it seems strange to have a goal and a secondary assist be given equal weighting.

Third, I'm not sure that even just using goals would be all that predictive. I imagine there's a ton of variance in dealing with goal scoring rates in a sample size of 50-60 games.

For example, I too looked at Ottawa's situation a couple months ago. Like you, I realized that the first line wasn't the problem. However, I also noticed that players like Antoine Vermette had ridiculously low shooting percentages (much lower than would be expected based on not only their shot distances, but also their career/previous season results).

JavaGeek said...

it's based on ice time per game with a cushion for injuries/changes on the line.

I basically add up the top players games played until I hit team's games played * 3 (+ cushion)

I don't see a problem with exponential growth of points in the data, the results look pretty reasonable.

T.C. said...

Yeah, I see Washington is not that balanced either yet...

Great table.

The Falconer said...

re: Ottawa's no depth. Back around Christmas time I noticed that Ottawa's 2nd line type players had a brutal Shot % so far on the season. They just didn't convert enough of their chances for whatever reason.