Toronto Marlies Makeshift PDO

A great disadvantage of doing statistical analysis on leagues outside of the NHL is utter lack of detailed information available to further enhance analysis. Cam, among others – have made this point.

Something as easy as PDO can be tracked to quantify momentum, however, the American Hockey league doesn’t differentiate shots on game sheets on special teams or at even strength.

At least, not publicly, which is a shame. I would assume teams do this on an individual club basis, but that is proprietary and I can understand why any data is kept under lock and key. I won’t let this degenerate into a rant on circumstances beyond my control, instead  I tried to maximize the available data with a little rejigging.

Advertisement - Continue Reading Below

In a feeder league like the AHL, with crazy, tightly compacted schedules including three games in three nights and various players inserted and scratched in a fluid lineup, I would presume to see some wild-ish swings in save percentage and shooting percentage, the components when added together represent a quantifying of momentum. We can make up for it with a makeshift PDO.

Without the ability to filter out power play shots, the numbers aren’t a true PDO, (perhaps a pseudo-PDO or Ted Dibiase’s PDO, or proxy-PDO), looking at the numbers at a team level is still helpful.

To simulate even strength shots as close as possible I subtracted one shot on goal for each power play goal scored. I considered removing one shot on goal for each power play opportunity, but that brings in to question other factors, like power plays that lasted a short time, say less than 30 seconds as an arbitrary point, which unscientifically seems to occur quite often in a penalty filled league.

Shootout goals through four games – two extra goals for and against were all removed as were all power play and shorthanded goals.

Advertisement - Continue Reading Below

The only definite is a power play goal scored and a shorthanded goal against. Similarly, I removed five shots on goal from the opposition for each of the shorthanded markers scored against the Marlies.

There should be a control and if I could, I’d like to see what it would take to produce this for the entire AHL, but there’s still a lot of manual work that would make the endeavor very labor intensive at least for the moment.

This was a strange season in itself, with a horde of NHL players early in the AHL and then the direct result of the exodus of NHL talent once the lockout ended.

The Toronto Marlies using these crude methods had a PDO of 103.09 during the period from the beginning of the AHL season, up until the NHL returned. Propping up those numbers was a 9.54 shooting percentage and .936 save percentage.

  Shoot% Save% PDO
Pre lockout 9.54 93.55 103.09
Post lockout 6.39 91.05 97.44

In the last 17 games, Toronto has amassed a 6-8-1-2 record, slipping to second in the division, tied with Abbotsford with five games at hand. The corresponding drop in PDO follows the trends when breaking down goals for/game and goals against/game pre- and post-lockout (goals-adjusted removing shootout goals).

Advertisement - Continue Reading Below
Date Rank GP W L OTL SOL PTS PCT GF GA GF/game GA/game
1/14/2013 1 36 22 11 1 2 47 0.653 124 91 3.59 2.56
2/24/2013 2 51 27 18 2 4 60 0.588 162 139 2.56 2.82

The Marlies lost a lot of important pieces that move on to the NHL in Nazem Kadri, Matt Frattin, Matt Fraser, Korbinian Holzer, Mike Kostka and Ben Scrivens, while Keith Aucoin was lost to the Islanders via a waiver claim.

Tim Connolly was added to the roster.

Add to that injuries to Reimer that moved up Jussi Rynnas and the Marlies were left to adjust  to post lockout mode and return to the insertion/deletion of players in a watered down roster.

Both goaltending and shooting percentage took a hit over the past 17 games as a result.

This isn’t a true PDO, but with some slight adjustments, the underlying results seem to mirror the actual on-ice results at a team level. If the numbers are correct, the Marlies are have faced some bad bounces and are bound to rebound just slightly.

Advertisement - Continue Reading Below

This site uses Akismet to reduce spam. Learn how your comment data is processed.