The Toronto Maple Leafs' looming PDO demon (what is PDO?)

Cam Charron
February 20 2013 01:26PM

"PDO", which doesn't stand for anything, can explain how hockey teams performed in a short amount of time.

It is the simple addition of a team's even strength shooting percentage and a team's even strength save percentage. Over the long run, it's been observed that the "PDO number" comes closer to 1.

It is also a very counter-intuitive concept, since it assumes that the value of every shot is equal, that teams have equally-good shooters and goaltenders. That said, teams that start the season with high PDOs tend to finish the season with lower PDOs. As of Monday morning, there were nine teams that had their goaltenders combine for .930 even strength save percentages or higher. After 82 games last season, only two teams managed that.

The overall worry here is that the Toronto Maple Leafs, as of Monday, had the third highest PDO in the National Hockey League. Their shooting percentage was 10.1% and their save percentage was .938. Add those two numbers together and you get 1.039. The highest that Randy Carlyle's Anaheim teams finished under him since we've had PDO data is 1.011.

I won't bug you much with the math behind regression, but I thought to bring out some real-world applications. I dug back through the last three seasons to find the teams with the highest PDO numbers after the first month:

Season Team Shooting % Save % PDO
2010 Washington 11.6% 0.927 1.043
2010 Edmonton 11.2% 0.926 1.038
2011 Dallas 9.9% 0.936 1.035
2012 NY Rangers 9.2% 0.943 1.035
2011 Boston 8.2% 0.952 1.034
2012 Dallas 8.6% 0.946 1.032
2012 Washington 10.9% 0.923 1.032
2010 Colorado 9.6% 0.935 1.031

There's a scattering of teams there in quality. The 2011 Boston Bruins won the Stanley Cup, and the 2010 Edmonton Oilers ended up in 30th and got the first overall pick. The 2010 Washington Capitals won the Presidents' Trophy, but got beaten in the first round. The 2011 Dallas Stars and the 2010 Colorado Avalanche became poster boys for regression: both teams had a strong start, and the Stars fell out of the playoffs on the last day. The Avalanche still managed to finish 8th.

Since the Leafs have played 17 games, here are those teams' records after the first 17 games:

After 17 W L OTL Points/48
2010 Capitals 10 3 4 67.8
2010 Oilers 8 8 1 48.0
2011 Stars 10 7 0 56.5
2012 Rangers 10 4 3 64.9
2011 Bruins 11 5 1 64.9
2012 Stars 11 6 0 62.1
2012 Capitals 10 6 1 59.3
2010 Avalanche 12 3 2 73.4

It's funny how similar PDOs can have such different effects on some teams. The Oilers had better shooting and save luck than six other teams here, but they were .500 after 17 games. The point of this is that a high PDO can make good teams look great, and bad teams look good.

Here are those same teams' records from Games 18 through 48 in that same season:

Games 18-48 W L OTL Points/48
2010 Capitals 20 9 2 65.0
2010 Oilers 8 19 4 31.0
2011 Stars 19 7 5 66.6
2012 Rangers 21 8 2 68.1
2011 Bruins 16 9 6 58.8
2012 Stars 14 15 2 46.5
2012 Capitals 16 13 2 52.6
2010 Avalanche 15 12 4 52.6

Only the 2011 Stars and 2012 New York Rangers improved their records. The Stars implosion would come later in the season, but the purpose of today's post is to see whether the Leafs have banked enough wins with an elevated team save percentage to make the playoffs. It's certainly possible: they have 31 games left and are three points clear of the 8th-place team. Last season with 31 to go, they were just two up of the next team, but the Capitals, who were chasing them, held a game in hand.

Around that time last winter I wrote a post about the Leafs' playoff fortunes, saying "Should the shooting [percentage] falter, I'm not convinced the team has the defence or goaltending to stay in a playoff spot."

A commenter named 'advancedstatsareafraud', a harbinger if there ever was one, wrote the following under that post:

When the leafs, with excessive PDO and way too low Fenwick, make the playoff - then finally people will come to realize that advanced stats are a complete fraud. The leafs laugh and are better then all this "regression to mean" talk.

I couldn't help but remember that comment when I saw 'leaferfan' post something similar on the post at NHLNumbers this week that showed the Leafs' in third place in the NHL in PDO:

I dont' care about this PDO nonsense. Mclaren and Orr will pull the leafs into the playoffs kicking in screaming with a few good punches along the way.

The bad news is that the Leafs are currently outplaying their means. They're 5-2 in one-goal games and 5-5 in contests decided by two goals or more. There's every indication that one shot or save that goes the wrong way could result in changing a game.

Anyway, total up the above teams' records and look at the key: points per 48 games:

  Wins Losses OTL Points/48
After 17 82 42 12 62.1
Games 18-48 129 92 27 55.2
After 48 211 134 39 57.6

It doesn't look like much of a drop considering nearly half the season has been played already. Over 82 the difference would be much greater, but it's still enough to change the Leafs' overall fortune. Toronto are at 10-7-0, on pace for 56.5 points over 48 games. It's expected that something around that number could be the number to get into the playoffs. Right now, I wouldn't bet money on the Leafs' maintaining their current level of winning, based on what has happened to teams in the past. They're right on the cusp where the four or five points that could be explained by PDO regression to the mean could put them on the outside-looking-in.

Right now, the Leafs are relying on out-scoring the opposition and not out-shooting them. While they've generally had more scoring chances than the opposition, it's been found that over time, scoring chances sync up with shot differential statistics. Right now the Leafs are benefitting from some good puck luck, and that's driving their record moreso than coaching, leadership, size or being an improved hockey team.

There is still room to improve. There is an unsigned, play-driving centreman ripe for a trade. The Leafs' best two-way centreman has played below 15 minutes in five of the last six contests. Maple Leafs defenceman Jake Gardiner is playing with the Marlies, and Marlies defenceman Korbinian Holzer is playing with the Maple Leafs. Reliable face-off and defensive centreman David Steckel is a healthy scratch nearly every game. Even with the winning record, the Leafs' underlying performance numbers paint a chilly picture.

So, no, if the Leafs keep doing what they're doing, I don't expect them to win 10 out of 17 games to make the playoffs. Teams over a 1.030 PDO in the last three seasons have had an observed drop in points per 48 games of seven. That would put Toronto on a 52-point pace for the remainder of this season with all the wins they have banked. Would that be enough to make the playoffs? Not unless the team improves.

63811cbf517d2d685ea09e103488ea3a
Cam Charron is a BC hockey fan that writes about hockey on many different websites including this one.
Avatar
#1 RexLibris
February 20 2013, 03:47PM
Trash it!
0
trashes
Props
0
props

I've ballparked the Leafs in the same territory that I expect the Jets, Flames and Oilers to occupy at season's end - somewhere south of the cutoff line but likely not in the top end of the draft order.

They have enough pieces to keep from being entirely terrible. I'm not certain that they can go forward without first moving back, though.

Excellent analysis, by the way. Discriminating the good from the "she ain't pretty she just looks that way" teams is a valuable exercise. Too bad management (of any team) never seems to take the lessons to heart.

Avatar
#2 Not Norm Ullman
February 20 2013, 06:06PM
Trash it!
0
trashes
Props
0
props

NON-BELIEVER! BURN HIM! *

* also, a canucks fan.

Avatar
#3 FlareKnight
February 20 2013, 10:42PM
Trash it!
1
trashes
Props
0
props

I'll say right now I hate math.

In the end it will be the play not the numbers that decide where this team finishes.

Yeah, I'm just going to not dive too much into this. Leafs get their top line winger, top 6 winger, top 4 defenceman, and their #1 goalie back into the lineup and they'll be fine.

Avatar
#4 BCapp
February 21 2013, 06:14AM
Trash it!
0
trashes
Props
0
props

Not Norm Ullman: I've missed you!!!

In regards to ther article I am just hoping he short season will allow teams to stay farther from regression. Smaller sample size should lead to a larger variance and all that...

Avatar
#5 lb71
February 21 2013, 07:34AM
Trash it!
0
trashes
Props
0
props

This season may cause an anomaly on all kinds of statistics, advanced or traditional. Year over year comparisons are only useful if the conditions are comparable. A compressed scheduled with more average games per week will have an impact on these stats. I'm not sure a 17 game comparison from a 48 game season vs an 82 game season would be valid. A comparison to the 1994-95 season would be more apt, but I doubt the stats are available. It may be necessary to wait out the season and see what these stats look like.

Avatar
#6 Pension Plan Puppets
February 21 2013, 08:32AM
Trash it!
0
trashes
Props
0
props
FlareKnight wrote:

I'll say right now I hate math.

In the end it will be the play not the numbers that decide where this team finishes.

Yeah, I'm just going to not dive too much into this. Leafs get their top line winger, top 6 winger, top 4 defenceman, and their #1 goalie back into the lineup and they'll be fine.

Where do you think the numbers come from? Thin air? No, they come from the play on the ice.

Why the fear of learning something?

Avatar
#8 RexLibris
February 21 2013, 05:19PM
Trash it!
0
trashes
Props
0
props
Pension Plan Puppets wrote:

Where do you think the numbers come from? Thin air? No, they come from the play on the ice.

Why the fear of learning something?

Jose Canseco believes that gravity is to blame for advanced analytics. He has a Twitter account, so he must be correct.

Avatar
#9 i_miss_mogilny
February 22 2013, 03:46AM
Trash it!
0
trashes
Props
0
props

If the Leafs management thinks they can secure a playoff spot riding unproven AHL-ers playing over their heads, then they're in for a rude awakening in a few weeks.

Avatar
#10 BCapp
February 22 2013, 06:28AM
Trash it!
0
trashes
Props
0
props

I miss mogilny

If gardiner comes back we actually won't have AHLers playing over their head. We'll have a top 4 D corps and a top 6 forward corps of NHLers. A sheltered 3rd forward line of top prospects (Kadri +Frattin) + a European veteran (komarov) and a 3rd pairing of AHLers

Avatar
#11 Topshelf
February 22 2013, 07:37AM
Trash it!
0
trashes
Props
0
props

It's nice to see the Leafs off to a good start this season. I thought they would struggle out of the gate and perhaps improve as the season went along, given more time to implement what Carlyle is asking of them. It's a completely different system than what Wilson employed, so some bumps are/were to be expected.

What I understand from reading the article is that the Leafs haven't been playing great to start the season and have essentially been lucky to have such a good record so far. Fair enough.

That said I would expect the team to play better as things progress, meaning they still MIGHT continue to win more than they lose down the stretch, they'll just have to win differently.

They'll have to play better overall and not give up so many shots/game. The shooting % will come down, so they need to start getting more shots. OK.

In a nutshell the Leafs have been lucky to have the record they do so far, and stil need to make improvements, makes sense.

Avatar
#12 Topshelf
February 22 2013, 07:50AM
Trash it!
0
trashes
Props
0
props

@Cam Charron

Since the Leafs have a relatively new coach, doesn't it make some sense that they MAY get better, or improve over time.

They were outshot against Buffalo last night, but the goals were on good quality chances thanks mostly to Kessel.

I suspect there will be some bumps ahead, but what you're saying is that despite the good record so far, the Leafs will have to play better if they're to make the playoffs.

Avatar
#13 lb71
February 22 2013, 08:15AM
Trash it!
0
trashes
Props
0
props

This year's 48 game season versus the first 48 games of an 82 game season may not be a valid comparison.

You are trying to extrapolate the performance of games 18-48 using statistics from seasons where games 18-48 were part of an 82 game season. This season's games 18-48 are the last 31 games of the season. In prior years, games 18-48 fall into the mostly the first half of the season.

There are a lot of factors that make this season an anomaly: games played on a more frequent basis, the quality of opponents are different (no cross divisional games), the meaning of games (we've hit the one-third mark already, in an 82 game season it would be the one-fifth mark), injuries/slumps on key players.

What might be more valid is looking at percentage of games played. Since they have played 35% of games, look at the first 29 games of the schedule versus the final 53.

Avatar
#14 Badger
February 22 2013, 09:06AM
Trash it!
0
trashes
Props
0
props

I really question the usefulness of PDO honestly.

Boston's PDO in their 2010-2011 season was 1027. Did that regress to the mean, resulting in an epic collapse? No. They won the cup.

PDO also fluctuates at a ridiculous rate due to small sample size. The Leafs' PDO was 1.039 after the Ottawa game. After three games it fell to 1.031. Yet they decisively won 2 of those 3 games.

Here's a good post on hfboards about PDO and how dubbing it a "luck" metric is misleading:

http://hfboards.hockeysfuture.com/showpost.php?p=53399873&postcount=20

Avatar
#15 MaxPower417
February 25 2013, 08:46PM
Trash it!
0
trashes
Props
0
props

PDO is an odd stat, because it is useful for quantifying luck, but it can also be impacted by skill. The 2010-2011 Bruins Badger mentioned for example, were probably a good enough team to exceed 1000 PDO without the aid of luck.

That said, the Leafs are not the 2010-2011 Bruins. I made this same argument about the Leafs on another board and caught a lot of flack for it, though I didn't have the historical evidence to back up my claim.

Luckily for us Leaf fans, in a short season, anything can happen and if they keep up this hot streak to bank 3 or 4 more quick wins, regular regression very well not catch up with them by seasons end. Not to mention the traditional hockey arguments of a young team, under a new coach improving, the return of important injured players and the hope that Kadri and others playing well on limited minutes are given more ice time.

Comments are closed for this article.