Are the Capitals and Jets PDO merchants ?

[Kalopsia]

I just ran into the Caps head coach in the concourse at Rogers Place

Did you ask him if the Caps are PDO merchants?

 

[Mattilaus]

I think there are definitely reasons why combining the two is bad but the reasoning is that shots and saves are zero-sum. Across all games played in the entire NHL this year, the amount of total shots taken is constant, and the amount of goals is exactly equal to the amount of total shots minus the amount of saves. The PDO of the entire league then is by definition always 100.

As for sustaining it I don't think anyone would argue that there isn't some skill involved and that good teams can generally maintain higher PDOs than bad teams. I don't think that ability is quite as repeatable as you would think though (especially in the modern era) and the degree to which teams are able to be above that 100 mark are pretty hard capped

Well said. Like a team with a sv% of .950 and a team sh% of 20% has a PDO of 115. Is there anyone that wants to argue this wouldn't be largely unsustainable over a full season? Great teams usually have a PDO above 100, but at a certain point above 100 it becomes clear it is unlikely to continue. I think at their height last year the Canucks were around a 109 and look where they finished. There was absolutely regression. They were still a good team (I still think they are despite recent results) but their sv% and sh% were not sustainable over the long term.

It has nothing to do with luck and the people who keep parroting that is all about luck don't understand or simply don't want to discuss stats. As said earlier, it's not even an "advanced" stat. Nobody gets this upset when you say a .950 goalie will come back down to earth or a player shooting 40% will regress. They only get mad when you combine the two. Then it's all "lol go home stats nerd! My team can keep doing it! They are DIFFERENT!"

 

[tarheelhockey]

The big insight isn't saying "they won't play this great forever" it's trying to provide some sort of quantitative measure to separate out how much of a team's performance is sustainable over a longer period. That is the entire point of statistics.

Sure, but PDO obscures the point by combining unrelated factors rather than separating them. In order to formulate PDO, you have to already have possession of the key factors — then you make them harder to understand by squashing them together for no reason.

It has this dynamic in common with +/-, another stat that gets formulated by taking some useful information and jumbling it up into a non-useful, often deceptive end product.

I don't think it's mostly "luck" in the way of getting crazy bounces or offsides reviews or the other team hitting all posts (although that definitely has some part in it) and I think that lots of stats people often overstate this aspect. Rather, it is "luck" in the sense that the Caps are lucky to have basically every single player -- McMichael, Protas, Chychrun, Ovi, Strome, Wilson, PLD, Sandin, Thompson, etc. -- playing the best hockey of their careers at the exact same time.

I hear you, but there’s no causative effect at play here. PDO focuses on the idea that shooting and save percentages “should” regress to a certain point (100) which matches the league average. But nothing causes a specific player or team to have that regression. The only causative factor is performance.

Take Ovie for example. Of course his game is going to regress soon, he’s pushing 40 years old. Only someone with a very childish view of pro sports would think Ovie’s not close to the end of his run of elite performance. Age will cause his performance to drop, which will cause his scoring to drop.

That said, Ovie’s career shooting average is 13%. The NHL’s average over the course of his career is 9% — he is a 44% better shooter than the average player, as proven by nearly 1500 games of steady reliable production. If he regresses from his current 19% shooting this season, 13% will be the benchmark that he regresses to. Not 9%. PDO doesn’t recognize the difference.

This concept applies across the board. Is it “lucky” that a bunch of Caps are having career years? Maybe the question should be, is this random? Or is there something causative associated with their production, like Strome getting away from poor situations in Arizona and Chicago and finding himself on a PP with Ovechkin and Carlson instead? If that’s the case, why would he not be having the best year of his career?

I disagree with your point about the playoffs too. We see this every single year when a big upset happens and you take a closer look and it's because a goalie gets hot or has a complete meltdown. Basically the whole catalyst for the "analytics era" in hockey was the 09-10 Caps and their .977 PDO getting Halaked by the Canadiens! It definitely was a huge talking point in 2023 too with Vegas and their 105.3 PDO. Getting lucky, sure, but also just having everyone click at the same time (and the right time)!

Again though, there’s no causative relationship between shooting and goaltending. Why would you be talking about Washington’s .977 PDO instead of Halak’s .939, or Mike Green and Alex Semin combining for 0 goals in 7 games? In order to get to PDO, you have to walk right past the information that actually tells the story of the series.

 

[qc14]

Sure, but PDO obscures the point by combining unrelated factors rather than separating them. In order to formulate PDO, you have to already have possession of the key factors — then you make them harder to understand by squashing them together for no reason.

It has this dynamic in common with +/-, another stat that gets formulated by taking some useful information and jumbling it up into a non-useful, often deceptive end product.

Again though, there’s no causative relationship between shooting and goaltending. Why would you be talking about Washington’s .977 PDO instead of Halak’s .939, or Mike Green and Alex Semin combining for 0 goals in 7 games? In order to get to PDO, you have to walk right past the information that actually tells the story of the series.

I generally agree with this but saves and goals (and thus save percentage and shooting percentage) are zero sum. It also measures a very specific thing -- how efficient a team is at turning a shot difference into a positive goal difference.

Why do we use xG% which combines xG for (offense) with xG against (defense)? Why are WAR-like stats so popular? Drilling down deeper is great and what fans and GMs and writers and coaches and scouts should all do. It is also extremely helpful to have broader, more overarching metrics

 

[SeanMoneyHands]

I'd say Washington is overachieving a bit this season but they're a lot better of a team than most people gave them credit for at the start of the season. Their roster is actually pretty solid.

 

[cupface52]

Since November 30(exactly half their season so far) caps are 16th in 5v5 on ice sh%. Only Edmonton has a better record in that time, by 1 point.

 

[Alexander the Gr8]

Honestly, couldn’t care less if the Caps are PDO merchants or not, it means nothing to me. I watched the Caps lose many winnable games last year due to shit goaltending from Kuemper.

Sure, some guys have been punching above their weight offensively and LT is having a fantastic season for us. However, many guys definitely cooled down.

- Strome, McMichael just haven’t been scoring at all since Ovechkin got injured.
- John Carlson’s production dipped but it was offset by Chychrun.
- Ovechkin seems to be having a hard time catching up to everyone else in terms of conditioning. His linemates being ice cold doesn’t help either.

They win close games by keeping the score low like they did last year, but they’ve shown the ability to open the floodgates or mount comebacks regularly, which they couldn’t do before.

They’re well coached and they form a tight knit group. We’ll see what kind of team they are in April, but at least they have their playoff spot secured already.

 

[gtrower]

The problem with discounting the Caps in particular as a legitimately good team / threat in the playoffs b/c of a high PDO is that they’re currently running away with the East.

I grabbed the records and PDO stats for the current season and the last 3 before that. So 2022-present which is almost a 300 game sample size for all teams (less for Utah). I plotted points vs PDO (projected points for current season). Then made a linear trend line to determine a rough metric for how many points on average a hundredth of PDO was worth at the end of the season. It came out to about 8.73 points / PDO hundredth.

If we take the Caps’ current point total (69, nice) and current PDO (1.034) we can adjust their PDO down 2 hundredths to a much more reasonable 1.014 (would rank 6th behind Jets, TB, LAK, Wild, TOR). That would only correlate to ~10 points (47/82*8.73*2) in the current standings. That would drop them to 3rd in the East on P%, 3 points behind TOR for the 1 seed with a game in hand.

So even if we arbitrarily discount a huge chunk of their quality play by manually adjusting their PDO down (while also not doing this to other teams) they’re still likely to only regress to still being squarely in the hunt for the 1 seed in the East.

TLDR: The Caps are likely over-performing in the standings to some extent based on their quality of play as hinted at by a league-leading PDO, but at the same time their accumulation of points has been substantial enough to suggest that they’re still likely one of the best performing teams in the league to this point regardless of their PDO.

 

[oXo Cube]

Since November 30(exactly half their season so far) caps are 16th in 5v5 on ice sh%. Only Edmonton has a better record in that time, by 1 point.

Once again I'll ask, why is 5v5 important at all here? Goals count no matter how many players on the ice.

Since November 30th, the Capitals are 5th in the NHL for all strengths sh% and have a PDO of 104(1st).

 

[JaegerDice]

If every team where just as good and every goalie the same that would be the case. They Are not and that should not be hard to understand.

I can see why you would think this intuitively, but the simple fact is that it's not true over repetitive testing. PDO varies wildly year over year, among good teams and bad teams alike.

This is because sh% and sv% are uncontrollable on a game to game, month to month, season to season basis. A player cannot simply 'choose' to score on every shot or they surely would (indeed, most shots a player takes over a season do not even register as SOG), and a goalie cannot simply 'choose' to save every shot, or they surely would.

These stats just reveal that most people don't understand what numbers really are. Thinking a team HAS to play worse in the future just because they have a high shooting % or high save % is insane. They have those high statistics BECAUSE OF excellent play, whether it be superior shot selection or a goalie on top of his game. Thinking PDO means anything is putting the cart before the horse.

People were making this idiotic 'their PDO is high because they do things to make their PDO high' argument through the entirety of the 2014 Colorado Avalanche's season.

It happens every time somebody refuses to come to grips with the fact their favorite sport really is incredibly random a lot of the time. People would rather ignore the numbers and pretend that now, our team is just super smart and extra good, and THATS why they're bucking all the historic trends.

 

[JaegerDice]

Sure, but PDO obscures the point by combining unrelated factors rather than separating them. In order to formulate PDO, you have to already have possession of the key factors — then you make them harder to understand by squashing them together for no reason.

It has this dynamic in common with +/-, another stat that gets formulated by taking some useful information and jumbling it up into a non-useful, often deceptive end product.

But they are related. Surely you understand that in any given game, a team's sh% is the inverse of the opposition's sv%?

For any X team to have Y on-ice sv% over Z games, the teams they have played over those X Z games must have had a combined sh% that's inverse of Y?

The reason team PDO tends to regress toward 1.00 across the league is because all these teams are taking shots and saving shots against each other. Not perfectly weighted, obviously, since the schedule isn't perfectly balanced, but these aren't two unrelated numbers being mashed together. They're opposing products of the same events (SOG).

If every one of the 32 teams played 82 games against 82 different teams, that none of the other 31 teams played against, THEN you could argue sh% and sv% are unrelated. But within the closed system of a league with the same 32 teams, they are absolutely related.

 

[philip]

What on earth does PDO merchant mean? At this stage of the season you are pretty much where you are on merit. If you're top it's because you've been better than all the other teams.

 

[tarheelhockey]

But they are related. Surely you understand that in any given game, a team's sh% is the inverse of the opposition's sv%?

For any X team to have Y on-ice sv% over Z games, the teams they have played over those X Z games must have had a combined sh% that's inverse of Y?

The reason team PDO tends to regress toward 1.00 across the league is because all these teams are taking shots and saving shots against each other. Not perfectly weighted, obviously, since the schedule isn't perfectly balanced, but these aren't two unrelated numbers being mashed together. They're opposing products of the same events (SOG).

If every one of the 32 teams played 82 games against 82 different teams, that none of the other 31 teams played against, THEN you could argue sh% and sv% are unrelated. But within the closed system of a league with the same 32 teams, they are absolutely related.

This would all be relevant if PDO was meant to measure leaguewide events (but then, why would we need a stat to tell us that leaguewide s% + sv% = 100? We know that by common sense).

But we don’t care about leaguewide PDO. The stat is measured on the team level. And no, there absolutely is not a causative relationship between a single team’s shooting % and that same team’s save %. Those are two completely independent dynamics.

 

[thadd]

I just learned what PD0 is.

Now, how can I unlearn i

 

[Dust]

The Jets are probably always high up on the PDO list as they have Helle and he also plays like 65 games a year. Probably also explains why he looks burnt out in the playoffs.

 

[vippe]

Caps has been playing really really good this year - almost always the better team on the ice, "PDO Merchants" or not - I dont really care.

 

[43sfriends]

I can see why you would think this intuitively, but the simple fact is that it's not true over repetitive testing. PDO varies wildly year over year, among good teams and bad teams alike.

This is because sh% and sv% are uncontrollable on a game to game, month to month, season to season basis. A player cannot simply 'choose' to score on every shot or they surely would (indeed, most shots a player takes over a season do not even register as SOG), and a goalie cannot simply 'choose' to save every shot, or they surely would.



People were making this idiotic 'their PDO is high because they do things to make their PDO high' argument through the entirety of the 2014 Colorado Avalanche's season.

It happens every time somebody refuses to come to grips with the fact their favorite sport really is incredibly random a lot of the time. People would rather ignore the numbers and pretend that now, our team is just super smart and extra good, and THATS why they're bucking all the historic trends.

I hate to be the one to tell you this, but just because you don't understand an event and decide it's "random" doesn't mean it is inherently so. It's arguable nothing in the UNIVERSE is random, let alone the closed physical system of a hockey game.