PDO and Unsustainable Stats

[AC]

No my point was that neither measure is really telling you an accurate story. Critics would like the Flames more if we were scoring less often and causing more rebounds, just seems funny.

Instead the storylines should be about if/how the Flames are generating higher probability scoring chances, while limiting the other teams'.

[Street Pharmacist]

Originally Posted by Jay Random

Reasonable, but completely uninformative.

PDO does not tell you anything that the separate components, shooting percentage and save percentage, don’t tell you. In fact, PDO gives less information, because it is one number instead of two. A team could have an unsustainably high shooting percentage and very poor goaltending, in which case it would have a PDO of 100 but still be due for a fall in the standings.

In the Flames’ case, they have roughly average goaltending (as measured by SV%) and a very high shooting percentage. The shooting percentage may or may not fall down to earth; because part of it may have been produced by their deliberate tactical choice to pass the puck more often instead of taking low-percentage shots. This may result in better percentages on the shots they do take. Even if not, the fact that they deliberately refrain from taking low-percentage shots will increase their average by omission.

Then consider that the team has had a huge amount of turnover in the past two years, and many of the offensive contributors are rookie or sophomore players. This means that we really have no baseline for what kind of shooting percentage we ought to expect. Is it unrealistic to suppose that the Flames have a lot of good shooters on their team? They didn’t have a couple of years ago – but most of those players aren’t on the roster anymore. Without a baseline, there is no possible way of measuring the deviation that might be attributed to luck.

To make my position perfectly clear:

I have nothing against advanced statistics. I like advanced statistics. But PDO is not an advanced statistic. It is an artefact produced by adding two unrelated statistics together, and it is junk.

Come to think of it, it tells you less than looking at shooting% and save% do. What about the tabs with an unsustainable shooting% but terrible save%? If you have terrible goalies and middling forwards, there'll still be PDO regression even though it may be near 100. Kind of more obfuscatory tab informative.

Flames are a great example. Save% is average, shooting% is too high. Why should they go to 100 unless you think their forwards can't score???

[opendoor]

Yeah PDO never made a lot of sense to me, especially on a team level. I can sort of see the value on a player level, but it's really no more valuable than just looking at the on-ice shooting and save percentages separately.

I mean, why would a team with an 11% shooting percentage and a crappy goalie putting up .895 goaltending (for a PDO of 100.5) be considered more sustainable than a team with a 9% shooting percentage and an elite goalie putting up .925 goaltending (for a PDO of 101.5). To me the first team is far more likely to see their percentages drop than the second.

[Jay Random]

Originally Posted by opendoor

Yeah PDO never made a lot of sense to me, especially on a team level. I can sort of see the value on a player level,

The thing is, no player has a PDO – not in any meaningful sense. Skaters don’t have a save percentage, and goalies don’t have a shooting percentage. So if you try to assign a PDO number to a player, you are giving him credit or blame for part of the game that is not his job.

[transplant99]

Originally Posted by AC

No my point was that neither measure is really telling you an accurate story. Critics would like the Flames more if we were scoring less often and causing more rebounds, just seems funny.

Instead the storylines should be about if/how the Flames are generating higher probability scoring chances, while limiting the other teams'.

Which makes the entire premise of this stuff mind bogglingly dumb.

I get that if the Flames were getting out shot and out chanced by a bunch regularly (like TO last year) then it would make sense that most would expect the team to fall back...but you don't need some goofy stat to tell you that, you only need common sense and a basic understanding of how the game works.

Originally Posted by Resolute14

Right now, we have a lot of rookies scoring a lot of points - the Flames lead the league in that regard by a ridiculous margin. There is no reason why people on the outside would not expect that the Flames will indeed regress toward the mean. And there is no reason why Flames fans should be upset by this. Otherwise, you all might as well get mad at yourselves and each other for predicting we'd finish near the bottom of the standings.

All true but isn't it the case that Calgary has started more rookies than most if not all teams? I mean no one expected the Jooris revelation, but the others were expected to at least contribute somewhat....which in turn would have had the Flames leading among rookie scoring overall, which they are. Or am I missing something here? (entirely possible)

[PeteMoss]

If you look at the team Hextally graphs on war-on-ice.com, you can see the Flames shot chart is much more narrow on shots for than shots allowed on shots from about the circles in.


Also interesting - if you uncheck the 'use only three blocks' checkbox, the only spot on the ice where the Flames take more shots (5 on 5) than the league average is the just behind the slot. They also have a much better shooting percentage from that spot than the rest of the league.

Could be a coincidence or it could point to something they've worked on to get effective shots from that spot on the ice on a regular basis.

[bigpete]

I dont get what this has to do with luck. If my beer league team joined the nhl tomorrow, we would put up terrible shooting and save percentages. And it would have everything to do with sucking, nothing to do with luck.

[opendoor]

Originally Posted by Jay Random

The thing is, no player has a PDO – not in any meaningful sense. Skaters don’t have a save percentage, and goalies don’t have a shooting percentage. So if you try to assign a PDO number to a player, you are giving him credit or blame for part of the game that is not his job.

That's kind of the point. If a guy has a great GF/GA but it's being driven largely by things out of his control (i.e. goalies' sv% or teammates' shooting percentage) then that's worth considering when you're assessing his performance.

On-ice percentages don't have a lot of value on their own, but they can help tell you what is driving a player's results.

[Jay Random]

Originally Posted by opendoor

That's kind of the point. If a guy has a great GF/GA but it's being driven largely by things out of his control (i.e. goalies' sv% or teammates' shooting percentage) then that's worth considering when you're assessing his performance.

On-ice percentages don't have a lot of value on their own, but they can help tell you what is driving a player's results.

Well, only if you think GF/GA is an individual stat in the first place. It is a dangerous business, deliberately making one mistake in order to correct another.

[Rick M.]

A few years ago plus/minus was all the rage. Then sophisticates began to ridicule the ludicrous conclusions this stat gave rise to. I don't see how PDO is any better. Trying to extract meaning where there is none always leads to error.

[monkeyman]

PDO numbers rage between ~1.03 and ~0.965 in any given season. In fact almost no team finished with a perfect 1.0. So why isn't a 1.03 and .965 sustainable over a season when they're within a proven historical range?

[opendoor]

Originally Posted by AC

No my point was that neither measure is really telling you an accurate story. Critics would like the Flames more if we were scoring less often and causing more rebounds, just seems funny.

Instead the storylines should be about if/how the Flames are generating higher probability scoring chances, while limiting the other teams'.

But you're vastly overestimating how many lost rebound opportunities are being missed because of the Flames' higher shooting percentage. I won't bore you with the numbers, but if the Flames' shooting percentage was 2% lower, they'd have probably generated 1-2 more shots over the season to date as a result of rebounds. It basically has zero effect.

[d_phaneuf]

Originally Posted by monkeyman

PDO numbers rage between ~1.03 and ~0.965 in any given season. In fact almost no team finished with a perfect 1.0. So why isn't a 1.03 and .965 sustainable over a season when they're within a proven historical range?

no one is saying it's not possible, just that it's unlikely

being in the range means little without knowing the mode and medium

[Enoch Root]

Originally Posted by opendoor

But you're vastly overestimating how many lost rebound opportunities are being missed because of the Flames' higher shooting percentage. I won't bore you with the numbers, but if the Flames' shooting percentage was 2% lower, they'd have probably generated 1-2 more shots over the season to date as a result of rebounds. It basically has zero effect.

But looking at rebounds as a direct result of a failed shot is too narrow of a view IMO.

I think the original point was that, when you score a goal, it ends your possession situation. If you don't score, your possession (zone time) continues (usually).

It doesn't necessarily mean rebounds, but continued zone time is not unusual after a failed shot.

Continued zone time is non-existent after a goal.

Based on those facts, failed shots must result in more follow-up shots than successful shots (goals) do.

[d_phaneuf]

I dont think anything really useful is going to come out publicly for awhile

I imagine most teams are already using things like SportVu, but until it can be tracked in shot quality/scoring chances etc.

for example look at the Sedins, they will at times have a shift with 1.5-2 minutes of zone time, but because they are looking for the perfect shot it might end up with 2 shot attempts, both A+ quality chances

yet the stats now would say a shift where someone took a shot from center ice on the goalie, the puck bounced to the corner off his pads and was thrown back at the net for an easy glove save, is essentially the same shift in advanced stats

it can't work like that

[Racki]

Originally Posted by Enoch Root

Whether or not they can sustain that is irrelevant because they don't have to.

They can regress significantly for the remainder of the season and they will still make the playoffs. So who cares?!?

Yes! This is what so many of the advanced stats guys are ignoring. You are allowed to have streaks where you get a little puck luck, shooters get hot or goalies steal a few games.

Just because you slide back to the pack afterwards, you still get to keep the points you banked earlier during your "lucky" stretch.

I also find it frustrating that they seem to allow no room for improvement. Teams and players have been known to get better during a season.

[monkeyman]

Originally Posted by diane_phaneuf

no one is saying it's not possible, just that it's unlikely

being in the range means little without knowing the mode and medium

My argument is that regardless of average, a PDO of 1.03 and 0.965 are not outliers, but rather high and low ends of the spectrum. Not historically proven unsustainable.
Actually, the Oilers are historically low at ~.960 so they could be an outlier or just setting a new historic lowe.

[belsarius]

Originally Posted by Jay Random

I have nothing against advanced statistics. I like advanced statistics. But PDO is not an advanced statistic. It is an artefact produced by adding two unrelated statistics together, and it is junk.

Quoted for truth. I could just as easily add PP% and PK% together and call it a stat and compare it to teams. I'm sure there will be a correlation to better teams having a higher number but they are really unrelated and that number means nothing.

PS thanks for the new sig!

[Resolute 14]

Originally Posted by belsarius

Quoted for truth. I could just as easily add PP% and PK% together and call it a stat and compare it to teams. I'm sure there will be a correlation to better teams having a higher number but they are really unrelated and that number means nothing.

PS thanks for the new sig!

That is, in fact, already done, albeit informally. Media types have long liked to add PP% and PK% together and call a team good on special teams if the sum exceeds 100.

[Resolute 14]

Originally Posted by monkeyman

My argument is that regardless of average, a PDO of 1.03 and 0.965 are not outliers, but rather high and low ends of the spectrum. Not historically proven unsustainable.
Actually, the Oilers are historically low at ~.960 so they could be an outlier or just setting a new historic lowe.

You don't understand what outliers are then. Case in point, over the last five full seasons, only one team managed a PDO of 1.03 or higher. That is one instance out of 150 team seasons. Only four posted a PDO of 1.025 or higher. Only eight managed 1.02 or higher. These are outliers.

So when people say the Flames' PDO of 1.031 is unsustainable, they do so because they do not believe the Flames are likely to finish at the extreme high end of the scale.

Bluntly, the Flames are not likely to finish at the extreme high end of the scale.

But that does not mean it can't happen. Colorado did it last year under the same circumstances.