Dave Hodge: Regular season data are useless in playoff predictions

[Bear of Bad News]

http://www.tsn.ca/talent/thumbs-down-to-regular-season-evidence-1.263583

His presented evidence consists of a three-game sample in the Detroit - Tampa series.

His claim is that "wild guesses" are just as good as anything: "But for every time you quote a regular-season stat to support a playoff pick, you should make a wild guess, because, chances are, the two will even out."

Good job, Dave. Cherry picker of the year.

Click to read more...

 

[hatterson]

His claim is that "wild guesses" are just as good as anything: "But for every time you quote a regular-season stat to support a playoff pick, you should make a wild guess, because, chances are, the two will even out."

Wow, that's not even really debatable, it's just demonstrably false.

Even naive, trivial predictions like using playoff seed, or regular season points, or goal differential are better than 50/50

 

[eklunds source]

There are people that get it and people that don't get it.

Statements like that are a great way for people to recognize who fits where...

 

[Pilky01]

God forbid a commentator be glib in his "thumbs down" column.

I'm not saying he is right, but I don't view regular season numbers as a be all and end all. Though I mostly base that belief on the Leafs-Sens series of the early 2000's when Ottawa would dominate in the regular season and then the Leafs would turn around and sweep them in the first round.

 

[Bear of Bad News]

God forbid a commentator be glib in his "thumbs down" column.

And God forbid we *discuss* his column on a *discussion* forum.

I'm not saying he is right, but I don't view regular season numbers as a be all and end all. Though I mostly base that belief on the Leafs-Sens series of the early 2000's when Ottawa would dominate in the regular season and then the Leafs would turn around and sweep them in the first round.

No one's ever said that regular season numbers are the be-all, end-all.

With that said, Hodge is far more wrong than right. Far more wrong, and I approve of "glibness" a lot more when the person using it knows what they're talking about.

 

[TheDevilMadeMe]

There is a legitimate case that regular season statistics are much less effective at predicting playoff success than they are at predicting future regular season success, because as anyone who watches hockey knows, the game changes at least somewhat in the playoffs.

But extrapolating anything from a sample that isn't even a full series long is absolutely brutal.

 

[Bear of Bad News]

There is a legitimate case that regular season statistics are much less effective at predicting playoff success than they are at predicting future regular season success, because as anyone who watches hockey knows, the game changes at least somewhat in the playoffs.

Without question.

On the whole, though, if you fell out of a boat and randomly hit a regular season metric, you'd probably do better than 50/50 with it on series predictions (which is contrary to Hodge's claim).

As an aside, I grew up near some choppy waters that were full of regular season metrics. Bring a life preserver when you go, friends.

 

[lomiller1]

God forbid a commentator be glib in his "thumbs down" column.

I'm not saying he is right, but I don't view regular season numbers as a be all and end all. Though I mostly base that belief on the Leafs-Sens series of the early 2000's when Ottawa would dominate in the regular season and then the Leafs would turn around and sweep them in the first round.

There is a whole branch of mathematics dedicated to finding out whether predictions are valid or not. Not only doesn't he know what the math says, he doesn't even know it exists, which is why people are rolling their eyes. He may as well be saying that "in his opinion 2+2=5". Ok so that's an exaggeration, but in principle that's pretty much what he's doing.

 

[Rob83]

Valid points raised!