dekedangle_rpf_mod: Hanson Bro from Slapshot (pic#7957183)
[personal profile] dekedangle_rpf_mod posting in [community profile] dekedangle_rpfanon
This is the twenty-fourth post of Deke Dangle RPF Anon, a community for all your ice hockey anon meme needs.

THE RULES


1. Mods retain the right to delete, freeze, and/or screen threads and comments.
2. Meme rules do not require warnings.
3. Respect flock. Do not repost or share information from private tumblrs, locked twitter accounts, flocked LJ posts, etc.
4. No linking fans to their real life identities.
5. No looks bashing or body shaming. This applies to players and people associated with those players and their clubs, as well as fellow fans.
6. No embedded music.
7. No embedded images.
8. No spamming the meme, whether through repeated comments or other means. 
9. No discussing trolling, individual trolls, or their efforts.





Meme rules do not require spoiler cuts. However, this layout does allow for them. Any of the following tags will create a spoiler cut when closed: <div cut>, <span cut>, <font cut>, <font color="white">

If you have any questions or concerns, please direct them to The Mod Post

Threaded View
Flat View
Top-Level Comment View

Next post opens at 5,000 comments.

Re: SCF Predictions Thread

From: (Anonymous)
https://twitter.com/omgitsdomi/status/737375687699529729

Pundits overwhelmingly in favor of Sharks, stats models overwhelmingly in favor of Pens. Any theories as to the reason for the discrepancy? Are people overestimating the Sharks (and if so, why?), or are analytics broadly failing to account for something (if so, what)?

Re: SCF Predictions Thread

From: (Anonymous)
On the Pens side, if they're using analytics from the regular season or even the previous rounds as a predictive model, they might not be taking into account Daley's injury which happened fairly late, and the fact that the Lightning straight up are weaker defensively and in goal (and also more injured) compared to the Sharks.

If the Penguins played weaker teams and had better stats, it doesn't necessarily translate to their matchup against the Sharks. Only the Caps were a fully healthy threat. The Rangers were not looking so great. And the Lightning would have been stronger with Bishop and Stamkos and a less rusty Stralman.

I don't think the Pens will be a pushover by any means, but just doing a matchup position by position (forwards, defensemen, goal) - the Sharks are both healthier and stronger. Especially the Pens blueline look to be getting pretty thin.

Re: SCF Predictions Thread

From: (Anonymous)
I believe at least two of the models for last round didn't account for the Stamkos and Stralman injuries, so I wouldn't be surprised if these don't account for Penguins injuries.

Re: SCF Predictions Thread

From: (Anonymous)
I know @IneffectiveMath's model only uses regular season data, but is sensitive to roster changes, at least for goalies. So his model would have reflected Vasilevskiy starting, and Stamkos's absence would have been baked in, since he'd been gone for a while before the season ended, right? I have no idea whether that model accounts for Schultz instead of Daley or whatever, though.

And MoneyPuck's model (I think?) uses playoff data as well, but IDK if roster changes are taken into account.

I haven't looked close enough into any of the others, although there have been several where I've tried and couldn't find documentation anywhere. I had to see several of omgitsdomi's charts before I could even figure out the difference between the purple bracket and the orange bracket (one gives a certain weight to expert opinion aka the eye test).

Re: SCF Predictions Thread

From: (Anonymous)
Lightning straight up are weaker defensively and in goal (and also more injured) compared to the Sharks

I... don't agree with this re: goaltending. Martin Jones has been great, but Vasilevsky's ECF performance was absolutely phenomenal -- most impressive player on either team.

Re: SCF Predictions Thread

From: (Anonymous)
+1
Vasilevskiy tried his hardest to make up for the Bolts defense forgetting how to do anything. I think the Sharks defense is much stronger comparatively, which gives Jones more help than the Bolts gave Vasilevskiy (and makes it easier for him to get shutouts, to be quite frank), but I think it's hard to argue that the Lightning have a weaker goalie than the Sharks when imo the Vasilevskiy+Bishop combo is one of the best in the league.

Re: SCF Predictions Thread

From: (Anonymous)
ayrt

Yeah -- defense absolutely does affect goaltending, mentally, in terms of confidence, but also in decision-making, in what the goalie can and can't trust his D to do. Lundqvist's deteriorating performance in the first round imo was mostly the result of the failure of the Rangers' defense. And Jones has had nights where he's let in tons of goals due to the failure of his D more than his own performance. And all that makes Vasilevsky's performance behind Tampa's D all the more impressive.

Re: SCF Predictions Thread

From: (Anonymous)
DA
Is there a way to do a defense-independent goalie stat at all?

Re: SCF Predictions Thread

From: (Anonymous)
ayrt

Not that I know of, but @nmercad and others do interesting stats work on goalies, there's probably links on his twitter!

Re: SCF Predictions Thread

From: (Anonymous)
Cool, thank you!

Re: SCF Predictions Thread

From: (Anonymous)
I have to agree. Jones had some great performances, but he also looked incredibly shaky at times. Vasilevskiy was spectacular from start to finish, imo.

Re: SCF Predictions Thread

From: (Anonymous)
They're all pretty close, though. Only one even goes over a 60% chance. Given how unpredictable hockey can be - just in this playoffs I know my team lost games we dominated and won games where the other time had us on the ropes - it looks like it's anyone's series. Exciting! Stressful. Exciting!

Re: SCF Predictions Thread

From: (Anonymous)
it looks like it's anyone's series.

Yep! I'm not sure a 55% chance of winning means a hill of beans in a sample size this small. Or, well, it means an even smaller hill of beans than a 65% chance of winning.

I kind of wish I didn't have a rooting interest in one of these teams, because man is this going to be some fun hockey for neutral bystanders. I also hope to have fun, but the stressed-out kind.

Re: SCF Predictions Thread

From: (Anonymous)
Yeah, even hardcore sabermetricians don't think predictive modeling means anything in the playoffs. How many games constitutes a representative sample is a pretty big question, and I like that my favorite baseball podcast defined knowing when stats stabilize as when Mike Trout is at the top of various leaderboards.

Re: SCF Predictions Thread

From: (Anonymous)
Lol, that's a pretty good description. Maybe the same thing in hockey is when Ovechkin has wrapped up the Rocket Richard.

Re: SCF Predictions Thread

From: (Anonymous)
I actually think it'd be interesting to look at how many games (and in what months, esp. in baseball where weather has a lot more of an effect) constitute the best model of the entire season, but that's a pretty complex question.

Re: SCF Predictions Thread

From: (Anonymous)
In previous seasons I could master most charts by checking to see where Sidney Crosby was and then looking for Adams or Glass. This season I haven't understood ANYTHING.

Re: SCF Predictions Thread

From: (Anonymous)
You must have been a fan of the Glass-to-Crosby scale :P

Re: SCF Predictions Thread

From: (Anonymous)
subthread OP

Yeah, I didn't mean to imply that the stats models were predicting probabilities overwhelmingly in favor of the Pens. I was just interested in the unanimity of different models being in favor of one team, versus pundits being nearly unanimously in favor of the other -- and what can be said about analytics, the eye test, etc. I can't recall another series where that was the case.

Re: SCF Predictions Thread

From: (Anonymous)
The numbers from the pundits and the numbers from the models don't even mean the same things, do they? The models are saying "this team has this % likelihood of winning," whereas the pundit numbers are "this % of people think this team will win." Right? Or am I misunderstanding? Those seem like different things.

Re: SCF Predictions Thread

From: (Anonymous)
That's right for the models, but I'm not sure how those pundit numbers are being generated and what they represent. I seem to recall reading that 17 out of 20 or so NHL pundits picked the Sharks winning, whereas only one USA Today person did.