The Advanced Stats Thread Episode VIII:

  • Xenforo Cloud has upgraded us to version 2.3.6. Please report any issues you experience.
Status
Not open for further replies.
Do we have anyone in here who's tracked games manually at all? I'm thinking about tracking Wolf Pack this season and could use some tips to make it smooth and help me save some time.

Also, suggestions on what to track apart from shot attempts? Thinking zone exits/entries.
 
Do we have anyone in here who's tracked games manually at all? I'm thinking about tracking Wolf Pack this season and could use some tips to make it smooth and help me save some time.

Also, suggestions on what to track apart from shot attempts? Thinking zone exits/entries.
I tried a few years ago, but it's a complete pain in the ass and requires a lot more time than you'd expect. Idk how the AHL PBP logs are, but if those can be scrapped, they're a good reference to pair with your results.

I think tracking shot attempts and zone exits/entries would be perfect. I'm gonna buy AHL live at some point, so if you wanna collaborate, let me know.
 
I bet it will be very time consuming hence why i'll take any tips/help i can get :P

Only tracking one team should make it easier though.
 
Do we have anyone in here who's tracked games manually at all? I'm thinking about tracking Wolf Pack this season and could use some tips to make it smooth and help me save some time.

Also, suggestions on what to track apart from shot attempts? Thinking zone exits/entries.
No tips, just the more you do it, the better you'll get. Two screens are clutch. The more you do the same team, the easier it'll be get to recognize players without having to see what number they're wearing. You can tell by their gear, or the way they skate.

False alarm on the PBP, and the code still works. God bless.

 
Last edited:
  • Like
Reactions: ManUtdTobbe
No tips, just the more you do it, the better you'll get. Two screens are clutch. The more you do the same team, the easier it'll be get to recognize players without having to see what number they're wearing. You can tell by their gear, or the way they skate.

False alarm on the PBP, and the code still works. God bless.



Kinda figured. Any tips on how to structure the data i pull?

For zone exits do i just go controlled, failed and dump out? Guess same for zone entry.

Edit: Maybe i'll just start with successful/unsuccessful for exits/entries.
 
Last edited:
Kinda figured. Any tips on how to structure the data i pull?

For zone exits do i just go controlled, failed and dump out? Guess same for zone entry.

Edit: Maybe i'll just start with successful/unsuccessful for exits/entries.
I wouldn't lump dump outs into one category. If it's a successful chip and retrieval, it's still an effective zone exit, IMO. If the other team retrieves it, it should be a failed zone exit.
 
  • Like
Reactions: ManUtdTobbe
Yeah, i'll just simplify to start.

Anything else you guys would want tracked if i feel like it won't add too much time for me?
 
Kinda figured. Any tips on how to structure the data i pull?

For zone exits do i just go controlled, failed and dump out? Guess same for zone entry.

Edit: Maybe i'll just start with successful/unsuccessful for exits/entries.
Structure it with the future in mind. Not what's the easiest data structure to track the game, but how can you simplify it while tracking it to make future Tobbe's life easier?
 
  • Like
Reactions: ManUtdTobbe
Problem is that i have no idea how to make future Tobbe's Life easier, but i'll just figure it out i guess :D
Well, what sort of analysis do you intend to run on the data? We can work through this together. I'm assuming we want the 'simple' stuff like CF, CA, GF, GA, Scoring Chances (these will be subjective, unless you gather enough data to build an xG model).

Take a look at some of the NHL pbp sheets to get an idea of what they're tracking that people are working off Play By Play

It's basically every event and who was on the ice during the event. From here, it's all about the code to clean it up, and analyze the way you want. If you remind me later, I can post what the end data frame could look like so you can get an idea.
 
It will be basic, probably just shot attempts and zone entries/exits to start. Scoring chances is too subjective to waste time on imo.

play by play like that would be really useful to use to double check though if available for AHL.
 
As long as the PBP files don't crap out, I'll be able to compile a full sheet of fancystats (no xG, because I failed my summer project) at the end of the preseason same as last year so we can all jump to ridiculous conclusions based off bad data! [this was last year shot attempts lol Gropp]

DKslyaoWsAE1b0m.jpg
 
  • Like
Reactions: Ori
Omfg Gropp 3 CF and 16 CA last night in 7:43 :biglaugh:
That line was beyond terrible. On the topic of preseason Corsi, Pionk sure was up to his old tricks against NJD with a solid -17%. If he keeps that up I wouldn't mind playing McQuaid and sending him down to the AHL.
 
Quinn pulled the goalie with 3:45 to go yesterday down 2. Analytically speaking, I still don't think this is "optimal" (it's too late), but I also suspect that he only pulled Toke that early because we were on the PP.

Still can't really wrap my head around the benefit of pulling the goalie 3 minutes for every goal you're down, but, I think that's what the math says, with more math coming out recently that says if you're down one goal you should pull with five minutes left. Would love to see an NHL coach do that.

Other random thought... I wonder how much TOI correlates to eye-test analysis.

I mean, I know for a fact that replay time and commentator attention and eye-test are very much so positively related.
 
Status
Not open for further replies.

Ad

Upcoming events

Ad