The Advanced Stats Thread Episode IX

[Ola]

Tableau Public
Is Deangelo not as good at entries ashes sometimes made out or am I missing something? Also, Smith is #notbad

Nah, I don't think so.

Its a great research, but remember that he has looked at like what 10 games on average. We as a team have been pretty destroyed on many nights, it will of course impact the data. He is right at the average for break-outs, don't think that is bad.

Preventing entries is also to a pretty large extent team driven. As a defender you cannot step up on a forward with the aim to both stop him and take control of the puck. You need a forward back that is in a position to take the puck if the defender takes the body. Otherwise its really risky to step up and take the body because you have no idea were the puck will go flying after a hit is made.

Look at a team like CBJ that has Torts that is keen on having his defenders keep the gap and collapse fairly deep, stay in the shooting lane, all their guys are from the league average and down. Boston is the polar opposite, and has been for years, and regularly executes those plays were Ds step up while having really good support, sure enough most of their guys are really high.

There is a divide in philosophy here and many coaches aren't at all sold on having Ds step up that high, and for good reasons. There are costs to it too. You will miss your guy at times, it becomes pretty chaotic, you need to keep a 3rd man high to a much larger extent -- and so forth. The benefit is that those pucks won high up ice easily can be translated to good transition plays. Defensively it is not the best option all the time, definitely not.

Lastly, Brendan Smith is perfectly solid at LD. If Gorts could manage to deal Staal -- I wouldn't be against keeping him around at LD as a vet guy. But what are the odds of that?

 

[Ola]

Isn't there any site that lets you do a custom query on skaters nowadays? Like CF% between specific dates etc???

 

[Irishguy42]

Isn't there any site that lets you do a custom query on skaters nowadays? Like CF% between specific dates etc???

NaturalStatTrick allows this, but not from a players page. It's a bit hidden. If you go to the "Players" tab and select "On-Ice" or "Individual," and then hit "More filters" you will get the ability to set the game range by date or by team games. Then adjust the other filters to show the team/position/etc. Granted, this won't show you just the player you're looking for, but it gets you what you want.

OffsideReview also has a date range feature, but you CAN do it for an individual player, or multiple.

 

[Ola]

NaturalStatTrick allows this, but not from a players page. It's a bit hidden. If you go to the "Players" tab and select "On-Ice" or "Individual," and then hit "More filters" you will get the ability to set the game range by date or by team games. Then adjust the other filters to show the team/position/etc. Granted, this won't show you just the player you're looking for, but it gets you what you want.

OffsideReview also has a date range feature, but you CAN do it for an individual player, or multiple.

Thanks!

 

[Ola]

@Irishguy42 When you export data from these sites, you always get all info from one row in just one field. Is there a way to sort it over several columns instead or another way to tackle the problem?

Like if the info in one field (B1) look like this: [NEAL PIONK,"D","R","2018","NYR","20006","2018-10-04","NSH","NYR","5x5","16.70","0","1","0.9","0.7","13","13","15","17","16","20","0.00","56.35","44.44","46.88","40.00","0.00","3.59","3.23","2.50","46.71","46.71","53.89","61.08","57.49","71.86","11.50","17.25","0.00","0.00","5.99","92.31","94.12","95.90","23.53"]

But I want like B1 to be NEAL PIONK, B2 to be D, B3 to be R and so forth. If I want to create a graph of Pionk's CF% in all games over a season, I could do it if the data was sorted in different columns, but I don't know how to handle it when all data from one game is in the same field.

NVM, I found it. People explaining excel online are extremely pedagogic.

 

[Irishguy42]

Yeah I have no idea about exporting data b/c I don't do that

 

[1Knee1T]

@Irishguy42 When you export data from these sites, you always get all info from one row in just one field. Is there a way to sort it over several columns instead or another way to tackle the problem?

Like if the info in one field (B1) look like this: [NEAL PIONK,"D","R","2018","NYR","20006","2018-10-04","NSH","NYR","5x5","16.70","0","1","0.9","0.7","13","13","15","17","16","20","0.00","56.35","44.44","46.88","40.00","0.00","3.59","3.23","2.50","46.71","46.71","53.89","61.08","57.49","71.86","11.50","17.25","0.00","0.00","5.99","92.31","94.12","95.90","23.53"]

But I want like B1 to be NEAL PIONK, B2 to be D, B3 to be R and so forth. If I want to create a graph of Pionk's CF% in all games over a season, I could do it if the data was sorted in different columns, but I don't know how to handle it when all data from one game is in the same field.

NVM, I found it. People explaining excel online are extremely pedagogic.

Those are comma separated values. You can just save the file as a CSV and open in it Excel. Excel will take care of the formatting for you. If you can't change the file to CSV directly, you can paste that string into Notepad, Save As, and name it however you'd like with ".csv" at the end (no quotes). The brackets aren't necessary though, so delete those if they bother you.

 

[Ola]

Those are comma separated values. You can just save the file as a CSV and open in it Excel. Excel will take care of the formatting for you. If you can't change the file to CSV directly, you can paste that string into Notepad, Save As, and name it however you'd like with ".csv" at the end (no quotes). The brackets aren't necessary though, so delete those if they bother you.

Yeah, I am sure that would have worked too.

I also found a button under data that let you divide data in one column over several columns if the data for example was separated by a comma which this data was.

 

[1Knee1T]

Yeah, I am sure that would have worked too.

I also found a button under data that let you divide data in one column over several columns if the data for example was separated by a comma which this data was.

You're setting the comma as the delimiter which is basically just turning it into a CSV on the fly. Simple enough.

 

[Ola]

Many parameters have been removed from these sites, anyone know how our CF% down by 1 rank against our CF% up by 1 compared to the other teams in the league?

Under AV, we used to be among the worst in the league while up a goal or two and like average when down a goal or 2. Or even a bit above average, but I am not sure how recently I checked that, but it was a trend at least.

Quinn is different in that regard, but it would be interesting to see how much of a change that really resulted in his first season. It seems like it takes time to change these things, sometimes more than one season.

 

[Ola]

1. Anyone know if there are any data available on like how likely a team is to finish at certain positions in the standings, in light of their result the previous year?

It would be interesting to see some raw data on that. A team finishing X in the standings 1 year, is Y % likely to finish 31 the next season, Z% likely to finish 30th next season and so forth... I was thinking about this in relation to Dallas pick.

2. If there were none data like that available, it would be interesting to look into it. I have an idea on how to write it in excel using 6 baskets (i.e. looking at positions 1-5 as one group, 6-10 as another group and so forth) to get a good overview. I would create conditions assigning pts for a position moving into an basket the following season by lining up the league based standing like 15 years in a row. But it would take quite many hours for me to get programming right -- anyone have any thoughts on how you could make it faster?

 

[Mac n Gs]

1. Anyone know if there are any data available on like how likely a team is to finish at certain positions in the standings, in light of their result the previous year?

It would be interesting to see some raw data on that. A team finishing X in the standings 1 year, is Y % likely to finish 31 the next season, Z% likely to finish 30th next season and so forth... I was thinking about this in relation to Dallas pick.

2. If there were none data like that available, it would be interesting to look into it. I have an idea on how to write it in excel using 6 baskets (i.e. looking at positions 1-5 as one group, 6-10 as another group and so forth) to get a good overview. I would create conditions assigning pts for a position moving into an basket the following season by lining up the league based standing like 15 years in a row. But it would take quite many hours for me to get programming right -- anyone have any thoughts on how you could make it faster?

Different writers release their projection models each year, but none state their specific methods.

My guess is thst most are probably inherently built from aggregate player value and running thousands of season simulations to come up with a final list. They’re flawed, but I can’t think of a better way to do it.

 

[Ola]

Different writers release their projection models each year, but none state their specific methods.

My guess is that most are probably inherently built from aggregate player value and running thousands of season simulations to come up with a final list. They’re flawed, but I can’t think of a better way to do it.

Its one of those cases where I think it would be best to just extract the probability from the actual numbers.

Like if you look at 15 years, you can see that the team that finished 1st overall like had a probability of finishing 1st overall again 10% of the time, 2nd overall 2.5% of the time, 3rd overall 5%. The sample size is too small to not have the numbers being a bit all over the place -- but if you created 6 baskets you could get more meaningful numbers. Like the team finishing 1st overall 1 year has a 36% likely-hood of finishing 1-5 the following season, 18% to finish 6-10, and so forth.

I recon its not that hard to create it in excel. You just run a script where all 6 baskets for one position each year tests if it shall receive +1 if a team at the position moved into the basket in question.

 

[Mac n Gs]

Its one of those cases where I think it would be best to just extract the probability from the actual numbers.

Like if you look at 15 years, you can see that the team that finished 1st overall like had a probability of finishing 1st overall again 10% of the time, 2nd overall 2.5% of the time, 3rd overall 5%. The sample size is too small to not have the numbers being a bit all over the place -- but if you created 6 baskets you could get more meaningful numbers. Like the team finishing 1st overall 1 year has a 36% likely-hood of finishing 1-5 the following season, 18% to finish 6-10, and so forth.

I recon its not that hard to create it in excel. You just run a script where all 6 baskets for one position each year tests if it shall receive +1 if a team at the position moved into the basket in question.

Doing it that way discounts roster turnover, which is why I’d avoid it. I mean, hey, maybe you’re right. If you figure it out, try it and let us know. I’d help, but I don’t know the code you’d need.

 

[romba]

Nah, I don't think so.

Its a great research, but remember that he has looked at like what 10 games on average. We as a team have been pretty destroyed on many nights, it will of course impact the data. He is right at the average for break-outs, don't think that is bad.

Preventing entries is also to a pretty large extent team driven. As a defender you cannot step up on a forward with the aim to both stop him and take control of the puck. You need a forward back that is in a position to take the puck if the defender takes the body. Otherwise its really risky to step up and take the body because you have no idea were the puck will go flying after a hit is made.

Look at a team like CBJ that has Torts that is keen on having his defenders keep the gap and collapse fairly deep, stay in the shooting lane, all their guys are from the league average and down. Boston is the polar opposite, and has been for years, and regularly executes those plays were Ds step up while having really good support, sure enough most of their guys are really high.

There is a divide in philosophy here and many coaches aren't at all sold on having Ds step up that high, and for good reasons. There are costs to it too. You will miss your guy at times, it becomes pretty chaotic, you need to keep a 3rd man high to a much larger extent -- and so forth. The benefit is that those pucks won high up ice easily can be translated to good transition plays. Defensively it is not the best option all the time, definitely not.

Lastly, Brendan Smith is perfectly solid at LD. If Gorts could manage to deal Staal -- I wouldn't be against keeping him around at LD as a vet guy. But what are the odds of that?

Nice post explaining the difference in team D philosophies, pros and cons and how it affects the zone entry and breakouts. How would you describe AV's style?

Edit: And I see you discovered Text to Columns delimiter to clean up your data, nice job!

 

[Ola]

Doing it that way discounts roster turnover, which is why I’d avoid it. I mean, hey, maybe you’re right. If you figure it out, try it and let us know. I’d help, but I don’t know the code you’d need.

I got the below form running 10 seasons. So of teams finishing 1st overall the last 10 seasons, 33% of them finished top 5 again the following season, 56% of them finished 6-10 the following season, none finished 11-15 and the remaining 11% finished 16-20.

So its kind of interesting, I wanted to check this number due to Dallas, thinking if Zucc resigns there and we get their 1st next season, like what is the standard movement of a team at a certain position.

Dallas finished 15th overall. The last decade, if you finish 15th overall, 11% of the time you finished 1-5 the following season, none of the teams finished 6-10, 33% of the team finished 11-15, 22% of the teams finished 16-20, none finished 21-25 and a full 3rd of all teams finishing 15th overall one year finished bottom 5 in the league the next year.

The numbers at individual positions are a bit too much all over the place since the sample size was so small, might add another 10 years later but with my choppy scripts it takes more time to add things than to do it right from the get go. But it still shows that teams in the 10-20 range are all over the place the following season. I think that is the big takeaway, if you can get a 1st from a team like that 1 year down the road -- you have more or less no idea were its going to be the next season. And I did notice some of the bigger droppers when I made the cheat, I doubt many called those before they happened. Some big risers were more predictable, but the teams that crashed down the standings weren't always that obvious.

Let me know if anyone want my sheet, its full of extremely efficient scripts. This is how one field look like, I doubt anyone in Silicon Valley could have made it more efficient (OM is IF in Swedish): =(OM(B3=$C$3; 1; 0)+OM(B3=$C$4; 1; 0)+OM(B3=$C$5; 1; 0)+OM(B3=$C$6; 1; 0)+OM(B3=$C$7; 1; 0)+OM(C3=$D$3; 1; 0)+OM(C3=$D$4; 1; 0)+OM(C3=$D$5; 1; 0)+OM(C3=$D$6; 1; 0)+OM(C3=$D$7; 1; 0)+OM(D3=$E$3; 1; 0)+OM(D3=$E$4; 1; 0)+OM(D3=$E$5; 1; 0)+OM(D3=$E$6; 1; 0)+OM(D3=$E$7; 1; 0)+OM(E3=$F$3; 1; 0)+OM(E3=$F$4; 1; 0)+OM(E3=$F$5; 1; 0)+OM(E3=$F$6; 1; 0)+OM(E3=$F$7; 1; 0)+OM(F3=$G$3; 1; 0)+OM(F3=$G$4; 1; 0)+OM(F3=$G$5; 1; 0)+OM(F3=$G$6; 1; 0)+OM(F3=$G$7; 1; 0)+OM(G3=$H$3; 1; 0)+OM(G3=$H$4; 1; 0)+OM(G3=$H$5; 1; 0)+OM(G3=$H$6; 1; 0)+OM(G3=$H$7; 1; 0)+OM(H3=$I$3; 1; 0)+OM(H3=$I$4; 1; 0)+OM(H3=$I$5; 1; 0)+OM(H3=$I$6; 1; 0)+OM(H3=$I$7; 1; 0)+OM(I3=$J$3; 1; 0)+OM(I3=$J$4; 1; 0)+OM(I3=$J$5; 1; 0)+OM(I3=$J$6; 1; 0)+OM(I3=$J$7; 1; 0)+OM(J3=$K$3; 1; 0)+OM(J3=$K$4; 1; 0)+OM(J3=$K$5; 1; 0)+OM(J3=$K$6; 1; 0)+OM(J3=$K$7; 1; 0))/9

 

[Irishguy42]

I'm assuming you did that with the 10 seasons up to Vegas' first season? Impressive btw

 

[Ola]

I'm assuming you did that with the 10 seasons up to Vegas' first season? Impressive btw

Thanks!

Nope, I surgically removed Vegas.

It also took me a long time to realize why the heck some numbers were off, before I realized it was due to PHX changing name to Arizona.

I also couldn’t get the OR command to work with names, hence why the script is like 5x longer than necessary.

 

[Ola]

I added the remaining seasons post 05' lock-out:

 

[PlamsUnlimited]

Alright, someone help.

Is there a way or a website to check/look up a player and see games by season? Like, I want to see what games he played in during certain seasons. I feel like there is but I can't find it.

@silverfish help pls

 

[Mac n Gs]

Alright, someone help.

Is there a way or a website to check/look up a player and see games by season? Like, I want to see what games he played in during certain seasons. I feel like there is but I can't find it.

@silverfish help pls

Are you looking for specific game logs or just an overview? Which stats are you trying to find?

 

[PlamsUnlimited]

Are you looking for specific game logs or just an overview? Which stats are you trying to find?

It's kind of asinine but it's because I got a couple white Liberty jerseys and want to be accurate with who I put on them. So I am trying to compile the dates in 98-99 they wore them and who played in those games.

 

[Mac n Gs]

It's kind of asinine but it's because I got a couple white Liberty jerseys and want to be accurate with who I put on them. So I am trying to compile the dates in 98-99 they wore them and who played in those games.

Hockey reference should have the full data base of those games with the game logs. This sounds like what you're looking for:
1998-99 New York Rangers Schedule and Results | Hockey-Reference.com

 

[East Coast Bias]

It's kind of asinine but it's because I got a couple white Liberty jerseys and want to be accurate with who I put on them. So I am trying to compile the dates in 98-99 they wore them and who played in those games.

Here’s the game log from the 98-99 season.

1998-99 New York Rangers Team Gamelog | Hockey-Reference.com

 

[East Coast Bias]

Holy shit. Same resource at the same time.