Some Draft Stats

[Vaasa]

The recent Mike Brown acquisition thread featured this post by Jux:

I couldn't give a **** what we gave up for him. A 4th has like a 5% chance of ever becoming an NHLer. What I care about is that Brown is a useless facepuncher doesn't belong in the NHL and who is worse at hockey than Burish and takes up a contract slot and cap space that would be better utilized elsewhere. The only way I could possibly be okay with this if it's because DW thinks he can trade Burish.

The part in bold made me wonder where the stat came from and whether it was true or not. After hunting around on the web and not finding decent stats on draft success, I decided to do some analysis myself. And since I'm a glutton for punishment and wanted to try and learn some new skills, I decided to do it in Excel.

~2700 lines of code and the copying of 15 years (1995 - 2010) worth of draft data from HockeyDB.com later, I have some early stats and may be able to answer some questions folks may have. I only have the data broken down by draft round and year so far, not by team, so I can't give you team info (yet).

But here a few stats to start things off (again, for the time period 1995-2010):

- Defensemen and wingers consistently make up around 30% of the draft picks in every round, but centers drop from around 30% of the first round to 20% or less for each subsequent round.

- Goaltenders generally make up 10% or less of the draft choices in every round but the 5th and 6th where they pop up to around 14%.

- Your chance of getting a player who has put in 120 NHL games or more by draft round are roughly: 1st = 42%, 2nd = 25%, 3rd = 18%, 4th = 13%, 5th = 10%, 6th = 10%, 7th = 10%, 8th = 11%, 9th = 9%

- Your chance of getting a consistent (over their entire career so far) 20-goal per 80 games played forward is almost 16% in the 1st round. But only 1.4% in the 4th round.

- Your chance of getting a consistent (over their entire career so far) 30-point per 80 games played defenseman is almost 5.5% in the 1st round. But only 1% in the 4th round.

- The best draft year for the % of players out of the total draft to get at least 1 NHL game was 1998, when just under 49% of all draftee's got at least one game. It's the only year under 50% and most years the total % who get 0 games is closer to 60%.


Anyway, that's a bit of early info. I'm still playing around with the data and am thinking I may try and create some simple infographics for the data.

If anyone has any questions or comments, please add them here. If you have ideas for data analysis I might do, put those here as well. I'm not saying I'll do it, or when I'll do it, but you never know.

Thanks

Click to read more...

 

[Led Zappa]

Interesting. Looking forward to more data. Thanks for putting in the time.

 

[Juxtaposer]

Good stuff. I pulled the 5% number out of my butt, but the sentiment was there.

 

[Alwalys]

Good stuff. I pulled the 5% number out of my butt, but the sentiment was there.

At what percentage does the sentiment change?

 

[magic school bus]

As you can see, there's little difference between a 4th and 7th round pick. I'm pretty sure the org is aware of this with how often we've seen these "worthless" 7ths throw into our end of deals.

Burish for a 7th and we're better off

 

[Led Zappa]

As you can see, there's little difference between a 4th and 7th round pick. I'm pretty sure the org is aware of this with how often we've seen these "worthless" 7ths throw into our end of deals.

Burish for a 7th and we're better off

But Pavelski.

Now excuse me, I'm off to get my lottery tickets.

 

[OrrNumber4]

Looking forward to seeing it.

It might show a better picture if you look at position vs. round, at least for the first 60 positions.

 

[magic school bus]

But Pavelski.

Now excuse me, I'm off to get my lottery tickets.

"worthless" picks to other teams, but not to us

 

[SJeasy]

Vaasa,

Thank you for the work. I look forward to more if you have the time.

A few notes:

I used 80 games as a cut line for an NHLer. It cut off failed prospects for the most part while including reserve players. I had heard 80 from somewhere else. TSN (Cullen) uses 50 games as the cut line in its analysis.

30 pts/year seems odd for a dman. 20 pts usually signifies a puckmover while 40 pts is a PP point man. Your choice.

You may also want to break down the first round. Within the round, I used 1-5, 6-15 and 15-30. There was huge difference between the spots within the round.

Another one of the hockey draft analysts generalized that rounds one and two were for NHLers, three through five were for the AHL, while rounds six and seven were hoping to get lucky. Essentially they said, it was a bust not to get an NHLer out of picks one through sixty and a bust not to get an AHLer out of picks sixty-one through one eighty.

When I did my analysis my numbers came up higher but that could easily be explained by the setting of the 120 game benchmark. By eyeball, it seemed that there was a real fall off from round six to round seven.

 

[hockfan1991]

Vaasa,

Thank you for the work. I look forward to more if you have the time.

A few notes:

I used 80 games as a cut line for an NHLer. It cut off failed prospects for the most part while including reserve players. I had heard 80 from somewhere else. TSN (Cullen) uses 50 games as the cut line in its analysis.

30 pts/year seems odd for a dman. 20 pts usually signifies a puckmover while 40 pts is a PP point man. Your choice.

You may also want to break down the first round. Within the round, I used 1-5, 6-15 and 15-30. There was huge difference between the spots within the round.

Another one of the hockey draft analysts generalized that rounds one and two were for NHLers, three through five were for the AHL, while rounds six and seven were hoping to get lucky. Essentially they said, it was a bust not to get an NHLer out of picks one through sixty and a bust not to get an AHLer out of picks sixty-one through one eighty.

When I did my analysis my numbers came up higher but that could easily be explained by the setting of the 120 game benchmark. By eyeball, it seemed that there was a real fall off from round six to round seven.

I believe 15 to 30 has hurt the sharks the most for a while. Best player until hertl has came around had probably been goc.may turn out to be Coyle but time will tell. Strum many years back. But there ability to find players like pavs by those stats would be in lucky category and wingels etc. has helped. Redo that 2003 draft and pavs may have been in top 20. We have been fortunate as well

 

[magic school bus]

Even with 160 games, there's not a huge difference

 

[Vaasa]

Looking forward to seeing it.

It might show a better picture if you look at position vs. round, at least for the first 60 positions.

Can you expand on this? Look at what for position vs round? Games played? Points?

 

[Vaasa]

It may help if I give you all an idea of what I have done data-wise. I have built code that goes through the 15-years of data I got from HockeyDB and it builds up these tables:

Draft picks per round (and total)
- # of D, G, C, W, and Other picks per round
- Total picks per round
- # of D, G, C, W, and Other picks per draft year
- % calculations for both round and year (1st and 3rd items above)

Games played per round
- Each round broken into brackets of 0, 1-19, 20-59, 60-119, 120-299, 300-499, and 500+ (Brackets arbitrarily chosen by me, NOT easy to change, sorry)
- # of picks that fit into each bracket above per round
- # of picks that fit into each bracket per year
- Totals across all drafts I have

Goal Average per game for forwards and defense (separate tables)
- broken into brackets again (<.06, .07-.12, .13-.18, .19-.24, .25-.31, .32-.37, .38-.49, >.49)
- Brackets were chosen because for an 80-game "season" .06 = 5 goals, .13 = 10 goals, .19 = 15 goals, .25 = 20 goals, etc.
- Like other tables, this is broken down by round, by year, and a total

Points Average per game for forwards and defense (separate tables)
- Same as the above excepts there are 3 extra points brackets.


So that gives you an idea of what I have so far. I'm sort of checking and looking over what the results are now. I may pull together a few graphics (simple pie charts maybe), and I'm thinking of trying pull out some data by team (starting with the Sharks of course).

Questions, ideas, or suggestions are welcome.

 

[Vaasa]

Vaasa,

Thank you for the work. I look forward to more if you have the time.

A few notes:

I used 80 games as a cut line for an NHLer. It cut off failed prospects for the most part while including reserve players. I had heard 80 from somewhere else. TSN (Cullen) uses 50 games as the cut line in its analysis.

30 pts/year seems odd for a dman. 20 pts usually signifies a puckmover while 40 pts is a PP point man. Your choice.

You may also want to break down the first round. Within the round, I used 1-5, 6-15 and 15-30. There was huge difference between the spots within the round.

Another one of the hockey draft analysts generalized that rounds one and two were for NHLers, three through five were for the AHL, while rounds six and seven were hoping to get lucky. Essentially they said, it was a bust not to get an NHLer out of picks one through sixty and a bust not to get an AHLer out of picks sixty-one through one eighty.

When I did my analysis my numbers came up higher but that could easily be explained by the setting of the 120 game benchmark. By eyeball, it seemed that there was a real fall off from round six to round seven.

I just picked 30 points as a random data point. As you can see from above I can derive different point levels as long as they fall within my brackets.

Hope that helps.

 

[OrrNumber4]

Can you expand on this? Look at what for position vs round? Games played? Points?

What I mean is, that instead of looking at how many NHL players are produced by rigid draft round, do it by a range of draft positions. The recommended 1-5, 6-12, 13-25, 26-40, 41-60, and then by rounds.

Games played is the best metric, since it holds through for eras. Points will always be an issue, due to era. 20 points was probably a puck-mover around 2001-2003, but I don't think it was one between 2005-2009.

 

[Vaasa]

What I mean is, that instead of looking at how many NHL players are produced by rigid draft round, do it by a range of draft positions. The recommended 1-5, 6-12, 13-25, 26-40, 41-60, and then by rounds.

Games played is the best metric, since it holds through for eras. Points will always be an issue, due to era. 20 points was probably a puck-mover around 2001-2003, but I don't think it was one between 2005-2009.

OK, I already have it by round. I don't have anything that works with draft position at the moment. Depending on the feedback I get, I may try and make that one of the future expansions to the code.

Thanks

 

[Vaasa]

Hehe, and the reason why I said these are preliminary numbers just became clear. I see I have some problems with the goals (and probably points) averages per round. Things are skewed to the low side and players like Tavares aren't showing up. So take any info related to points or goals above with a grain of salt. The draft picks and positions data seems solid so far though.

 

[OrrNumber4]

Hehe, and the reason why I said these are preliminary numbers just became clear. I see I have some problems with the goals (and probably points) averages per round. Things are skewed to the low side and players like Tavares aren't showing up. So take any info related to points or goals above with a grain of salt. The draft picks and positions data seems solid so far though.

You will also want to limit the year to 2007 or 2008. When determining "consistent" players, players drafted so recently are going to have skewed statistics.

 

[Vaasa]

You will also want to limit the year to 2007 or 2008. When determining "consistent" players, players drafted so recently are going to have skewed statistics.

I know. That's why I have data broken out by year, round, AND totals. One, it helps me check that the code is working correctly by making it easier to spot problems. But I was thinking that when I get around to designing the visuals that I may make some interactive charts where I could select a range of years (and or rounds) and have the graphics update to show just that set of data. I've done similar stuff in the past with Excel, but you have to get your data correct, and in the right format first.

 

[WineShark]

Awesome effort. Thanks for doing it!!

 

[whlscowt]

Really interesting stuff thanks for the effort.

 

[SJeasy]

You will also want to limit the year to 2007 or 2008. When determining "consistent" players, players drafted so recently are going to have skewed statistics.

You have to drop the first two years or so of NHL time to get to continuing production levels for forwards, but you can pick them off by eyeball in that early two year span by looking at usage. Henrique would be a case in point.

I found that you could pretty much tell on forwards by 5 years after the draft. Dmen were a bit less certain. You need at least 8 years to get goalies at what amounts to the 5 year level on forwards. You will always get late-bloomers in all categories so none of the early analysis is 100% (eg. St. Louis, Tim Thomas, Doug Murray).

I use Phillips as a case in point for the cut line on puck-moving dmen. He is a little bit of a trigger but he isn't a puckmover. Same for Garrison. Puckmoving has not been an emphasis until recently.

 

[Vaasa]

Here a couple of early attempts at creating a visualization, just using the built-in charting tools of Excel:

Edit, either I don't know how to imbed photo's from Flikr or I can't. Here's a link: http://www.flickr.com/photos/vaasa/

 

[Barrie22]

Here a couple of early attempts at creating a visualization, just using the built-in charting tools of Excel:

Edit, either I don't know how to imbed photo's from Flikr or I can't. Here's a link: http://www.flickr.com/photos/vaasa/

so literally anything from the 4th round is basically a throw of the dice and hope and pray to god that they make it?

 

[Led Zappa]

Here a couple of early attempts at creating a visualization, just using the built-in charting tools of Excel:

Edit, either I don't know how to imbed photo's from Flikr or I can't. Here's a link: http://www.flickr.com/photos/vaasa/

You should try Dropbox or something like it. Can't really view it that well on Flickr. Enlarging it puts it in some weird full screen video mode and I can't figure out a way to save it so I could enlarge it myself.