Probability of a First Round Pick = Top 6/4

[Grind]

First round only

Suprising. Was definitly expecting more goaly and dmen busts, though i geuss thats accounted for in the more recetn drafts. very cool none the less. I would definitely heavily favor forwards in the first going forward

 

[DaveG]

so breaking it down farther:

top 5:
8 of 44 forwards didn't pan out (.181)
5 of 21 dmen didn't pan out (.238)
all 5 goalies panned out... shockingly. People can argue semantics with DP, but he was a starter for the Isles for a while and a pretty solid one at that.
overall 13 of 70 picks in that span were busts (.186)

6 through 10:
here it starts to get dicey
20 of 46 forwards didn't pan out (.465)
9 of 18 dmen didn't pan out (.500)
all 6 goalies did not pan out
overall, selecting in this zone over the 1994-07 time frame meant batting .500

11 through 15:
24 of 46 forwards selected didn't pan out (.522)
7 of 18 dmen selected didn't pan out (.389)
3 of 6 goalies selected didn't pan out (.500)
34 of 70 picks in this range over that time didn't pan out (.486)
#15 overall pick was particularly a death sentence spot, as only 2 weren't complete misses and one of those two was Radulov.

16 through 20:
34 of 46 forwards didn't pan out (.739)
8 of 21 dmen didn't pan out (.381)
2 of 3 goalies didn't pan out (.667)
44 of 70 players selected in this range did not pan out (.629) basically leaving teams slightly better then 1 in 3 chance of selecting a solid contributor to their team here.

21 through 25:
17 of 37 forwards didn't pan out (.459)
14 of 23 dmen didn't pan out (.609)
5 of 10 goalies didn't pan out (.500)
36 of 70 picks in this draft range didn't pan out (.514), though the question is why it's so much better then the 16-20 zone. Great place to pick a goalie if you're taking one in the first round.

26 through 30:
20 of 30 forwards didn't pan out (.667)
22 of 32 dmen didn't pan out (.688)
5 of 8 goalies didn't pan out (.625)
overall 47 of 70 (.671) selected in this range did not pan out.
#30 is the only draft position worse then #15, with all but 1 being a total bust, and that one (Daron Quint) wasn't a star by any stretch.

 

[Grind]

Awesome, now i don't have to do that

Hope you don't mind me reposting this to the initial post. Very cool to see.

What was your criteria (sorry if i missed that) for determining success/failure?

 

[Fourier]

Fun Thread.


The first round is quite sensitive to where the pick was made. This seems to rapidly disappear once you get to round 2 and beyond.

 

[Grind]

Fun Thread.


The first round is quite sensitive to where the pick was made. This seems to rapidly disappear once you get to round 2 and beyond.

mmhmm... i always thought this was probably the case which was why i found so many past draft studies very frustrating as they'd go to macro, usually only breaking down by round. This is fine for the 2-7th rounds i think, (i'd like more info on the second round, but i definitely think you get to a pretty general wash by the 3rd/4th round) but I was fairly certain their'd be quite a but I really wanted to see the returns pick by pick within the first.
and thanks a lot to DG for doing the D man - Forward -Goalies numbers, very cool

 

[DaveG]

Awesome, now i don't have to do that

Hope you don't mind me reposting this to the initial post. Very cool to see.

What was your criteria (sorry if i missed that) for determining success/failure?

Basically I took a look at a mix of games played and offensive production (mostly for forwards). If a player has say >500 games played but not a lot of offensive output I'll take a look at their TOI and their reputation. So a guy like Wayne Primeau who wasn't much of a top 6er will still get credit as he was a solid #3 centerman for quite a few years and managed to get a lot of GP. Whereas a guy like Wade Belak never really managed to have the same level of impact despite the number of games played. With dmen it was mostly games played then TOI unless it was an obvious case. There were a few instances (Babchuk being one) that the player would have a significant number of games played, but never really had high ice time or if they did it was only for one or two seasons and was not sustained by them beyond that.

Suprising. Was definitly expecting more goaly and dmen busts, though i geuss thats accounted for in the more recetn drafts. very cool none the less. I would definitely heavily favor forwards in the first going forward

Honestly a good bit of that may have a good bit to do with the adage that dmen take longer to develop. IMO with drafting a forward you should expect to see them become whatever kind of impact player they will turn out to be by age 23, late bloomers like Penner not withstanding. On the other hand with dmen you can often see someone not manage to crack a roster season after season but eventually emerge to be a pretty solid player even heading into their late 20s like Colaiacovo and Harrison have managed to do. From say the 07 draft alone it wouldn't in the least bit surprise me to see Plante and Petrecki turn out to be solid pros, even if they're listed as busts at the moment due to having had minimal NHL impact to date.

 

[Yossarian54]

mmhmm... i always thought this was probably the case which was why i found so many past draft studies very frustrating as they'd go to macro, usually only breaking down by round. This is fine for the 2-7th rounds i think, (i'd like more info on the second round, but i definitely think you get to a pretty general wash by the 3rd/4th round) but I was fairly certain their'd be quite a but I really wanted to see the returns pick by pick within the first.
and thanks a lot to DG for doing the D man - Forward -Goalies numbers, very cool

It could potentially be biased by teams - if they rank choices by the 'round' they estimate them to go in as opposed to a straight forward list. You may see some small differential between say pick 62 and pick 82 in round 3. But as you say, pretty much all the analyses are round by round, making it impossible to tell.

I also think that the abberations seen in pick 16, and the odd variation in blocks 5-10, 11-15, 16-20 and 21-25 are most likely due to the small sample size. Which is really unavoidable here.

Otherwise really good. Ideally a spilt could be done - chance of being elite, chance of being good, chance of being mediocre, chance of busting.

 

[Grind]

It could potentially be biased by teams - if they rank choices by the 'round' they estimate them to go in as opposed to a straight forward list. You may see some small differential between say pick 62 and pick 82 in round 3. But as you say, pretty much all the analyses are round by round, making it impossible to tell.

I also think that the abberations seen in pick 16, and the odd variation in blocks 5-10, 11-15, 16-20 and 21-25 are most likely due to the small sample size. Which is really unavoidable here.

Otherwise really good. Ideally a spilt could be done - chance of being elite, chance of being good, chance of being mediocre, chance of busting.

Totally... again with something that happens so rarely it's impossible to ever get a good handle on it (as stated, by the time you've reached a reasonable sample size, your spanning 15-20 years and its a whole new ball game).


I'd like to try and qualify some theories i have about the returns not diminishing in the expected curve.

2 of which being
1) a lot of "late risers" push into the top 10, where players with a better/longer track record are dropped for players with a "newer" track record
2) Teams stick with "safer" players from 5-10, and are more willing to risk it on boom/bust picks from 10-15 or 20.

Now those are pure speculation, but a breakdown of catagories (Elite[stamkos/crosby], Star[firstline], good[top6], career[3rd], bust[4th and depth players]) could be used to explore theory 2 (IE: if we found that 10-15 creates more "good" to "elite" players, but also has more complete busts, and that almost all of picks 5-10 were "good" and "career" players, it could validate this)

 

[seventieslord]

In other words, I'm hungry for an analysis that offers the following percentages for each draft position:

• chance of being elite
• chance of being very good
• chance of being good
• chance of being a 3rd or 4th liner
• chance of not spending significant time in the NHL

Those percentages then could be fed into a weighted formula (because the value of those results decreases as you move down the list) that determines the 10-year success rate of the draft position.

I've done this exact study. I haven't put it together into a comprehensive form for easy digestion but I have used pieces of it to help settle debates around the board from time to time. Try searching my name and the word "draft" for a primer. Eventually I'll get to releasing the whole thing.

I also think it's much too early to "judge" the 2006-2008 drafts, especially for defensemen and goalies. If you stick just to drafts where you're 100% sure how the players have panned out, you'll have more accurate results.

But the results of this study show the exact same thing I've always said - the higher up you're drafting, the more likely you are to get any caliber of player, regardless of what you set the threshold at - franchise player, star, above average player, average player, regular NHLer.

 

[DaveG]

I'd also urge people to be careful when evaluating old drafts. There reaches a point where things start to normalize but before say 1992 there was still a pretty big stigma about drafting Soviet and Czechoslovakian players. It led to a lot of unpredictability about whether or not said players would even be able to make it over, which led to a lot of elite level players even then slipping. 90 for example had Jagr, who was the consensus #1 or #2 player in the draft IIRC, ending up at #5, Kozlov in the 3rd, Zhamnov in the 4th, and Zubov in the 5th. Even 5 years later there's no way that happens.

91 had Kovalev at #15 and Ozolinsh at #30 for similar reasons.

It's the biggest reason I typically use 1994 as my base date when I do these compilations. Though an argument can certainly be used for 1992 as well with Hamrlik, Yashin, Kasparaitis, Petrovicky, and Nazarov all going top 10, with Gonchar and Straka also being first rounders.

And any Soviet Bloc pick from before 1989 can't be taken seriously when using it to make a list like this. The guy that was arguably the best goalie of all time was taken in the 10th round in 1983 for that reason.

 

[Grind]

I'd also urge people to be careful when evaluating old drafts. There reaches a point where things start to normalize but before say 1992 there was still a pretty big stigma about drafting Soviet and Czechoslovakian players. It led to a lot of unpredictability about whether or not said players would even be able to make it over, which led to a lot of elite level players even then slipping. 90 for example had Jagr, who was the consensus #1 or #2 player in the draft IIRC, ending up at #5, Kozlov in the 3rd, Zhamnov in the 4th, and Zubov in the 5th. Even 5 years later there's no way that happens.

91 had Kovalev at #15 and Ozolinsh at #30 for similar reasons.

It's the biggest reason I typically use 1994 as my base date when I do these compilations. Though an argument can certainly be used for 1992 as well with Hamrlik, Yashin, Kasparaitis, Petrovicky, and Nazarov all going top 10, with Gonchar and Straka also being first rounders.

And any Soviet Bloc pick from before 1989 can't be taken seriously when using it to make a list like this. The guy that was arguably the best goalie of all time was taken in the 10th round in 1983 for that reason.

this is definitely the biggest roadblock to creating a truly comprehensive study to drafting/likely-hood/etc. With such small sample sizes done so infrequently, any attempt at gathering a large amount of data gets undermined by changing outside factors over the necessary time frame. Changes in bias's have had a pretty major role in what types of players get drafted how, not to mention what types of players are able to be successful.

really the most you can get for a study of draft picks in a similar set of bias's and that have fully developed by the current date, is going to be about 6-8 years max- not nearly enough to establish anything solid.

 

[seventieslord]

I'd also urge people to be careful when evaluating old drafts. There reaches a point where things start to normalize but before say 1992 there was still a pretty big stigma about drafting Soviet and Czechoslovakian players. It led to a lot of unpredictability about whether or not said players would even be able to make it over, which led to a lot of elite level players even then slipping. 90 for example had Jagr, who was the consensus #1 or #2 player in the draft IIRC, ending up at #5, Kozlov in the 3rd, Zhamnov in the 4th, and Zubov in the 5th. Even 5 years later there's no way that happens.

91 had Kovalev at #15 and Ozolinsh at #30 for similar reasons.

It's the biggest reason I typically use 1994 as my base date when I do these compilations. Though an argument can certainly be used for 1992 as well with Hamrlik, Yashin, Kasparaitis, Petrovicky, and Nazarov all going top 10, with Gonchar and Straka also being first rounders.

And any Soviet Bloc pick from before 1989 can't be taken seriously when using it to make a list like this. The guy that was arguably the best goalie of all time was taken in the 10th round in 1983 for that reason.

You’re right, but my project doesn’t exclude players like that, nor do I plan on it.

At any given time, teams draft players based on their upside and their likelihood of ever reaching that upside at the NHL level, as well as how widely known the previous two factors are. Some teams thought highly of Hasek’s potential but apparently not very highly of his likelihood of playing. In the annals of history, the guy is just a late round steal, like Alfredsson or Kaberle.

There are players just like that today, the only difference being is that there are no political roadblocks anymore, just skill-based ones. But in principle players are drafted in the same way.

 

[BenchBrawl]

It would also be interesting to evaluate the success rate of individual head scouts.While a certain pick might have a certain % chance to become a top 6/top 4 when you take the average of all head scouts , it doesn't mean the % is true if you're smarter at picking up talents and therefore we should try to evaluate the degree to which talent in scouting could dominate these average % (if we want to jump to conclusions as far as the value of certain picks).

 

[Bomber0104]

Interesting summaries and findings Grind!

I just pity the poor forward that scored 40-44 points every year with lots of hard work shifts along the boards

 

[seventieslord]

Interesting summaries and findings Grind!

I just pity the poor forward that scored 40-44 points every year with lots of hard work shifts along the boards

good reason to not have a "hit or miss" metric and instead assign a wide range of values to the players. That kind of guy you mentioned would be a 6.5 or 7.

 

[Gregsky99]

This thread could almost be moved to the prospects board. Some of those guys definitely need to see it.

 

[Grind]

good reason to not have a "hit or miss" metric and instead assign a wide range of values to the players. That kind of guy you mentioned would be a 6.5 or 7.

agreed, a tiered analysis would be better, but again thatis not what the purpose nor scope of this study was.

I had to draw a line somewhere so i did, and i think it's fairly accurate, at least as far as percieved worth.

I'd love to see yoru finished stud Seven, have a pet project that the stats could be helpful for.

 

[seventieslord]

agreed, a tiered analysis would be better, but again thatis not what the purpose nor scope of this study was.

I had to draw a line somewhere so i did, and i think it's fairly accurate, at least as far as percieved worth.

I'd love to see yoru finished stud Seven, have a pet project that the stats could be helpful for.

I am a horrible procrasitnator.

Knowing that there is some demand for it helps.

 

[mopigeons]

Draft picks

So, I'm new to these here parts.
I've long been curious about the established value of a draft pick. IE: what can someone expect out of a draft pick of any given round? How long does it usually take the player to get to the NHL, what is their usual success afterwards, etc.
I'd been thinking of starting this project for a while, but of course, before doing so: if any of you know if this has already been done, can you point me to the study in question?

Otherwise, I might be looking for a few pointers before getting started.
I had a look through some of the sticky threads, and I'm left with the question: what kind of data sources do you use when working with hockey stats? Is it up to us to record them on each go, or are there some complete data sources available? (Databases, Excel files, etc... just something other than bringing up every player's page on HockeyDB or something..)?
[EDIT: Just saw I can get some of the required info from hockey-reference.com in CSV format]

 

[Yossarian54]

Bit o' discussion in this thread:

http://hfboards.mandatory.com/showthread.php?t=1298875

Maybe have a look at this and the references contained within: http://myslu.stlawu.edu/~msch/sports/Schuckers_NHL_Draft.pdf

I'm not sure anyone's really nailed an effective 'measure' of success or failure. For example, Scott Cullen's weighting here http://www.tsn.ca/blogs/scott_cullen/?id=398986 I think slightly inflates the lower pick values by assigning a reasonably high value to good or okay minor leaguers - which are essentially a failure for an NHL team in the 1st round..

 

[mopigeons]

Thanks for the links!

Scott Cullen's analysis I was familiar with, definitely one of the things that piqued my interest in the subject. I'm looking to go a little bit deeper though, and probe into the later rounds, and also perhaps try to establish something more concrete in terms of a draft pick/player's contribution to a team.

For that, the second link has proved useful (http://myslu.stlawu.edu/~msch/sports..._NHL_Draft.pdf). While similar in scope to what I was looking at, the study looks at years 1988-97; whereas I'd like to broaden that horizon to something closer to 1980-today.

One of the challenges I'm facing is in assessing whether to evaluate draft picks by selection rank, round, or some adjusted factor.

I'm leaning towards an adjusted rank, given that the number of teams and rounds in the draft changed on a rather regular basis particularly through the 80s and 90s until the last CBA was signed in 2005.

I'm also facing some challenges with what metrics to use to properly evaluate players. Given that I'd like to go back to the 80s if possible (to get a broader sense of a "modern era" of hockey), I can't seem to find much data other than the commonly available offensive statistics (goals, assists) for skaters, which aren't my favourite, but might have to do if I want to include those years. I believe ice time only starts being tracked in the late 90s, which hinders my ability to use that data unless I also constrict the time frame.

If I do end up using the full timeframe I'd originally wanted (1980 onwards), then so far, I was partial to using this method to evaluate offensive statistics for skaters: http://hfboards.mandatory.com/showthread.php?t=589911

If anyone has any insight, or if I've overlooked any info on boards, please feel free to share!

 

[Grind]

Thanks for the links!

Scott Cullen's analysis I was familiar with, definitely one of the things that piqued my interest in the subject. I'm looking to go a little bit deeper though, and probe into the later rounds, and also perhaps try to establish something more concrete in terms of a draft pick/player's contribution to a team.

For that, the second link has proved useful (http://myslu.stlawu.edu/~msch/sports..._NHL_Draft.pdf). While similar in scope to what I was looking at, the study looks at years 1988-97; whereas I'd like to broaden that horizon to something closer to 1980-today.

One of the challenges I'm facing is in assessing whether to evaluate draft picks by selection rank, round, or some adjusted factor.

I'm leaning towards an adjusted rank, given that the number of teams and rounds in the draft changed on a rather regular basis particularly through the 80s and 90s until the last CBA was signed in 2005.

I'm also facing some challenges with what metrics to use to properly evaluate players. Given that I'd like to go back to the 80s if possible (to get a broader sense of a "modern era" of hockey), I can't seem to find much data other than the commonly available offensive statistics (goals, assists) for skaters, which aren't my favourite, but might have to do if I want to include those years. I believe ice time only starts being tracked in the late 90s, which hinders my ability to use that data unless I also constrict the time frame.

If I do end up using the full timeframe I'd originally wanted (1980 onwards), then so far, I was partial to using this method to evaluate offensive statistics for skaters: http://hfboards.mandatory.com/showthread.php?t=589911

If anyone has any insight, or if I've overlooked any info on boards, please feel free to share!

Very interesting.. I'd love to see what you come up with.

The only thign i'd caution against using a time frame that goes all the way back to the 80's is i think you'll find success rate considerablly different, so using the two in a combined pool will skew "relevent"/recent numbers to seem less succesful.

It is the big issue with an event that happens once a year, the changes in the nature of the game, drafting, scouting, development all play apart so that the data your studieing from the 80's is from a very different creature then what your studying in the last 10 years.

just important to keep this in mind.

 

[mopigeons]

Very interesting.. I'd love to see what you come up with.

The only thign i'd caution against using a time frame that goes all the way back to the 80's is i think you'll find success rate considerablly different, so using the two in a combined pool will skew "relevent"/recent numbers to seem less succesful.

It is the big issue with an event that happens once a year, the changes in the nature of the game, drafting, scouting, development all play apart so that the data your studieing from the 80's is from a very different creature then what your studying in the last 10 years.

just important to keep this in mind.

I know what you mean. While it's my goal to go that far back, I may start off with something "closer" and keep it to the 30-team era. Already, I can acknowledge that even in the 30-team era there are important differences in terms of statistics produced by different styles of play (ie: the crackdown on obstruction post-lockout), but I hope I can find some way to normalize statistics in there.

Once I get something that works for the 30-team NHL, I'll try to expand my timeframe further back.

I'm still working on putting the data together, but look forward to discussing it once results trickle in.

 

[The Sweetness]

Interesting analysis, thanks! However, I do not think it is fair to conclude that first round picks are overvalued based in the evidence you presented. While a large number may bust it is also one of the few ways to get a truly elite player. If the chances if getting an elite ppg forward are low in the first, and lower outside the first, how do you think franchises should rebuild without first round picks, which is where the majority of the talent comes from?

You need something to give more weight to the homerun picks if you hope to conclude something about the value of first rounders imo.

Thanks again!

 

[Grind]

Interesting analysis, thanks! However, I do not think it is fair to conclude that first round picks are overvalued based in the evidence you presented. While a large number may bust it is also one of the few ways to get a truly elite player. If the chances if getting an elite ppg forward are low in the first, and lower outside the first, how do you think franchises should rebuild without first round picks, which is where the majority of the talent comes from?

You need something to give more weight to the homerun picks if you hope to conclude something about the value of first rounders imo.

Thanks again!

To be fair a late first round pick isn't really over rated. If a contender making a push trades a first between say 15-30 for a low end second liner or high end third that's pretty decent return. That first round pick only has a 30% chance of being anything better then that. Furthermore, they've got about 30% chance of being equal ( a third liner) and about a 30% chance of busting completely. The liklihood of getting a "home run" or top line player in the back half of the first is not usually significantly higher then a second round pick, its pretty much a complete lotto roll at that point.



To be honest, the most overrated picks would be those in the 6-10 range. Fans especially seem to be attache dot the diea that a pick in this range requires legitimate top 6 player to come back. There's a 60% chance that player doesn't reach top 6 ability. of course, theres about a 20% chance he exceeds that talent, but you'll find the amount of "home runs" outside the top 5 is nearly identical from picks 6-30, at that point, most GM's will admit it's just luck.

My point beign theirs a signifact drop off after the top 5, and that in reality, picks 6-10, which often seem to be as coveted as the top 5 are more often then not no better then picks 10-15. The liklihood ofa ctually getting "homeruns" or first liners outisde of the top 5 is pretty well consistent from 6 all the way to 26