Scouting, Statistics, and St. Louis

[Iain Fyffe]

NCAA players or NCAA bound players are drafted by NHL teams after being qualified by the scouts as to whether they are willing to leave early and turn pro if the opportunity exists.

You're reaching. Jason Botterill was drafted 20th in 1994, he finished college. Tony Tuzzolino, #113, finished college. Luciano Caravaggio, #155, finished college. Ashlin Halfnight #213, finished college. Tim Thomas, #217, finished college. Jeff Mikesch, #231, finished college.

Those are the six NCAA players drafted in 1994. All were drafted after their first season of NCAA play, all played pro after their NCAA career, and all finished the college years before turning pro. Not a single NCAA player drafted that year left college early.

 

[Iain Fyffe]

For every st luis there is like 50 guys who busted. You can take a chance, but no way can you tell if it is an educated one. players get drafted on IQ, skill, size, strength, etc. Skilled guys always get passed up, usually its for good reasons, some times its a mistake.

For every St. Louis, there are 50 guys you busted is true, in the sense that for every player the calibre of St. Louis, there are many who are not. There are very few small players with St. Louis' results in 94-95. You can't lump him in with all small players. That's not his defining feature. His defining feature is how good he is at hockey.

And then, of course, there's the ratio of big players who make it to those who don't. Historical the big player busts are more costly to teams, because they used high draft picks on them. If an 8th-round pick busts out, you really haven't lost anything. If a 1st-rounder does, you've lost a lot.

I didn't read the whole thread but when it comes to scouting small players, there is a perverse consideration I wanna talk about.

If I'm a scout from a an average NHL team and I wanna make a name for myself. Why would I draft a small player even if I do believe that size isn't that important? I know that:

1) My people will not give him a proper opportunity
2) Even in the event that he is given an opportunity, the coach and the GM will not rate him objectively and he is very likely to discriminated against in the future.
3) if my average-to-good size player busts, the player will take most of the blame. If a small player busts, it'll be charged doubly harshly to my account.

I think there's an element of truth to this. Many teams seem to draft "safe" players, of course we can't tell on whose recommendation that is. They'd apparently rather draft a player who has a good chance to be a fourth-liner in the late rounds, than a player who has big upside potential but has a greater chance to bust. Such picks make no sense in terms of the economics of draft picks, but may very well be done for the reasons you mention above.

 

[Canadiens1958]

Results

You're reaching. Jason Botterill was drafted 20th in 1994, he finished college. Tony Tuzzolino, #113, finished college. Luciano Caravaggio, #155, finished college. Ashlin Halfnight #213, finished college. Tim Thomas, #217, finished college. Jeff Mikesch, #231, finished college.

Those are the six NCAA players drafted in 1994. All were drafted after their first season of NCAA play, all played pro after their NCAA career, and all finished the college years before turning pro. Not a single NCAA player drafted that year left college early.

Because the teams were willing to wait. Not all teams are patient. A GM is not interested in leaving a legacy for his successor at the cost of his job.

You miss the results component of your model. Specifically from your 1994 model / system 5 of the top thirty did not play 1 game in the NHL or 16.67%

Looking at the 1990-2004 NHL entry drafts and the top 30 picks each year we see that the 0 game draftees were the following:

1990-4, 1991-1, 1992-3, 1993-2, 1994-3,1995-1,1996-2, 1997-3, 1998-1, 1999-3, 2000-1, 2001-3, 2002-4, 2003-0,2004-2. Total = 33 or 2.2 per year.

The NHL approach produced much better results.

 

[Iain Fyffe]

Because the teams were willing to wait. Not all teams are patient. A GM is not interested in leaving a legacy for his successor at the cost of his job.

So just the patient teams screwed up their evaluation of St. Louis then?

You miss the results component of your model. Specifically from your 1994 model / system 5 of the top thirty did not play 1 game in the NHL or 16.67%

In the top 30 players actually drafted, three played no NHL games. But playing an NHL game is not a measure of success. "As long as this guy plays at least 1 NHL game, this pick is a success" is not something you should ever be thinking.

The total NHL games played by the first 30 actual picks is 13,372, while my system produces 12,522. So when you say the NHL produced "much" better results, apparently you mean 7%?

But if you look at, say, the top 11 players in my system, only two players failed to play at least 600 NHL games. Of the top 11 actual draftees, four of them failed to play even 220 NHL games. If you make anything between 220 and 245 the arbitrary cutoff, I get 10 of 11, the scouts get 7 of 11.

So what else do you have? These little piecemeal things go nowhere; you need to be systematic and thorough.

 

[Iain Fyffe]

Further on the patient teams, between 1992 and 1996, of the players drafted from the NCAA about 2/3 finished college. Also in this time, 19 of 26 teams made at least one "patient" pick. It seems patient teams were not in short supply.

 

[Canadiens1958]

NHL Teams

So just the patient teams screwed up their evaluation of St. Louis then?


In the top 30 players actually drafted, three played no NHL games. But playing an NHL game is not a measure of success. "As long as this guy plays at least 1 NHL game, this pick is a success" is not something you should ever be thinking.

The total NHL games played by the first 30 actual picks is 13,372, while my system produces 12,522. So when you say the NHL produced "much" better results, apparently you mean 7%?

But if you look at, say, the top 11 players in my system, only two players failed to play at least 600 NHL games. Of the top 11 actual draftees, four of them failed to play even 220 NHL games. If you make anything between 220 and 245 the arbitrary cutoff, I get 10 of 11, the scouts get 7 of 11.

So what else do you have? These little piecemeal things go nowhere; you need to be systematic and thorough.

So the NHL teams collectively had the sense not to draft two of your top 30. This is a tribute to them and a negative for your system. Either way the NHL system was more effective either by 7% or looking at top 30 busts.Arbitrarily setting the number at 11 does not work. No NHL job or academic situation allows self grading or self definition of criteria.

As stated previously, teams are not willing to wait as long as it took the St. Louis to make the NHL. 50 pro contracts mean that time runs out on draft picks quickly. In the case of St. Louis it would have been after 3 seasons. Perrin and St. Louis did not sign with an NHL team once their NCAA careers ended but with an IHL team. Which means even if drafted they more than likely would not have been tendered.

 

[Iain Fyffe]

So the NHL teams collectively had the sense not to draft two of your top 30. This is a tribute to them and a negative for your system. Either way the NHL system was more effective either by 7% or looking at top 30 busts.Arbitrarily setting the number at 11 does not work. No NHL job or academic situation allows self grading or self definition of criteria.

30 is an arbitrary number (that was the whole point of my selecting some other arbitrary number). There's no reason that 30 is a better number than 11, other than it's a bigger number. A bigger number still would be even better presumably. The only non-arbitrary number is to use all of them, which we haven't done. Using your arbitrary number of 30, it seems that the collective wisdom of hundreds of professional scouts produced 7% better results than one guy and a computer.

As stated previously, teams are not willing to wait as long as it took the St. Louis to make the NHL. 50 pro contracts mean that time runs out on draft picks quickly.

Just because you state it, doesn't mean it's true. With an NCAA draft pick you have the luxury of not having to sign him in the normal time period. You're in no danger of losing his rights until he's finished college, so he could sit on your reserve list, not taking up a pro contract, and you can watch him develop further.

And as I pointed out a couple of posts ago, the great majority of NHL teams were willing to draft a player they had to wait to finish college, and most NCAA players drafted at this time did finish college before turning pro. So your claims that teams in general were not willing to wait is clearly and flatly false.

In the case of St. Louis it would have been after 3 seasons. Perrin and St. Louis did not sign with an NHL team once their NCAA careers ended but with an IHL team. Which means even if drafted they more than likely would not have been tendered.

Not hardly. If a team had been able to see past their size to draft them, they would have seen past their size to sign them. Their not having signed with an NHL after college just means that they were still facing the biases that prevented them from being drafted. Their not being signed after college also punches holes in your idea that it was a strategic lack of patience that prevented them from being drafted. They could have been had that same day with a contract, and were still not signed.

 

[overpass]

No it's not. As I said, numerical analysis could have been done at the time, using information available at the time, that would indicate St. Louis (and Perrin) were players you really wanted to look at.

I took another hour today and adapted the QMJHL method I developed yesterday, this time using ECAC player data from 1980-1982. All of the data used existed, and was easily available, in 1994 and 1995. Again, the output in this case is peak (best 5 of 10 seasons) points-per-game production at the NHL level, assuming a scoring environment of 3.75 goals per game. Results for contemporaries of St. Louis and Perrin:

Player|Season|Est PPG
St Louis M|1995|1.22
Perrin E|1995|1.21
St Louis M|1994|0.98
Perrin E|1994|0.77
Palmer S|1994|0.55
Martins S|1992|0.54
Fogarty R|1992|0.52
Ducharme D|1993|0.50
Majic X|1993|0.48
Ducharme D|1992|0.48
Richardson B|1993|0.47
Sancimino M|1994|0.21
Majic X|1992|0.19
Perreault N|1992|.018
Murphy B|1993|0.18
Clark W|1992|0.17
Houle J|1994|0.16
Regan T|1994|0.16
Tuomainen M|1992|0.15
Regan T|1993|0.15
Palmer S|1993|0.15

So now we can stop pretending that numerical analysis could not have been done, using only information available at the time, that would show the seasons put up by St. Louis and Perrin to be as impressive as they actually are.

I appreciate the work done here, Iain. I'm convinced that junior league production above a certain threshold is a good predictor of NHL production. But it doesn't entirely address the Martin St. Louis issue.

Among the largest reasons that St. Louis was passed over in the draft was his short stature. I'm sure you would agree with that - you've often written that short players have been overlooked by NHL teams. Martin St. Louis himself is a data point in favour of your argument, as are players like Daniel Briere and Brian Campbell. But as of 1994, what was the track record of players 5'9" or shorter at transitioning to the NHL? Should NHL teams have not considered the height or size of players, or had height and size been important factors in the ability of players to make the NHL?

Anecdotally I can think of Theo Fleury who had a lot of success. But there were also some high profile smaller players who failed to turn junior success into NHL success.

 

[Canadiens1958]

Realities

30 is an arbitrary number (that was the whole point of my selecting some other arbitrary number). There's no reason that 30 is a better number than 11, other than it's a bigger number. A bigger number still would be even better presumably. The only non-arbitrary number is to use all of them, which we haven't done. Using your arbitrary number of 30, it seems that the collective wisdom of hundreds of professional scouts produced 7% better results than one guy and a computer.


Just because you state it, doesn't mean it's true. With an NCAA draft pick you have the luxury of not having to sign him in the normal time period. You're in no danger of losing his rights until he's finished college, so he could sit on your reserve list, not taking up a pro contract, and you can watch him develop further.

And as I pointed out a couple of posts ago, the great majority of NHL teams were willing to draft a player they had to wait to finish college, and most NCAA players drafted at this time did finish college before turning pro. So your claims that teams in general were not willing to wait is clearly and flatly false.


Not hardly. If a team had been able to see past their size to draft them, they would have seen past their size to sign them. Their not having signed with an NHL after college just means that they were still facing the biases that prevented them from being drafted. Their not being signed after college also punches holes in your idea that it was a strategic lack of patience that prevented them from being drafted. They could have been had that same day with a contract, and were still not signed.

You submitted the arbitrary number of 30 in your article. I just went with it. It is the number you selected.

You still have to stock your minor league system with players and develop players to fit with the NHL team. This is not happening when the player is in university or outside your control. Teams wait on NCAA players who are drafted but not progressing as expected. See 2006 David Fischer vs other picks from the same era, 2000s - Max Pacioretty, Louis Leblanc, Mike Komisarek, Chris Higgins who left the NCAA early. Do not mistake disappointment for patience. All you have shown is that teams were disappointed with multiple NCAA picks in the 1990s. With CHL players the disappointment came sooner so the roster or reserve list spots were churned more efficiently.

Your theory about St. Louis and Perrin is based on assumptions that are very questionable. You assume they actually wanted to be NHL drafted as early as possible, putting themselves in the least favourable position for choosing their life path and/or contract negotiations at each step of the process.

Consider that their career path was as a duo from bantam thru their first pro contract with Cleveland in the IHL.

1991-92 played Midget AAA in their hometown of Laval. 1992 QMJHL draft saw a scenario where the duo would be split. The top four Q draft spots were held by weak teams run by poor organizations with no desirable educational opportunities in town.The duo avoided the Q, took a year to get their grade 12 qualifications for an NCAA school while playing hockey within an hour of their Laval home. Then they both chose the University of Vermont, the closest NCAA school to their home, a comfortable 1 1/2 hours from home.. Both were committed to completing their university career with a degree with no early departure.

If hockey was still an option post university then they were in a position to optimize their payday which was not the case for drafted NCAA players who were limited to one team in negotiations.
Your point and your complete position re the NHL draft has the effect of penalizing the player. You claim that he should have been drafted earlier fits the abstract of your system / model but does not fit the players' goal of optimizing life opportunities and revenues.

This is also true for the NHL approach but at least the NHL system recognizes that certain players - Adam Oates, Martin St. Louis, Curtis Joseph, Ed Belfour will choose alternative routes with great success, while others will do so with limited or no success in the NHL.

 

[Iain Fyffe]

Among the largest reasons that St. Louis was passed over in the draft was his short stature. I'm sure you would agree with that - you've often written that short players have been overlooked by NHL teams. Martin St. Louis himself is a data point in favour of your argument, as are players like Daniel Briere and Brian Campbell. But as of 1994, what was the track record of players 5'9" or shorter at transitioning to the NHL? Should NHL teams have not considered the height or size of players, or had height and size been important factors in the ability of players to make the NHL?

There's quite a list of successful small NHL players at the end of this article, including many pre-dating St. Louis.

The Projectinator actually does consider size. Smaller forwards are given lower rankings than larger one with the same numbers. And the adjustment is significantly larger for defencemen. Just as with age, it's something both the system and the scouts take into consideration, but with apparently very different degrees of importance.

 

[Iain Fyffe]

You submitted the arbitrary number of 30 in your article. I just went with it. It is the number you selected.

Yes, and if I had selected 11 we'd be using 11 then. It remains an arbitrary number and therefore not a useful one for evaluating the system.

You still have to stock your minor league system with players and develop players to fit with the NHL team. This is not happening when the player is in university or outside your control.

No kidding. Here's one solution: don't only draft players from the NCAA. I think NHL teams understand that, because no team drafts exclusively from the NCAA.

All you have shown is that teams were disappointed with multiple NCAA picks in the 1990s. With CHL players the disappointment came sooner so the roster or reserve list spots were churned more efficiently.

If you say so. Seems just as reasonable to say that with a low draft pick, a team is willing to wait to see if the player does develop as they hope. I can see your argument flying with a 1st-rounder, but the boggling fact about St. Louis is that no one was willing to even through a 9th-round pick at him and see what happened. A pick that would otherwise be used on some guy who might make the fourth line some day.

Your theory about St. Louis and Perrin is based on assumptions that are very questionable.

I'd say the same about your hypotheses. My theory about St. Louis and Perrin is that they were not drafted because they were seen as being too small for the NHL. The Entry Draft size fetish, which peaked in 1998, was well on its way at this time. My theory is that, especially at the time, NHL teams put way too much emphasis on size at the draft, and unjustifiable emphasis. And I have the research to back that up.

 

[Iain Fyffe]

Project some of the undrafted 2012 players and we'll see roughly 3 - 7 years down the road.

We're getting close. I projected the 2009 draft on Hockey Prospectus back in, well, 2009. So those players are 3 years along already!

 

[Canadiens1958]

If

Yes, and if I had selected 11 we'd be using 11 then. It remains an arbitrary number and therefore not a useful one for evaluating the system.


No kidding. Here's one solution: don't only draft players from the NCAA. I think NHL teams understand that, because no team drafts exclusively from the NCAA.


If you say so. Seems just as reasonable to say that with a low draft pick, a team is willing to wait to see if the player does develop as they hope. I can see your argument flying with a 1st-rounder, but the boggling fact about St. Louis is that no one was willing to even through a 9th-round pick at him and see what happened. A pick that would otherwise be used on some guy who might make the fourth line some day.


I'd say the same about your hypotheses. My theory about St. Louis and Perrin is that they were not drafted because they were seen as being too small for the NHL. The Entry Draft size fetish, which peaked in 1998, was well on its way at this time. My theory is that, especially at the time, NHL teams put way too much emphasis on size at the draft, and unjustifiable emphasis. And I have the research to back that up.

The old qualifier if. 11 would have been rejected as insufficient for a sample space. 30 barely makes the cut since it covers the bare minimum of a first round.

St.Louis and Perrin were committed to the NCAA route. They also realized almost a generation ago what Justin Schultz realized this year and a few others did in between, that being drafted is not optimal in all situations. It was not for them in the Q or the NHL.
Free agency, the earlier the better, is the best option for the player.

No one is questioning the size emphasis in the NHL nor the research that you have in this regard. In various threads I have made the same point.

Your theory is lacking because it does not ask the necessary questions about how do players who are disadvantaged by the NHL draft tendencies go about creating the best situation for themselves. Factor in these questions and the understanding of the NHL draft mechanics will improve.

 

[Iain Fyffe]

30 barely makes the cut since it covers the bare minimum of a first round.

30 doesn't make the cut. To analyze the results you need to be thorough and systematic, and you're being neither.

St.Louis and Perrin were committed to the NCAA route. They also realized almost a generation ago what Justin Schultz realized this year and a few others did in between, that being drafted is not optimal in all situations. It was not for them in the Q or the NHL.
Free agency, the earlier the better, is the best option for the player.

Do you have evidence that St. Louis did not want to be drafted, or is this you guessing?

Comments shown here from St. Louis suggest this was not the case. St. Louis says he was naive about the process when he was young, that if he played well he would get a fair shot.

So even if his plan was to go undrafted and earn a FA contract, as you suggest, he now believes that to be a naive course of action.

Your theory is lacking because it does not ask the necessary questions about how do players who are disadvantaged by the NHL draft tendencies go about creating the best situation for themselves. Factor in these questions and the understanding of the NHL draft mechanics will improve.

The problem is, St. Louis and Perrin are outliers. Small players with their qualifications were usually drafted, just lower than they deserved to be. But there's a difference between being under-drafted and being un-drafted.

 

[wgknestrick]

So just the patient teams screwed up their evaluation of St. Louis then?


In the top 30 players actually drafted, three played no NHL games. But playing an NHL game is not a measure of success. "As long as this guy plays at least 1 NHL game, this pick is a success" is not something you should ever be thinking.

The total NHL games played by the first 30 actual picks is 13,372, while my system produces 12,522. So when you say the NHL produced "much" better results, apparently you mean 7%?

But if you look at, say, the top 11 players in my system, only two players failed to play at least 600 NHL games. Of the top 11 actual draftees, four of them failed to play even 220 NHL games. If you make anything between 220 and 245 the arbitrary cutoff, I get 10 of 11, the scouts get 7 of 11.

So what else do you have? These little piecemeal things go nowhere; you need to be systematic and thorough.

IMO # of games is not a telling metric for how well either system drafted. You have significant quality of those games that needs to be accounted for as well as RANKING those players. Total number of games is irrelevant, but which system ranked those players best compared to actual production rates is the true test. The # of games will always favor the actual (NHL) drafting method due to selection bias. NHL teams are much more likely to play their top draft picks regardless of true talent. Iain, your diamonds in the rough may be very talented, but get little opportunity to ever show it.

I still don't quite understand why production based evaluation isn't the primary factor in draft order. It just seems like a sound metric to base a decision on rather than opinions. Who cares what players look like out on the ice, if they can get the job done (cough, Max Talbot)?

 

[bluesfan94]

The problem is, St. Louis and Perrin are outliers. Small players with their qualifications were usually drafted, just lower than they deserved to be. But there's a difference between being under-drafted and being un-drafted.

I don't think that that is necessarily true. If someone is projected to go in the 6th round, and is underdrafted, that could lead to them being undrafted. One can lead to the next. Basically, there could be teams that thought MSL or Perrin were worthy of being drafted, but when they picked, there were higher ranked players still available. That late in the draft, there is much less of a consensus as to a draft board. One team could have had MSL ranked in the 5th round, for example, but in that round, and the following rounds, someone ranked higher than MSL on that team's draft board was still available. I think that this is the difficulty with saying that all teams missed on MSL. Sure, they may not have drafted him, but if they had him as a 5th round talent, they just misranked someone, which is common that late in the draft.

 

[Canadiens1958]

Comments

30 doesn't make the cut. To analyze the results you need to be thorough and systematic, and you're being neither.


Do you have evidence that St. Louis did not want to be drafted, or is this you guessing?

Comments shown here from St. Louis suggest this was not the case. St. Louis says he was naive about the process when he was young, that if he played well he would get a fair shot.

So even if his plan was to go undrafted and earn a FA contract, as you suggest, he now believes that to be a naive course of action.


The problem is, St. Louis and Perrin are outliers. Small players with their qualifications were usually drafted, just lower than they deserved to be. But there's a difference between being under-drafted and being un-drafted.

I comment on what is presented. You chose not to present more and I have no obligation to do your work or presentations.

Comments attributed to St.Louis are about his NHL camp experiences post IHL and in no way touch on the pre QMJHL draft or NHL draft situations.

The approach taken by St. Louis and Perrin was obvious from the moment they finished their Midget AAA season. The QMJHL and NHL scouts understood and adjusted accordingly. As did the Montreal media. There was no hue and cry about any perceived injustice that they were not drafted by the media people in the loop. Outside or transient media types may have commented but that is the way it is.

 

[Canadiens1958]

Signing

IMO # of games is not a telling metric for how well either system drafted. You have significant quality of those games that needs to be accounted for as well as RANKING those players. Total number of games is irrelevant, but which system ranked those players best compared to actual production rates is the true test. The # of games will always favor the actual (NHL) drafting method due to selection bias. NHL teams are much more likely to play their top draft picks regardless of true talent. Iain, your diamonds in the rough may be very talented, but get little opportunity to ever show it.

I still don't quite understand why production based evaluation isn't the primary factor in draft order. It just seems like a sound metric to base a decision on rather than opinions. Who cares what players look like out on the ice, if they can get the job done (cough, Max Talbot)?

Signing the player is the main factor. Red flags that indicate a player will not sign when the team wishes to sign him are weighed in the process.

 

[overpass]

We're getting close. I projected the 2009 draft on Hockey Prospectus back in, well, 2009. So those players are 3 years along already!

Ryan Ellis should be an interesting test case for your rankings. We'll see how he does.

His development has been behind several players that you projected for similar peaks. Possibly part of a trend where smaller players take longer to develop? See St. Louis, Rafalski, Briere, Campbell, etc.

 

[Ogopogo*]

The last thing I'll say about the "eyes" thing is this: if eyes capture everything, why was Martin St. Louis not drafted in 1994 or 1995? Trained eyes, professional eyes, saw him play and decided he was simply too small to play in the NHL. The numbers at least should have said "you really ought to consider this young fellow, what else are you going to use your 8th-round pick on?". The numbers ain't perfect, but the eyes sure ain't either.

Eyes are biased and forgetful. Scoring stats have a large role to play in evaluation - eyes help too but they are not the be-all, end-all.

 

[Canadiens1958]

1992 Draft

30 doesn't make the cut. To analyze the results you need to be thorough and systematic, and you're being neither.


Do you have evidence that St. Louis did not want to be drafted, or is this you guessing?

Comments shown here from St. Louis suggest this was not the case. St. Louis says he was naive about the process when he was young, that if he played well he would get a fair shot.

So even if his plan was to go undrafted and earn a FA contract, as you suggest, he now believes that to be a naive course of action.


The problem is, St. Louis and Perrin are outliers. Small players with their qualifications were usually drafted, just lower than they deserved to be. But there's a difference between being under-drafted and being un-drafted.

Let's look at other NHL Entry Drafts like the 1992 and your parallel projectination:

http://www.puckprospectus.com/article.php?articleid=1187&mode=print

In the 1992 NHL Entry draft the scouts and GMs combined to pick three players, amongst the top 30 picks, who never played an NHL game. Your top 30 has nine such players. Paul Brousseau, Mikko Luovi,Steve, Suk,Derek Wilkenson, Tomas Klimt, Marc Hussey, Phillippe deRouville,Ryan Smith-D, Keli Corpse.

Also there is the interesting case of Todd Marchant, NCAA player not drafted in 1992 but drafted in the 7th round by the Rangers in 1993, leaving the university program early to sign. Signability
being the key. Scouts doing their job getting an indication of what it would take to sign the player.

More we look at your projectinating efforts the weaker they seem.

 

[Canadiens1958]

Eyes

Eyes are biased and forgetful. Scoring stats have a large role to play in evaluation - eyes help too but they are not the be-all, end-all.

Big misrepresentation. Scouts use their eyes to observe and they use their hands,ears, experience and intellect to record, grade the players on what they saw and heard in conversations with the players, coaches, arena confidants and sources. Their stats extend well beyond simple scoring stats to micro stats looking at facets of the game that few even consider. It is a very complete package.

 

[Iain Fyffe]

IMO # of games is not a telling metric for how well either system drafted.

I agree - it's just a quick-and-dirty way of looking at it.

The # of games will always favor the actual (NHL) drafting method due to selection bias. NHL teams are much more likely to play their top draft picks regardless of true talent.

Yes, this is an important consideration. As I said, the system does consider a player's size, because in the data set the larger the player, in general the better the career. However, whether this is due to actual ability, or due to selection bias by those who distribute ice time to the players, is unknown.

I think that this is the difficulty with saying that all teams missed on MSL. Sure, they may not have drafted him, but if they had him as a 5th round talent, they just misranked someone, which is common that late in the draft.

Absolutely, in general teams missed on him. There may have been some teams that were eyeing him, but didn't get the chance to draft him. I'm not sure how likely it is there were very many of these teams, however, since there were 520 players drafted in his draft years.

Ryan Ellis should be an interesting test case for your rankings. We'll see how he does.

His development has been behind several players that you projected for similar peaks. Possibly part of a trend where smaller players take longer to develop? See St. Louis, Rafalski, Briere, Campbell, etc.

Defencemen in general take longer to develop. He will be a very interesting one to watch, bearing in mind of course that ultimately he was drafted right around where the system suggests he should have been.

Eyes are biased and forgetful. Scoring stats have a large role to play in evaluation - eyes help too but they are not the be-all, end-all.

This is basically my point. Both factors are very important IMO - I think scouting deserves more credit than you're giving it here. But the numerical analysis deserves way more credit than C1958 is giving it here.

Let's look at other NHL Entry Drafts like the 1992 and your parallel projectination:

Let's look at all of them. Why pick one or two when you can look at all of them? In 1991 the system comes out looking mighty fine. But why cherry-pick examples when you can look at all of them?

Also there is the interesting case of Todd Marchant, NCAA player not drafted in 1992 but drafted in the 7th round by the Rangers in 1993, leaving the university program early to sign. Signability
being the key. Scouts doing their job getting an indication of what it would take to sign the player.

What it would take apparently amounting to drafting him in the 7th round. Are you sure the reason he wasn't drafted in 1992 was signability rather than size? He left college early in 1993; are you saying you believe he simply wouldn't have done the same in 1992?

More we look at your projectinating efforts the weaker they seem.

If you say so.

 

[Iain Fyffe]

Their stats extend well beyond simple scoring stats to micro stats looking at facets of the game that few even consider. It is a very complete package.

Wow. So you obviously agree that using stats aids in the scouting process. And yet, my suggestion that such is true is met with complaint after complaint. You're a piece of work, you are.

My entire point is that with a fairly simple system using only very basic stats, you can get pretty good draft rankings. I can only imagine what can be done with a much more thorough system with access to better data.

 

[overpass]

Defencemen in general take longer to develop. He will be a very interesting one to watch, bearing in mind of course that ultimately he was drafted right around where the system suggests he should have been.

Looking at your article from 2009, it looks like the system would put him higher. Correct me if I'm reading this wrong, but his "Projectinator estimate in the last column (10YE)" is 0.94, or a close third behind John Tavares and Brayden Schenn, ahead of Duchene, Kane, Kadri, Glennie, and Cowen.