ATD 2011 Draft Thread III

[Velociraptor]

Dawson City is pleased to select, from the Buffalo Sabres, Gilbert Perreault.

That's a great pick, he wasn't on my radar because I didn't think he'd be available. If I had him and Esposito for a 1-2 punch down the middle. But I think I will focus on other positions seeing as Esposito is a legitimate first line center. Would've been nice to get Perreault this late in the draft.

 

[Hawkey Town 18]

In theory, in the exact same seasons, with all things being equal, yes, the guy who was 1st once and 9th another time would have done better on the aggregate (i.e. maybe he finishes 25 points ahead of 5th when he's 1st, but just 12 points behind 5th when he's 9th) - but I also don't think that was the point he was trying to make at all... at least I hope not. What an awful thing to argue about !



I don't think it would. You need to compare everything to a known value and not change the goalposts year by year. I could select random placements from different generations too, and show that they are all exactly 70% away from eachother but I have better things to do - like pick my nose.

Probably the best thing you can do is put it into practice. Next time you need to compare your player to someone else's, try out what you're saying. I think it would be a mess. When you're done that, try using a comparison to #2 as a benchmark. You'll like it a lot better, I reckon.

I was thinking that instead of my benchmark being a specific no. rank (like #2), that it would be a specific percentile rank (like 80th percentile of however many first liners are in the league). And then you would do multiple comparisons of that type.

 

[seventieslord]

Just to clarify this to anyone reading, here's where I'm coming from.

Anyone who thinks that a 10th place finish in 1930, or being the 10th-best defenseman in 1930, or whatever, is the same as doing that in 2010, is telling you that they think the talent pool and the skills of top level players have looked like this over time:

Perfectly constant.

We know that is incorrect. It has to be. Players are bigger, faster, stronger, and the population that feeds the NHL is so much bigger now, so that just the law of averages alone would make the top-end players that make the NHL, better.

But it is equally, if not more, incorrect, to begin to equate a 6th in a 6-team league as a 30th in a 30-team league. If the talent pool really did get 5X bigger/stronger, sure, that works, but it hasn't. If you believe that line of thinking, then what you are saying is that the talent pool has followed the size of the NHL (or before the NHL, the number of teams from assorted leagues competing for the cup):

Look at the red lines that follow the 2s, 5s, 10s, 15s, and 20s around. Ugly! How can that be? And if you follow that line of thinking, then finishing 10th in the 10-team NHL of 1930 is actually better than finishing 8th in the 6-team NHL of 1960. That simply can't be true. It's not about league size.

So if it's not perfectly constant and it's not about league size, then what do we do? I submit the following:

I contend that the talent pool feeding the highest levels of hockey has been growing steadily over the past century, just as sure as the population and interest in the sport have grown.

Keep in mind that I don't expect this to be seen as exact at all, just a representation of what's been happening. For example, the numbers in 1960 line up to numbers twice as high in 2010. Has the talent pool feeding the NHL really doubled? I don't know. It might be up only 30%, it might be up 300%. That's work for someone else to do. But regardless, in principle it still stands.

Also, keep in mind that this is "ideal", but events such as the world wars, the pre-merger world, senior stars staying amateur in the 20s, the WHA, and elite europeans being slow to come over in the past 50 years, mean that the principle of this graph holds true only a global perspective and not the NHL in particular. For example, if the placements of these numbers were, after furious calculations, deemed to be correct, and someone was talking about a guy who was 15th in scoring in 1975, it has to be noted that that is just 15th in the NHL and not globally, and there were perhaps 5 Europeans and 5 WHA players who were likely better too, making him 25th globally.

As far as the #1s go, yes, the way I aligned them there, they follow a curve just like the other numbers below them. But in reality, the best player in the world can be significantly better than any other (like Orr, Gretzky) or not much better than the next (Hull in the 60s, Crosby today). Pretend that the #1s have a bottom where they stand, and a top of the ceiling of the chart. #1s are not constant. But I do think #2s are a constant; at least compared to #1s. That's why I like comparisons to #2 so much. That is still a curve, but it's the least-sloped line on the graph. For the purposes of analysis, if we assume #2 throughout history is constant, it opens up a world of possibilities for comparisons. But even those are still just starting points. Always consider who was #2. there's a big difference between Frank Boucher and Mario Lemieux.

 

[BenchBrawl]

As far as the #1s go, yes, the way I aligned them there, they follow a curve just like the other numbers below them. But in reality, the best player in the world can be significantly better than any other (like Orr, Gretzky) or not much better than the next (Hull in the 60s, Crosby today). Pretend that the #1s have a bottom where they stand, and a top of the ceiling of the chart. #1s are not constant. But I do think #2s are a constant; at least compared to #1s. That's why I like comparisons to #2 so much. That is still a curve, but it's the least-sloped line on the graph. For the purposes of analysis, if we assume #2 throughout history is constant, it opens up a world of possibilities for comparisons. But even those are still just starting points. Always consider who was #2. there's a big difference between Frank Boucher and Mario Lemieux.

Mario Lemieux was a number 1 for a decade or so for sure.But even when he was number 2 , he was so much of an exception that we can almost make a special rule for the Gretzky-Lemieux era.Even if I'm impressed with the chart for originality and vulgarisation , it's not an exact science.I think it's dangerous to look at it this way if it's being thrown around in debate too much without looking who those numbers X are/were.

1,2,3,4,5 in a specific year could be as good as 1,2,3,4,5 of today , but then the 6th is just as good as the 25th of today. ( just an example )

100 years is a small sample , the variance is going to be high regardless.

 

[seventieslord]

Mario Lemieux was a number 1 for a decade or so for sure.But even when he was number 2 , he was so much of an exception that we can almost make a special rule for the Gretzky-Lemieux era.Even if I'm impressed with the chart for originality and vulgarisation , it's not an exact science.I think it's dangerous to look at it this way if it's being thrown around in debate too much without looking who those numbers X are/were..

- Yes, if there was ever a time when there was #2 was an outlier, it was 1988-1992, when Mario and Wayne were the two best players on earth by far. (in 1986, 1987, 1993, and 1994, injuries aside, it is arguable that they were the two best but if they were it wasn't by so much that it made #2 an outlier)

- I know it's not an exact science. But I'd love it if we could all agree with it in principle. (And I'm referring in particular to the gradual but sure increase in quality of the talent pool, not the #2 comparisons.) Page 3 makes a lot more sense than page 1 or page 2. I don't think anyone is really on page 1 anymore; Dreakmur may have been the last. Page 2 is going extinct as well - If I understand Johnny D correctly, he believes in it, and there may be one or two others but I don't want to misstate their positions.

 

[seventieslord]

1,2,3,4,5 in a specific year could be as good as 1,2,3,4,5 of today , but then the 6th is just as good as the 25th of today. ( just an example )

I highly doubt that.

 

[Derick*]

When you're looking for a sample to adjust to, the higher you go the more variability you're going to have for which period happens to have more talent, and the lower you go the more things like powerplay time and number of teams is going to be a factor. That's why I proposed using 3rd - 18th as a metric. Two for there never being more than two transcendental talents in the league at any one point, and 18 based on the number of first line forwards during the original six era. I created a "scoring rate" by averaging the number of points earned by players in that sample. Here were my results:

Year Scoring Rate
1968 70
1969 82.0714
1970 72.7214
1971 82.2143
1972 86.2857
1973 92
1974 83.8571
1975 99.4286
1976 100.5
1977 81.7453
1978 104.75
1979 97.7143
1980 99.0714
1981 105.7143
1982 111.8571
1983 103.5
1984 106.9285
1985 108
1986 110.7857
1987 95.8571
1988 106
1989 107.0714
1990 106.0714
1991 101.2857
1992 101
1993 121.57
1994 101.0714
1996 107.1428
1997 89.9285
1998 81.6428
1999 87.5
2000 80.5714
2001 89.3571
2002 83
2003 88.0714
2004 78.7857
2006 95.9286
2007 96.2142
2008 88.5714
2009 88.3571
2010 89.5714

It's the same principle 70s is using by adjusting to the second highest scorer but I think in a much more accurate way (I haven't gotten around to doing earlier eras yet).

And then if you want to adjust for size of the talent pool and such factors, you can add another adjustment on top of that.

But, ignoring talent pool, lets say we want to estimate how many points Gretzky would score if his 85-86 season were played in the 09-10 season. 89.5714/110.7857 = 0.8085. (0.8085)(215) = 174 points.

The margin Gretzky led his peers by in the 80s is about the same margin someone who scored 174 points today would lead his peers by.

Intuitively I think most people would guess he'd score less than that, and that goes to show that by only remembering Gretzky's totals people are acting as though all the leading scorers of the 80s were scoring 170 - 200 points, when in reality scoring has only gone down about 20% and the highest total ever not be Lemieux or Gretzky was only 155 points.

 

[monster_bertuzzi]

Dawson City is pleased to select, from the Buffalo Sabres, Gilbert Perreault.

Thats who I wanted to trade up for!!

 

[BenchBrawl]

I highly doubt that.

My example was an exageration , but having said that , even if you highly doubt it doesn't mean it can't happen.In fact , if the NHL survive for 5000 years and we could see it , I would be willing to bet it's certain to happen or close to it.

 

[seventieslord]

When you're looking for a sample to adjust to, the higher you go the more variability you're going to have for which period happens to have more talent, and the lower you go the more things like powerplay time and number of teams is going to be a factor. That's why I proposed using 3rd - 18th as a metric. Two for there never being more than two transcendental talents in the league at any one point, and 18 based on the number of first line forwards during the original six era. I created a "scoring rate" by averaging the number of points earned by players in that sample. Here were my results:

Year Scoring Rate
1968 70
1969 82.0714
1970 72.7214
1971 82.2143
1972 86.2857
1973 92
1974 83.8571
1975 99.4286
1976 100.5
1977 81.7453
1978 104.75
1979 97.7143
1980 99.0714
1981 105.7143
1982 111.8571
1983 103.5
1984 106.9285
1985 108
1986 110.7857
1987 95.8571
1988 106
1989 107.0714
1990 106.0714
1991 101.2857
1992 101
1993 121.57
1994 101.0714
1996 107.1428
1997 89.9285
1998 81.6428
1999 87.5
2000 80.5714
2001 89.3571
2002 83
2003 88.0714
2004 78.7857
2006 95.9286
2007 96.2142
2008 88.5714
2009 88.3571
2010 89.5714

It's the same principle 70s is using by adjusting to the second highest scorer but I think in a much more accurate way (I haven't gotten around to doing earlier eras yet).

And then if you want to adjust for size of the talent pool and such factors, you can add another adjustment on top of that.

If it's based on first line jobs in the O6 how come it starts in 1968?

Also, prior to 1960 it's going to give you some awfully funky results. There were big dropoffs in the results of the 10th/15th/20th scorers compared to 2nd and 5th in the 40s and 50s.

 

[seventieslord]

My example was an exageration , but having said that , even if you highly doubt it doesn't mean it can't happen.In fact , if the NHL survive for 5000 years and we could see it , I would be willing to bet it's certain to happen or close to it.

isn't that the same principle that states that if you give a billion monkeys a billion typewriters, eventually one of them will type out the entire works of shakespeare?

 

[Derick*]

If it's based on first line jobs in the O6 how come it starts in 1968?

Also, prior to 1960 it's going to give you some awfully funky results. There were big dropoffs in the results of the 10th/15th/20th scorers compared to 2nd and 5th in the 40s and 50s.

Because I designed it so that, in principle, I could apply it further back. 1968 was just when I decided to stop the night I started calculating, and I haven't bothered to go further back yet. It's a lot of busy work to add that up you know, unless I find a way to download the data in a spreadsheet.

We'll see. For the seasons I've shown it passes the giggle test, and actually matches people's eyes a lot better than traditional adjusted points (which are entirely based on GPG, which is ridiculous really, for a couple reasons). For instance, Yzerman's 1989 season still ranks higher than a lot of Art Ross seasons, I think justifiably so.

 

[Hawkey Town 18]

Just to clarify this to anyone reading, here's where I'm coming from.

But it is equally, if not more, incorrect, to begin to equate a 6th in a 6-team league as a 30th in a 30-team league. If the talent pool really did get 5X bigger/stronger, sure, that works, but it hasn't. If you believe that line of thinking, then what you are saying is that the talent pool has followed the size of the NHL (or before the NHL, the number of teams from assorted leagues competing for the cup):

This is what I realized earlier, then my thought was to try to account for different talent pool sizes by doing multiple comparisons to create a picture of the what the talent pool looked like at the time...If the gap between bottom end and top end 1st liners was bigger in one year compared to another, then the talent pool of that year was smaller (compared to league size). Anyways, I guess in the end it wouldn't work, and even if it would it's way too much work.

I do like the #2 comparison. I wonder if a #3 or #4 comparison would work better to account for those (Lemieux as a #2 situations)?

 

[Derick*]

This is what I realized earlier, then my thought was to try to account for different talent pool sizes by doing multiple comparisons to create a picture of the what the talent pool looked like at the time...If the gap between bottom end and top end 1st liners was bigger in one year compared to another, then the talent pool of that year was smaller (compared to league size). Anyways, I guess in the end it wouldn't work, and even if it would it's way too much work.

I do like the #2 comparison. I wonder if a #3 or #4 comparison would work better to account for those (Lemieux as a #2 situations)?

Or an average of 3rd-18th...

 

[seventieslord]

I do like the #2 comparison. I wonder if a #3 or #4 comparison would work better to account for those (Lemieux as a #2 situations)?

Hockey Outsider did what he considered an improvement on BM67's vs. #2 system. Instead of #2, it compared to the average of #2-4.

It may or may not be better, I never really decided. However, I know it wouldn't "fix" the issue with the 1989 season's results

 

[monster_bertuzzi]

LL could have taken Perreault and had the best 1st line in the draft easily....Lindsay-Perreault-Conacher, sick.

 

[Leafs Forever]

LL could have taken Perreault and had the best 1st line in the draft easily....Lindsay-Perreault-Conacher, sick.

It's close, but Hull-Forsberg-Hull is quite possibly better.

 

[monster_bertuzzi]

It's close, but Hull-Forsberg-Hull is quite possibly better.

Didn't even realize they were put together already. Hull-Forsberg-Hull is better for sure, lol.

 

[seventieslord]

LL could have taken Perreault and had the best 1st line in the draft easily....Lindsay-Perreault-Conacher, sick.

but then what would his defense have looked like?

He already started with a below-average #1 defenseman; how bad would it have been if he waited yet another round?

Centers tend to fall. LL is doing what I would be doing right now, if I had been able to resist the call of the Moose.

 

[Leafs Forever]

Didn't even realize they were put together already. Hull-Forsberg-Hull is better for sure, lol.

Not sure if they're going together or not since jokerit hasn't updated their roster post, but if they are, welll then they'd be better.

 

[Dr Pepper]

I haven't even touched my blueline yet.....will probably get on that in the near future. Either that or finish off my top line of _____-Perreault-Bossy.

 

[seventieslord]

Is it actually in nik's plans to put those two together, though? I think it's a bad idea, and nik's a smart guy, something tells me he doesn't want them together either.

 

[monster_bertuzzi]

Hmmm maybe re-unite the Hull's and have Forsberg carry a 2nd line? That's what I would do.

 

[BenchBrawl]

isn't that the same principle that states that if you give a billion monkeys a billion typewriters, eventually one of them will type out the entire works of shakespeare?

I think it would take more than billions , but it is almost sure to happen yeah.Or they could all write the same thing.

 

[Hawkey Town 18]

Is it actually in nik's plans to put those two together, though? I think it's a bad idea, and nik's a smart guy, something tells me he doesn't want them together either.

The first two things I thought of when I saw that they were together...the Hulls are definitely better off on separate lines and watch out for that PP.