2008-09-29

Smid: sink? or swim? Some stats to set the stage



Why the heck are the Oilers playing Ladi Smid at such a young age? Shouldn’t he be developing his craft in the AHL rather than getting his baptism-by-fire at the NHL level? Haven’t we seen enough scenes like the above?

Such questions are cannon fodder for the ongoing Oilogosphere debate between those who like watching young players develop vs. those who lean towards veteran solutions to plug perceived holes on the roster. Those who have read my commentary over the past year will likely realize I have a strong preference towards the former. Bear in mind that I watched all the myriad of mistakes of a club that simultaneously * featured six emerging youngsters age 21 and under; today all six are in (or en route) the Hockey Hall of Fame, so that time it worked out all right. It doesn’t always, obviously, but perhaps that explains my bias.

(*1981-82)

This week the experience-vs.-potential debate has turned to young Smid, both in my previous post on Oil Droppings and in this comments thread over at
Lowetide. Are the Oilers rushing Ladi? In his defence I mentioned names like Chris Pronger, Jay Bouwmeester, Brent Burns, and Robyn Regehr as guys who were minus machines at the start of their career but took off once they gained the requisite experience. I immediately accused myself of cherry-picking and decided to do a little more rigorous research on the subject. So here goes …

Once again availing myself of the on-line bible Hockey-Reference.com, I went through 15 years of entry draft records, 1990-2004, identifying defencemen who were picked among the top 10 selections. HR doesn’t list their positions on that database, but let’s just say there were a lot more familiar names in that elite draft list than unfamiliar ones, so I just spot-checked the doubtfuls. I came up with a list of 45 blue-chip blue-liners that went near the top of their draft class. Interesting fact #1 is that 45/150 = exactly 30%, which “coincidentally” is the solution to the well-known roster equations 6 of 20 and 7 of 23.

None of the 45 were selected by the Oilers, and while a few (Hamrlik, Pronger, Brewer, Pitkanen) subsequently spent time in Edmonton, Smid is the only one to actually begin his career here.

Interestingly, every last one of the 45 blue-chippers has subsequently played in the NHL, anywhere from a cup of coffee (Boris Valabik, 2004, 7 games to date) to a long, distinguished career (Darryl Sydor, 1990, 1171 GP). As a group they have so far played 22,998 regular season games, more than 500 per, an average that will continue to soar since over 30 of these guys are still active. NHL scouts know their stuff.

[Amazing trivia: of the 25 Entry Drafts 1979-2003, 248 of the 250 players chosen in the Top 10 have actually played in the NHL! For extra points, name the two hotshots who never got a single cup of coffee. Answer below]

I’m particularly interested in the blue-chip defencemen’s performance as youngsters, and since Smid himself has just completed his 21-year-old season I chose that age as the cutoff (“hockey age” as defined by Hockey-reference.com). Of the 45 guys all but three (Ruslan Salei, Lance Ward, Lars Jonsson) played their first games at or before that age. 32 of the 45 had played at least 50 GP and 23, a rank majority, had over 100 NHL games by that age. As a group they averaged 113 GP through their 21-year-old season. Having completed his 21-year-old season with 142 GP, Ladi Smid ranks 21st, right in the middle of the pack.

So how did these guys fare in their early years? I chose +/- as the best available statistic of record. Interestingly, as a group these hot shot draft choices are virtually at the waterline with a collective -21 in those 23000 GP. Even more interesting are the splits:

Up to 21: 5069 GP, -291 (-0.057/GP)
22 & up: 17929 GP, +270 (+0.015/GP)
Totals: 22998 GP, -21, (-0.001/GP)


Surely that’s enough of a difference to be statistically significant.

Like most statistical studies, the samples are far from pure and we must recognize their limitations:
 For starters, +/- is not a perfect stat. (Eh, David?)
 Where ATOI numbers are available, virtually all played fewer minutes/GP in the early years, very probably against lesser QualComp (which is unavailable except for the most recent draft classes). Presumably coaches generally “sheltered” these guys early in their career, which would tend to reduce their raw minus figures.
 OTOH, guys drafted in the top 10 are generally coming in to non-playoff teams, so there might be an early-career bias towards minus figures due to team effects.
 As the team improves to the mean, the player’s stats should improve with them. But the player’s own improvement would contribute to that team improvement. Such self-referential data is apt to be corrupted no matter how you slice it.
 Conclusion: Handle with caution, but the career arcs from minus to plus of entire groups of players must to some extent reflect their growth as players.

One problem is that the data includes the early careers of all 45 players, but the mid-careers of a smaller number and the late careers of relatively few. Since two-thirds of them, 28 of 42, were minus players in the under-22 portion of their careers, this skews the overall result to make the entire group appear to be average (even) players when clearly they are not. To eliminate this built-in bias I therefore did a separate sort of just those 15 blue-chippers -- again, exactly 30% -- drafted in the Top 10 in the five drafts 1990-94. Some of these guys are still going strong, but their careers are pretty well defined at this point. I will show just this smaller table to confirm the method:


Year Pick Player ******** up to 21 22 & up

*************************** GP +/- GP +/-
---------------------------------------------
1990 7 Darryl Sydor ****** 182 -14 989 +38
1990 8 Derian Hatcher ***** 193 -1 852 +75
1990 9 John Slaney ********* 47 +3 221 -29
1990 10 Drake Berehowsky *** 50 -5 499 -43
1991 3 Scott Niedermayer * 213 +62 888 +122
1991 4 Scott Lachance ***** 166 +7 653 -83
1991 5 Aaron Ward *********** 5 +2 692 -44
1991 8 Richard Matvichuk ** 92 -14 704 +60
1992 1 Roman Hamrlik ***** 261 -77 815 +13
1992 3 Mike Rathje ******* 116 -26 652 +52
1992 5 Darius Kasparaitis * 155 +9 708 +30
1993 2 Chris Pronger ***** 202 -33 738 +186
1994 1 Ed Jovanovski ***** 212 -16 609 -26
1994 2 Oleg Tverdovsky *** 246 -17 467 -4
1994 10 Nolan Baumgartner *** 5 -1 126 +5

Up to 21: 2145 GP, -121 (-0.056/GP)
22 & up: 9613 GP, +352 (+0.037/GP)
Totals: 11758 GP, +231 (0.020/GP)

Given that exactly 2/3 (10 of 15) were minus players through age 21 this player sample would seem to be representative of the larger group. Results varied from player to player, from the freakish Niedermayer to the slow-developing Pronger, but as a group they improved by about +1 for every 10 GP after their 22nd birthdays.

The question remains why not play them in the AHL ‘til they’re 22 to learn the pro game and avoid those early-career minuses? Anaheim did this with Salei, who is not a bad comparable to Smid btw. But it was Ruslan's treatment that was exceptional for a high draft pick, not Ladi’s. Those same 15 top-drawer prospects that played 2145 GP at the NHL level by age 21 played just 402 AHL/IHL games during the same stage of their careers. Per player that's 143 GP in the NHL, just 27 in the minors. It would seem most teams conclude that for development purposes, AHL ≠ NHL. Defencemen training to compete at the top level must be exposed to opposition at the top level. Or so goes the common wisdom in the NHL. What the Oilers are doing with Smid is nothing out of the ordinary at all for a player of his pedigree.

Finally there is the matter of the CBA. As of 2008-09, players are free to move on after 7 years of NHL experience OR age 27 (with at least 4 years experience). Since virtually every player on the list would meet the second criterion, there is no reason to delay their start on the first. Thus the UFA clock begins ticking when the player graduates from junior, whether he is in the AHL or NHL. Time spent developing a guy more slowly in the AHL is potentially time lost at the productive end of those first 7 years.

So it is with Ladi Smid, who has surely already played the two “worst” seasons of his career. Now that Oilers and their fans have made the investment, it’s time to start reaping the return. I anticipate Smid building off of last year’s strong finish – he was an even player the last 14 games after returning (prematurely) from his knee injury, despite being paired with the guy many/most saw as Edmonton’s weakest defender, Matt Greene -- and become at least a break-even player in 2008-09.


Trivia answer: Ryan Sittler ( Philadelphia, 7th overall,1992) and Brent Krahn (Calgary, 6th overall, 2000)]

10 comments:

dstaples said...

Bruce, I've been thinking about the use of comparables for this kind of study, and I want to run a few things by you.

I believe it was Bill James who really got into the use of comparables, and his method was to look at all kind of players, and find one or two who really matched up, then compare the careers. One thing he didn't have to factor in much, though, was the style of a player and the role of that player on the team.

What I mean is, all second basemen have pretty much the exact same job, and it doesn't matte if you're 150 pounds or 225 pounds, you just have to field the ball and hit the ball. So you can compare big guys, little guys, smooth guys, choppy guys, doesn't matter. You just look at their results and compare them.

In hockey, though, not all defenders have the same role. A small guy almost always plays a much different kind of game than a big guy. They have different roles. For instance, one guy might be physical, the other guy a finesse player, and that will have an impact on their longevity.

So when I do these comparisons, I've come to think it's best -- crucial in fact -- to compare only apples with apples. Actually to compare only Gala apples with Gala apples, not Gala with Macs.

So in looking at how Smid might perform in the long term, I would be looking for guys who were kinda big and kinda fast, kinda physical and kinda good with the puck, guys who broke in really young, guys who were top picks, guys who were minus players in their first few years.

If we zero in on those 10 or 15 guys from, say, 1980 to 1997 (so we can look at pretty much full careers) that might give us a good idea. Or if we just want to see how quick these young guys turn it around, we can look at players right up until the 2002 or 2003 drafts.

In any case, I've seen LT and Tyler do some comparables work, and I've done some, and that's where I've ended up, that it's crucial to compare only sports cars with sports car, SUVs with SUVs, tanks with tanks.

Bruce said...

I believe it was Bill James who really got into the use of comparables, and his method was to look at all kind of players, and find one or two who really matched up, then compare the careers. One thing he didn't have to factor in much, though, was the style of a player and the role of that player on the team.

David: I'm a disciple of Bill James myself; his Baseball Abstracts of the 1980s taught me a huge amount about not just baseball but the science, and the art, of interpreting statistics.

What I mean is, all second basemen have pretty much the exact same job, and it doesn't matte if you're 150 pounds or 225 pounds, you just have to field the ball and hit the ball. So you can compare big guys, little guys, smooth guys, choppy guys, doesn't matter. You just look at their results and compare them.

I'd be willing to bet that the 150 pounders have different results from the 225 pounders. And whilst all second basemen have pretty much the exact same job in the field, their contribution on the offensive side of the game can be entirely different. One 2B could be a singles-hitting, base-stealing OBP fiend and another could be a slugger. Or both, if he's Joe Morgan.

James of course had many columns of statistics in doing his comparables and tried to match up as many of them as possible to find similar players. As I recall he would go across positions; Cal Ripken, say, might have no comparison at shortstop but his batting stats would compare to a slugging first baseman like Eddie Murray. As a "good hitting shortstop" Ripken would have more value; as a hitter maybe things aren't so cut and dried.

In hockey of course there are not that many columns of statistics, especially ones that measure the Smids and Saleis of the league in any sort of a meaningful way. Therefore such comparisons are often steeped more in observation and less on data. It's less reliable, but in a sport that can't easily be measured in discrete "events", we don't have much choice.

dstaples said...

I agreed, this comparison stuff is less reliable in hockey, and is more of an aid to observation.

I mean, just by watching Smid we can see a kid who has so many tools -- size, skating, stickhandling -- and that should tell us he has a future, if he can calm down a bit and read the game better, play with more confidence passing the puck, the kinds of things that come with experience. But it's nice to see the numbers mesh with those observations.



I'd have to go back and look more closely how James did his comparables. I'm going by memory, 20 years ago.

kinger said...

Nice post Bruce, it was a good read.

Jonathan said...

It's a nice post Bruce. I've thought it was a huge mistake to start smid two seasons ago; he should have been getting his feet wet last season, but that ship has sailed, and he's a good fit for a regular spot.

Outside the top-4.

Bruce said...

Jonathan: I agree third pairing is still best, and we got the depth to cover that.

At least Smid got one full year in the AHL, which even as an underage probably did him more good than a year of junior. That's one more year of minor-league ice time than many of the other hotshots from their respective draft classes.

Tangotiger said...

Good job.

If you start off at 18, you are UFA at 25. If you are 19, then 26. If you are 20, 21, 22, 23, then UFA at 27.

So, it would seem to me that you want the clock to start ticking at age 20. This way, you get him at age 26, 27, and not 18, 19.

Bruce said...

Good job.

Thanks, Tangotiger, and welcome!

So, it would seem to me that you want the clock to start ticking at age 20.

Exactly. When I wrote:
"Thus the UFA clock begins ticking when the player graduates from junior, whether he is in the AHL or NHL"
... I was referring primarily (if not specifically) to 20-year-old graduating juniors, at which age the AHL option becomes available to most (Smid being an exception). At that point 7 years is 7 years, so why wait?

This way, you get him at age 26, 27, and not 18, 19.

Yes and no. "Yes" when it comes to negotiating rights, but often if not usually "no" in reality where teams mostly lock up their own for the long term. Thus they get 18, 19 and 26, 27. A team like Toronto has to be confident they will be able to keep Luke Schenn under contract for as long as they want him, so it then becomes a question of what's in the short-term benefit of the player and the organization. Will Schenn make the Leafs better this year? Will he be a better player next year by having played in the NHL this year?? Will he be two steps further ahead at 20 fluttering as a Leaf instead of flying as a Rocket???

:D

If the answer to those quetsions is "probably, yes", then the only competitive reason to hold him back might be to insulate him from a losing environment.

On the financial side, the team's capologist has to figure out how much more expensive those 26 and 27 years might be: the price of experience. Given the CBA itself will have expired before then, it may well not matter a whole hell of a lot.

And on the marketing side, having a guy like Schenn or Sam Gagner flashing his potential under the bright lights sells legitimate hope in the future, an essential ingredient for any rebuilding club. My motto -- well, one of them -- as a fan is "Youth movement? Show me."

On the other hand, if he's Jay Bouwmeester and his team is still losing by the time he's 25, maybe it's just better for both player and organization to go their separate ways and start over. The team has obviously got other pressing needs for that cap space, and the current formula, Bouwmeester and all, isn't quite cutting it.

Robert Vollman said...

Brent Krahn suited up for the Calgary Flames for 5 play-off games starting April 22, 2007. He didn't get any ice-time, but I'm sure he had a cup of coffee.

Bruce said...

Robert: Nice catch, I remember that now that you mention it, and yes I would call that a cup o' coffee, even if he didn't get any game action.

Meaning Ryan Sittler was the only Top 10 draft pick in 25 years to never dress for an NHL game. Which is pretty amazing when you think about it.