2009-04-08

The missing link


I know the pieces fit cuz I watched them tumble down
No fault, none to blame it doesn't mean I don't desire to
Point the finger, blame the other, watch the temple topple over
To bring the pieces back together, rediscover communication

Tool -- Schism
*****

I cited this tune over at Lowetide's place the other day, finding some of Maynard James Keenan's lyrics resonated with the (hotly-denied!) reports of a rift in the Oilers' room and with the state of the roster in general. Rather than dally with the rumours or try to divine the true intelligence behind Ethan Moreau's remarks, I focussed on matters of record, namely birth certificates. These reveal a stunning gap where the heart of the roster should be. To recap the findings of my previous post, here are Oilers players and GP by age group through 78 games in 2008-09:

Age
Group ..... # ..... GP ... AvGP .. %team
----------------------------------------
36-39 ..... 1 ..... 61 ..... 61 ..... 4%
33-35 ..... 3 .... 217 ..... 72 .... 15%
30-32 ..... 8 .... 332 ..... 42 .... 22%
27-29 ..... 1 ..... 20 ..... 20 ..... 1%
24-26 ..... 9 .... 443 ..... 49 .... 30%
21-23 .... 10 .... 343 ..... 34 .... 23%
18-20 ..... 1 ..... 72 ..... 72 ..... 5%

----------------------------------------
Total .... 33 ... 1488 ..... 45 ... 100%

I have added two columns. The first is simply the average number of GP for players in the given age group; the second expresses the percentage of team GP for each age group. Here's a crude graph showing that % of GP distribution, featuring a very odd double-peaked curve.


That looks not so much like a team, but two different ones.

Comparison is the next stage of the analysis; it is obvious at first glance that the current age distribution is out of the ordinary, but how much so? Surely the place to start is with the '05-06 Oilers, the last time we know the pieces fit. Here was the distribution by age of the House of Cards that Kevin Built.


Age
Group ..... # ..... GP ... AvGP .. %team
----------------------------------------
36-39 ..... 2 ..... 56 ..... 28 ..... 4%
33-35 ..... 2 ..... 43 ..... 22 ..... 3%
30-32 ..... 9 .... 536 ..... 60 .... 34%
27-29 ..... 9 .... 472 ..... 52 .... 30%
24-26 ..... 7 .... 235 ..... 33 .... 15%
21-23 ..... 6 .... 204 ..... 34 .... 13%
18-20 ..... 3 ..... 25 ...... 8 ..... 2%
----------------------------------------
Total .... 38 ... 1571 ..... 41 ... 100%

In each column there's a fairly normal Bell curve with a single peak. Almost half (18 of 38, or 47%) of the players who suited up for the Oil during that chaotic season were between 27 and 32; moreover the same group played more games per player than the other groups; combine the two and the 27-to-32-year olds played fully 64% of the games. It's an interesting contrast to 2008-09, where not only is that huge hole at the age 27-29 level, but the distribution within groups is strange. Players over 33 averaged 70 GP; those between 27-32 averaged just 39. In 2008-09 the 27-32 y.o. group comprised just 27% of the full roster and played only 23% of the games. Let's compare the two graphically, with the 2005-06 Oilers represented in copper, the '08-09 squad in blue.


There are superficial similarities between the central curve of the '05-06 Oilers and the younger lobe of the current version, but there is a 5-year difference in maturity. More generally, the above starkly demonstrates the '08-09 Oilers have far more players near the beginning or end of their careers, while the '05-06 club had a critical mass of players near their peak.

Of course the 2005-06 regular season Oilers that stumbled and bumbled their way to 8th place was different from the leaner version that was wildly successful in that same post-season. Some guys were traded off mid-season; others who had a cup o' coffee at some point were out of the mix come playoff time, as was the case with all three 20-year-olds (Pouliot, Jacques, Syvret).

Age
Group ..... # ..... GP ... %team
--------------------------------
36-39 ..... 1 ..... 18 ...... 4%
33-35 ..... 1 ..... 24 ...... 5%
30-32 ..... 9 .... 169 ..... 37%
27-29 ..... 8 .... 130 ..... 28%
24-26 ..... 4 ..... 74 ..... 16%
21-23 ..... 2 ..... 42 ...... 9%
18-20 ..... 0 ...... 0 ...... 0%
--------------------------------
Total .... 25 .... 457 .... 100%

The 27-32 y.o. set comprised 68% of the active players and played 65% of the games, while the under-24s played an even smaller role with only Hemsky and Greene getting a sniff of playoff action. I won't show a graph, which varies but little between regular season and playoffs. I have noted for future reference that playoff rosters might provide a more streamlined methodology, a little less clutter with fewer players and a more constant "best possible" line-up. Of course, for that method to work the team in question has to actually make the @#$%^&* playoffs.

Which brings us back to 2008-09, where it's instructive to compare the Oilers with the competition. I'm using brute force, not code, so won't do the whole league. Let's just compare with geographic rivals and Northwest Division co-leaders Calgary and Vancouver. Here I have reduced each team to just the number of players in each age category, and the percentage of GP for each group.

Age ..... Edmonton .. Vancouver .. Calgary
Group ... # / %team . # / %team . # / %team
-------------------------------------------
36-39 ... 1 // 4% ... 1 // 3% ... 1 // 5%
33-35 ... 3 / 15% ... 3 // 8% ... 3 / 12%
30-32 ... 8 / 22% ... 3 / 14% ... 6 / 23%
27-29 ... 1 // 1% .. 13 / 38% ... 6 / 20%
24-26 ... 9 / 30% ... 5 / 16% .. 11 / 30%
21-23 .. 10 / 23% ... 7 / 21% ... 5 / 11%
18-20 ... 1 // 5% ... 0 // 0% ... 0 // 0%
-------------------------------------------
Total .. 33 /100% .. 32 /100% .. 32 /100%

The Oilers have gotten more GP out of guys 33 and older than either rival, and more GP from guys 23 and younger. In the middle both divisional rivals far outstrip the Oilers; note the 13 players Vancouver has used in the 27-29 group, compared to Edmonton's 1. Calgary's curve peaks a little younger, but fully 73% of their games have been played by players aged 24-32, compared to 68% for Vancouver and just 53% for Edmonton. Here's a graph showing Vancouver in green, Calgary in red, Edmonton in blue.

Next year Dustin Penner and Tom Gilbert -- dangerously assuming either is still around -- will turn 27; Pouliot, Jacques, Stortini, Reddox and O'Sullivan all "graduate" to age 24, and their whole generation of players will be one year older and presumably better. At the same time, Sheldon Souray, Lubo Visnovsky, and Fernando Pisani all turn 33, Steve Staios turns 36, and lest we forget, Dwayne Roloson turns 40. Whatever the validity of my artificial brackets, it's pretty tough to argue those guys on the far side of 30 will be getting much better; they'll just be getting older. The wheel of time keeps turning, but given the void of players cycling through the top of the career curve ours seems to be missing an axle.

It is incumbent on Oilers management to address this gaping hole in the coming l-o-n-g off-season. Such a hole cannot be readily filled from within.

8 comments:

B.C.B. said...

I love that graph at the end, I think it shows why the Oilers had a tough year. That being said, I think the number of 18-20 year olds is going to be increasing as younger players on entry level deals are cheaper then middling vets, and you need to develop the top end kids somewhere (lets just hope it is on the fourth line, or bottom 2 d pairing).

PS: how do you get charts in blogger?

Scott Reynolds said...

Good stuff Bruce. I think that Mirtle posted a nice long while back (making it particularly difficult to find) a graph of the number of NHL players and their ages. Such a chart might prove helpful in developing a choice for the age ranges. I know that Dan Tolensky did an article which tracks just the top goal scorers of all time and measures only goal production. At the same time, it should give a decent overview of what the career curve has been. His study would seem to suggest age ranges of:

18-20 (generally poor performance)
21-22 (improved performance)
23-29 (core of career)
30-32 (decline in performance)
33+ (heavy decline in performance)

This is obviously very close to what you have already. So congrats on your intuition!

From a team building perspective, it would seem to make sense to start moving players along in their age 29 or so season just before their decline while their value remains (I would imagine) quite high. I don't know that you really need to hold on to some veterans, but I think it's probably wise. From an Oilers perspective the guys to consider moving along at the right price are Horcoff, Pisani, Moreau, Visnovsky, Souray and Staios. Now, you probably don't want to move all of those guys out, but at this point they're all probably going to be taking on water going forward in terms of their quality of play. I know that it's not going to be popular around this part of the hockey world, but I think this is the most compelling argument to move Horcoff. Paying him for his 30-35 seasons after underpaying for his previoius work might have us all sobbing once he hits the "heavy decline" stage of his career. I hope all of that fitness stuff helps.

Scott Reynolds said...

I don't mean to hog up your comments Bruce but here's a link to the Tolensky article:

http://www.hockeybuzz.com/blog.php?post_id=8663

Also, Scott Howson is a smart cat. He has 0 players on a long term deal past the age of 32. Anyone older than 32 expires next summer at the latest.

Bruce said...

I love that graph at the end, I think it shows why the Oilers had a tough year.

BCB: Thanks. I liked that graph too, so much so that I have done a minor revision to the post and added a couple more. I work with tables a lot and can "see" a graph by simply looking at the numbers, but sometimes the visual tells the story at a glance. As it does in this case.

To get the chart in there, I copied it (using Shift+Edit in Excel) to "Paint", then saved it as an image file which I inserted as a picture. Crude, but when it comes to computing chops, "crude" is the top of the mountain AFAIC. :)

I don't mean to hog up your comments Bruce but here's a link to the Tolensky article:

Scott: No worries about that on my lowly blog, I welcome all the comments I can get. Thanks for the link, that was an excellent read. To some extent it's apples and oranges, the peaks of fully mature hockey players occur later than those of pure goal scorers; it's ideal to have the average age of your team a little higher than of its top snipers, as we saw in spades here in the 1980s. Still, it's a single smooth curve without a gaping hole in the centre of it.

Also, Scott Howson is a smart cat. He has 0 players on a long term deal past the age of 32. Anyone older than 32 expires next summer at the latest.

Yeah, we could use a guy like that around here. The Oilers have Visnovsky, Souray, Horcoff, Staios and Moreau each signed for several more seasons through age 35 or 36 in each case. That's a $21.2 MM cap hit to begin the season, all on the right end of the already-misshapen age distribution graph. That would be "right" as in "wrong", as that entire core group will be redshifting away from their peaks even as Oilers continue to pay them peak value. To a significant extent the team is hooped; those contracts and many others like them around the league are going to be very difficult to move in the current environment.

B.C.B. said...

Thanks Bruce, for the idea. Stupid Macs don't have paint, so I am going to have to find another way of doing it.

Just letting you know, the last installment of the Aggressive Penalties and Lazy Penalties is up (over at my site), and as usual it shows 'our' boy Zorg in a great light.

raventalon40 said...

Very interesting stuff, Bruce. I think even if there is no verbally acknowledged schism there might be an unavoidably physical and temporal schism in the room - albeit unspoken.

Doogie2K said...

I think when I took screencaps on my Mac, I used Shift-Command-3 to get the screenshot on the desktop, then used Preview or something like that to crop. (I may have also moved to my PC and used Paint or IrfanView; I'm not sure now.)

MikeP said...

re Macs, I use Grab to get screenshots and then Gimp to edit a bit if necessary. Gimp's pretty heavyweight, but it gets the job done and it's free.