Thursday, June 22, 2006

2006 Draft Outlook

Context Matters

Even before the puck dropped in 05-06 there were a group of Oiler fans out there who felt that the 05-06 season was the season that Oilers should be trading in some future for some right now. The feeling was the drafting and free agent collection process had put a solid foundation in place for later, and that if there ever was a time to "go for it"...well, 05-06 was a top contender because of its strategic importance. Many reasons could be readily listed: the cap was going to rise faster the than the budget so 05-06 was the lone opportunity to spend on par with the field; the place was unattractive to UFAs; the fan base were expecting a post-cap resurgence and post-season appearance; and to avoid making an omelet facial the ownership needed to be proved right about just needing a level playing field.

As I recall it there was debate about how to "go for it". Given the known budget restraints, certain Oilers fans suggested a suitable plan would be to save up budget space by gliding through the season with the cheapest possible squad that could still cling to a playoff spot when the trade deadline came. Then the plan would be max out the rest and to load up full bore to try to become the first 8th-seeded team to reach the finals. Given that picks seemed to be the trading chip of choice at that part of the season, such a plan seemed to all but ensure that Lowe and Co. would be scarce commodities at the 2006 NHL Entry Draft.

As it turns out — whether by design or accident — Lowe's dealings seemed to follow the broad strokes of such a plan. Prospects were traded in early, saving the picks for the deadline. Two prospect/suspect Ds plus a bag of cash were used in to upgrade Brewer. Some budget room turned York into Peca. Then Spacek and Tarnstrom were obtained for some prospects and toe jam. And finally the deadline is when we saw the top three picks leave. 1st, 2nd, and 3rd round 2006 picks for Samsonov and Roloson, with Stastny and Reasoner thrown in to seal the deal.

Prior to last seasons' trades the Oilers had added significant pieces to their stable of prospects via assets acquired through earlier deals. They acquired enough picks to draft an incredible 14 prospects in 2002. Then 12, 10, and 8 prospects in the years to 2005, two fewer each year. The trend may continue too since they have only 6 picks in 2006, 5 if Florida resigns Semenov. That's less than half of the number of picks early in Lowe's tenure. A precipitous drop, and nothing in the first round.

Considering the number of prospects and picks they've sent packing recently, you almost have to wonder if the dolts in the main office should have thought ahead and shelved their AHL team for a few years!

2006 Expected Values

I have heard this year is looking like a poor draft year, with the talent dropoff occurring unusually early. If this is true, the 2005-06 year was a good year to make a push at the deadline, adding another possible rationale to the list. From what I've seen of the draft, the normal course appears to be that the majority of good players can be pegged early and the GMs do a decent job of selecting them in the right order. At some point in the draft the ability to predict player value approaches random chance and the later rounds tend to look just like what you'd expect a series of guesses to look like. If the draft is shallow, the tail-end "guesswork" part of distribution should appear earlier.

With that context and assuming no trades, here is an estimation of the number player games and points that the various teams can expect for the upcoming draft:

You Should See A Table Here

The table is ordered by the expected number of games picked by teams. Edmonton comes in third last, just above the cup champs (still smarts to type that). Edmonton was #8 by this measure last year, with about double the expected games played. Its obvious that having top picks generally will push you towards the top.

Columns 2-11 list the picks for each team, and define a set of brackets. In past versions of this table, I've used the per-round brackets of 1-30, 31-60, etc. However this linear scale makes little sense applied to the expected value estimation process. Any reasonable look at the draft history shows the pick values non-linearly decrease. It looks roughly logarithmic to me so I've used a log scale for the brackets (picked a log factor that made them look about right to me). Nothing terribly meaningful there but I find it easier to read the importance of the picks.

To estimate the games and points expected I used the pick weighting technique by Dawson and Magee, using the 13 years of historical draft data from 1987-2000. I didn't do anything to discount the draft year, which the model has provision to do. So if the draft class is shallow, consider the table to be an optimistic estimation, probably especially for the lower half of the table. In case you're curious, by this measure the 1st round pick the Oilers gave up to get to the finals was an expected 289/263 games (forwards/D) and 155/52 points (forwards/D) in an average draft class.

If you believe that Lowe's regime isn't strong at the draft, maybe this turn is encouraging, because they've been pretty successful in the trade and sign mode of building a better team, and not all that many GMs can claim to have succeeded there. I'm interested to see if Lowe makes deals with the picks, to see if he's looking for more or fewer swings at bat.


Blogger speeds said...

to what extent has the expected games measure changed since a team now only holds a player for a maximum of 7 years (574 games), as compared to far more during the previous CBA's?

6/22/2006 9:24 pm  
Anonymous oilswell said...

Both expected values relate to the player's total NHL career. Players could be traded before they ever play for their original team, but you can still try to evaluate the quality of the original pick by GP or P. But you nail another important point about determining pick worth to the club, of course. Obviously it would be nice how many games on average it takes players at various positions (in aggregate and by pick position) to become a useful NHL player.

6/22/2006 10:13 pm  
Blogger Vic Ferrari said...

Good stuff, and I'll echo speeds' sentiment as well, esp with regard to Defencemen.

Anyhow, just to put the average player that you show this draft position produces (289 games/155 points) into terms that are simpler for me to understand ... I looked up the guys with career results closest to that in the past 15 years.

GP Pts
Jim Campbell 285 136
Alexander Semak 289 174
Tim Sweeney 291 138

Those are the three best fits that I could find.

J.Money at OilFans once called first round draft picks the gold bars of hockey forums. It's funny cuz it's true. :D

6/24/2006 10:10 am  
Anonymous Anonymous said...

Well there's a lot of variability there so the busts (0 GP, 0 PTS) certainly bring the averages down. Bust rate is pretty good by that point. I don't have the deviation, but if you kept 10 1st rounders like that it's probably a decent bet you'd only get about 2900 games and 1600 points. That's what? 3 Todd Marchants + 7 busts? A Stillman and 2 Maltbys + 7 busts?

6/24/2006 9:14 pm  
Anonymous oilswell said...

Crap. That's me again up there.

6/24/2006 9:15 pm  
Blogger Jonh Neo said...

This is a some good links : Google win money playing game
Do you like car game ?
If you have free time, I comment that you should visit win money playing game
Google win money playing game

7/10/2006 7:32 pm  

Post a Comment

<< Home