There is little point in ‘evaluating’ the entry draft in the days or weeks after it happens. Draft ‘evaluation’, in this sense, basically consists of comparing what teams did to either the CSS rankings or one’s personal sentiments, then praising those who did similar things and condemning those who varied drastically. As I said before the draft, if you are one of the prospect experts who spends years watching and studying these children, then perhaps your evaluation matters. If you didn’t, you’re mostly just comparing lists with no added value. It’s a waste of time.

However, standing back from specific team’s specific picks and observing the draft as a whole, there is one evaluation we might definitely make: most NHL teams either don’t read or don’t believe the statistical research on draft strategies. The research says that Central Scouting Services does a fairly good job in the high rounds, yet this year NHL teams merrily went way off-board as high as the 21st pick. The research says that highly-ranked forwards are far more certain values than highly-ranked defensemen, yet eight teams decided to pass on Filip Forsberg in favor of a defenseman, and nine passed on erstwhile top-three pick Mikhail Grigorenko. And the research says it’s poor value to draft projected 3rd and 4th line talent, given how cheaply and readily available it is, yet several teams picked players scout-described as grinders in the top sixty.

One such pick was the Oilers 32nd overall selection, a one Mitch Moroz. While some observers hypothesized that the Oilers saw some nascent power-forward qualities in the young man, CSS had Moroz ranked at 72, a ridiculous forty slots lower than the Oilers’ estimation, and Redline Report specifically described him as “a feared enforcer”. The sheer apparent irrationality of this pick provoked the most industrious Tyler Dellow to do some excellent research on the provenance of bottom-six players, which can be found here and should be read immediately.

While there’s a wealth of interesting data in the piece that could be extrapolated in any number of different directions, Dellow’s major thesis might be summarized as a restatement of the ‘swing for the fences’ doctrine of drafting: that NHL teams should not bother drafting players with stereotypically bottom-six skills sets, but rather those who have potential high-end offensive upside, even where that potential seems extremely small. While there are always exceptions, it is clear that many NHL teams, at some point in the draft, switch their strategy from drafting for skill, smarts, and offensive upside to drafting for size, truculence, and defensive responsibility. Although some GMs start doing it earlier and more eagerly than others, almost all teams pick up a Coke machine or two every draft.

The research suggests this may be a poor strategy in terms of maximizing the talent level of your roster, but nevertheless, it may not be a wholly irrational one. Which is to say: I believe that NHL teams do not necessarily draft efficiently outside of the first round, but in this case the inefficiencies have a logic of their own.

For one thing, it is not entirely clear what ‘swinging for the fences’ is outside of the top two rounds. Most of the players with a clear, substantial offensive skill set are gone by then, and the players who remain all either suffer from clear drawbacks or are still of entirely uncertain quality. To be able to draft the player with the highest potential upside, one has to first be able to identify the player with said upside, and outside of the first round- hell, outside of the top-ten- no one is very good at doing that. Consider, if you will, Ben Wendorf’s reviews of scouting reports on earlier draft classes: often it is not a case of the teams passing over players they know to have skill in favor of those who don’t, but simply that the scouts didn’t realize the skill was there at all. Perhaps they don’t know what to look for, or perhaps it honestly hadn’t developed yet. Either way, there may be no difference between swinging for the fences and swinging blindly.

To a certain extent, then, the preference for Coke machines in the later rounds is a rational response to imprecise knowledge of drafting and the irrational values of the League. The kind of characteristics that popular low-round players have- size, aggression, fighting ability, defensive responsibility, work ethic- are the kind of traits hockey has always valued, and indeed overvalued. They are therefore the traits that hockey is most likely to continue to value in the future. And they have the obvious virtue of being facts about 18-20 year olds rather than speculations about 18-20 year olds. A fourth-round player’s offensive upside in the NHL is a hypothesis. It may or may not come to pass. A 200 pound body and a punchy temperment, however, are established traits.  Useless traits, maybe, but real ones.

Imagine two players with limited upside: one is small, smart, and non-physical while the other his huge, hard working, and aggressive. Assuming the scoring never materializes, which one of them is likely to contribute some sliver of value to your organization in the future? Especially considering, as Dellow points out, that you’re likely to be trading a bottom-six forward rather than using him yourself? If you’re a traditional GM, you believe size, work ethic, and fighting have inherent value. If you’re a non-traditional GM, you know that those things have inherent value to many of your traditionalist peers. Either way, your chances of squeezing some use out of a low-end pick who exemplifies hockey’s customary biases is far higher than your chances of getting value out of one who thwarts them.

The fundamental irrationality of the way bottom six forwards are drafted derives from the irrationality of the way bottom-six forward roles are assigned. The lower a player is on the roster, the greater chance that he has his position due to something other than skill level. Maybe he’s good in the room, maybe he’s a great role model for the rookies Maybe he maintains perfect conditioning, works real hard, never complains. Probably he fights; probably he’s good with the media and nice to the fans. These are the positions where GMs indulge their sentiments and ideologies- they prefer size over speed, obedience over creativity, North Americans over Europeans. They’re not giving out these jobs to the best players available. They’re giving them out to whichever of the very similar range of players available best fits their biases.

Ironically, one of these biases is itself draft pedigree, which may be responsible in part for Dellow’s finding that a bottom-six player is often a projected top-six player who didn’t work out. It may be that the high-round picks are actually better hockey players than those drafted lower and specifically for bottom-six roles. Or it may be that high picks simply get more chances from their teams, including the chance to reinvent themselves. I admit I haven’t done the research yet, but anecdotally it seems that first round picks are automatically given much longer grace periods to prove themselves as worthy NHLers than fourth-round picks. Some players have to prove that they belong in the show, others have to prove that they don’t.

It’s all very well to say that teams shouldn’t hand out roster spots on intangibles, they should rationalize and optimize, but given that a proportion of bottom-six jobs are already filled by greater talents on their way up or down, it is unclear how much real advantage a team would get from optimizing the 3-5 ideologically-filled slots on their bottom lines.These players just don’t play that much, and if the side benefits you get from them in terms of attitude/customer relations outweigh the relatively trivial talent differential, it might be a valid compromise. Scrawny 4th liners are a PR problem. Huge 4th liners are a PR boost. If I’m a smart GM and I know that the scrawny player might be a hair better in terms of possession, but only a hair, and possibly I’ll have to take shit from the media for hiring him, I might go with the big guy just because it makes my job easier and my reputation better.

There is often a seed of social rationality behind apparently irrational behaviors, and this is certainly the case with low-round drafting for bottom-six players. Faced with the difficulty, and indeed perhaps impossibility, of identifying offensive upside in the later round of the draft, scouts and GMs lean towards players who have lesser but more certain value. That value is driven by the irrational biases of their profession, but those biases are so widespread that they essentially push even those who don’t believe them to play along. There are definite benefits to be had, in terms of trades and PR, for having a few spare enormous WHL boys in your depth positions, to be weighed against the indefinite and possibly nonexistant benefits of trying to guess the future offensive skill of ambiguous, late-blooming fifth rounders. The Oilers may have been wrong to take Moroz quite as high as they did, but it’s not all that far below him that all teams are drafting future third-liners no matter what their intention. Some just choose to do it explicitly.

However, the interesting take-away from Dellow’s article is that the opportunity cost of optimization is relatively low. Older players with the qualities of career fourth-liners are cheap and plentiful, many of them come from outside of the draft, and they’re usually available for trade with the people who do draft them. If a team wanted to experiment with disregarding size and truculence and attempted to draft only the skilliest skill forward available at any given moment, they might find they get a significant bonus. They might find that swinging for the fences pays off, in terms of either being able to find future top-tier guys at a higher rate than normal or in terms of icing a more competent third/fourth line. Or they might not, and all those low-round skill guys will crumble and blow away like dust as just as the conventional wisdom suggests, and they could just let them fly away and start buying up all the discount Coke machines out there. There may or may not be benefits to be had by flouting lower-round drafting conventions, but either way, it costs almost nothing to test it.

[N.B.: Of course, the research suggest the smartest thing to do with lower-round picks is not to draft draft forwards at all, but rather defensemen and goalies. But we've already established that NHL teams don't read the research.]

[Also N.B.:  There are several online articles I remember reading in the past on A) the virtues of drafting for highest possible upside and B) the usual accuracy of CSS in the high rounds that I was unable to locate for this article despite several afternoons of vigorous Googling.  If you did relevant research on the topic and I neglected to refer to it here, PLEASE leave a link in the comments.]