When putting value on draft picks goes wrong

Trying to establish some sort of value system for draft picks has been the focus of many analytics-focused front offices and independent sports outlets for a little while now. And it’s no doubt a worthwhile endeavor that can likely help teams stay disciplined at the draft and avoid letting their gut get the best of them.

But the general takeaway of draft pick value in the NHL is that as you get outside the first few selections, the probability of landing an impact player essentially goes off a cliff, so you might be best off accumulating a few extra lottery tickets in the later rounds and increase your odds that one hits.

It’s easy to get behind this sort of approach later down the line, but trying to apply this sort of thinking to the top of the draft can be tricky. Or, in the case of this piece from TSN yesterday, almost disastrous.

The big problem with this way of looking at draft pick value is it uses expected games played as currency. That might be useful when we talk about flipping a fourth rounder for two fifths, since, again, at that point finding an NHL player at all is incredibly unlikely. But up near the top, where you have Auston Matthews, the same standards can’t apply.

Take this proposed deal, for example:

4

Maybe a slight overpay here, but considering the premiums most teams have to pay to move up at the draft, I think this might be the most reasonable package yet. This could, if you pay attention to the mock drafts, look something like Auston Matthews to Arizona in exchange for one of Alexander Nylander or Mikhail Sergachev plus one of Adam Mascherin or Luke Kunin.

A slight overpay. BY ARIZONA.

If the Leafs made a move like this, they’d have to turf their entire front office. Sure, value-wise, you can expect more games played here. But those games won’t be played by Auston Matthews.

There are other proposals sprinkled throughout this piece, including one where Philadelphia uses a bunch of later round selections to make up enough value for the top spot. Again, this approach doesn’t factor in for high-end players versus fourth-line grinders, so I guess it works if you don’t care which of these you build your team out of.

While I do think it’s more likely than not that Toronto stays at one and selects Matthews with the pick, I don’t think it’s an absolute guarantee. Doubly so for a numbers-savvy Leafs organization that you can bet is doing the exact type of draft valuation exercise I went through here, looking for opportunities to accumulate a higher volume of picks

Make no mistake, the Leafs are numbers-savvy, which is why I know they’ll put a premium on top line skill. They’re also likely a team that factors in risk, which these pick valuations do not. When you have an NHL-ready elite center at the top of the draft, you don’t move that pick to get two or three prospects who’ll likely go back to junior and hopefully turn into pros. You just make the pick. 

Even the slide-down from first overall to second is hardly worth exploring since, as the TSN piece even points out, using this value method, the gap is closed by a fourth round pick. A fourth round pick to land you Connor McDavid over Jack Eichel. Think about that.

Lastly, like most hockey pundits who dig into this idea of trading down picks for volume, we again get a mention of Bill Belichick and the Patriots’ approach in the draft over the last couple decades. If you want the Coles Notes on that whole thing, the basic premise is that Bill Belichick (arguably the smartest sports human on this planet) takes advantage of this simple fact:

Perhaps the most common characteristic of bad organizations is that they
make a habit of giving away draft picks to trade up and acquire a
player with an earlier selection…The evidence suggests the NFL draft is most likely a crapshoot, so even if your team’s draft board has a first-round grade on a player left in the middle of the second round, chances are that the rest of the league is right and you’re wrong. Teams do trade up and succeed, of course, in the same way that a drunk blackjack player hits on 16 against a five and wins sometimes, but it’s not an optimal strategy.

The NHL draft, like the NFL’s and other leagues’, is mostly a crapshoot. That much is true. But the key word there is “mostly”, and at the top of the order is where it isn’t. Why on earth would a team trade essentially a can’t-miss player from the number one selection to grab more tickets for the actual crapshoot portion of the draft? It makes no sense.

Also, it’s worth noting that Belichick article is commonly shared, and oversimplified when it is. If you look through and view some of what are considered his best trades, it isn’t always just a trade-down scenario. In some cases the Pats will trade for higher picks in the following draft, or even unload picks for a roster player. The main takeaway from that piece is actually that New England doesn’t get too attached to its aging free agents, lets them walk, and accumulates value through compensatory picks. Calling every trade-down at the draft “Belichicking” or “The Hoodie Play” and applying some level of genius to it is a bit foolish, and misrepresents what is actually going on there. Not to mention, NHL teams have been trading down in the draft literally every year. It’s nothing new. 

Toronto’s front office is going to crunch the numbers and do their homework on every pick. We know that. But they’re also going to look at managing risk, and moving the Matthews pick would obviously present way too much of it.

If the end-goal was to see how many prospects you can turn into SPC slots (pro contracts), maybe the volume plays put forth at TSN would make sense. But since those slots are limited, it might be better to try to make the most of them and win games with high-end talent instead. 

  • Mitch92

    Great article, can’t believe TSN had that piece.

    If you’re using games played as your measurement Pittsburgh should have traded the 1st overall in 2005 for, let’s say the 7th and 25th.

    1st overall = Sidney Crosby (707 games)

    7th overall = Jack Skille (313 games)
    25th overall = Andrew Cogliano (704 games)

    What a steal for Pittsburgh, they get 1,017 games!!

  • magesticRAGE

    Yost’s analysis is far too simplistic. He completely ignores potential point production. I know its very difficult to project how a player’s junior game will translate to the pros but simply valuing based on games played doesn’t tell you anything about impact. I guess it allows you to ignore position but not much else. I imagine the teams are doing a much much more intricate pick valuation.

  • BarelyComments

    BTW: Great critique, Ryan. Sadly, there are many people reading Yost’s article right now that are nodding their heads and going, “Boy I hope the Leafs trade that pick for a bunch of other picks.”

    • JB#1

      Oh I was shaking my head alright… as in “No, no, no” way would anyone in their right mind follow what the Yost article is advocating concerning the trading of the 1st overall pick.

      That Yost article was nothing more than click-bait.

    • MaxPower417

      Doubtful. Draft pick valuation tables were all the rage on Twitter during last years draft. EVERYONE was espousing warnings that they break down around the top of the draft, and to not take those results seriously.

      This is incredibly poor work by Yost, but I don’t see it swaying too many peoples thinking in any other way than “Draft pick valuation tables are useless!1!!” which is unfortunate, because as Ryan points out, they are still a useful tool for the large majority of draft positions.

  • magesticRAGE

    Great article, couldn’t agree more.

    GM’s typically trade down to acquire volume of picks, pick plus an equal talent, based on need, or the next year draft is better to commit to. The Leafs are not in a position to trade down. They need a center, it’s a good draft, there is no equal talent available, and they already have a good volume of picks, especially in the first 3 rounds.
    Further more, last year was focused on getting as many picks as possible and fill the bare cupboards, which they did, and happened to acquire good skill with a lot of those picks. With 12 picks and limited SPC’s to hand out, they can now focus on quality instead of quantity. The Leafs are in a “best player available” mode, and some may require trading up to get.

  • JB#1

    You miss the important point that these “draft values” work over large samples. So for example if a team traded that 1st pick for pick 7 and pick 20 over the past 10 or 20 drafts then the team trading that 1st pick would “win” more often then not. An example of this would be trading Yakupov pick.

    That said, Matthews may not be “average” 1st round pick we have seen over the past 20 years. And hence trading him requires more then the 7 and 20th pick.

    • JB#1

      I am missing the important point? Really…

      Did you even look at the data before cherry-picking the Yakupov draft?

      Let’s have a look at the data:

      I’m going to leave 2015 out of it as McDavid takes this analytical theory out behind the barn and puts a bullet in its’ head.

      2014
      1. Aaron Ekblad
      7. Haydn Fleury
      20. Nick Schmaltz

      2013
      1. Nathan McKinnon
      7. Darnell Nurse
      20. Anthony Mantha

      2012
      1. Nail Yakupov
      7. Matthew Dumba
      20. Scott Laughton

      2011
      1. Ryan Nugent-Hopkins
      7. Mark Schiefele
      20. Connor Murphy

      2010
      1. Taylor Hall
      7. Jeff Skinner
      20. Beau Bennet

      2009
      1. John Tavares
      7. Naxem Kadri
      20. Jacob Josefson

      2008
      1. Steven Stamkos
      7. Colin Wilson
      20. Michael Del Zotto

      2007
      1. Patrick Kane
      7. Jakub Voracek
      20. Angelo Esposito

      2006
      1. Erik Johnson
      7. Kyle Okposo
      20. David Fischer

      2005
      1. Sidney Crosby
      7. Jack Skille
      20. Kenndal McArdle

      2004
      1. Alex Ovechkin
      7. Rostislav Olesz
      20. Travis Zajac

      2003
      1. Marc-Andre Fleury
      7. Ryan Suter
      20. Brent Burns

      2002
      1. Rick Nash
      7. Joffrey Lupul
      20. Daniel Paille

      2001
      1. Iiya Kovalchuk
      7. Mike Komisarek
      20. Marcel Goc

      2000
      1. Rick DiPietro
      7. Lars Jonsson
      20. Alexander Frolov

      Oh dear, oh dear, oh dear… I think this theory still has a date behind the barn…

      Years I win – 2013, 2010, 2009, 2008, 2007, 2005, 2004, 2002, 2001

      Years you win – 2012, 2003,

      Debatable Years – 2011, 2006, 2000

      I’m giving you the 2012 Yakupov draft even though it could be debatable who’s the better player between Yakupov and Dumba. Even with that, you still need one of the deepest drafts (2003) in the history of NHL drafts, and the Penguins going off the board and picking a goalie 1st overall, to rescue your “results”.

      Even leaving out 2015 and 2014, which I think would be Years I would win, these results point out just how dangerous (to your GM job security) following a plan as poorly laid out as this would be.

      P.S. Sorry for the long post folks.

      • magesticRAGE

        basically the original draft value ranking is erroneous. And so both Yost and the mittenstringer bloggers here are using bad rank value rankings data to write a column

        garbage in = garbage out

  • MaxPower417

    I’m stunned that Yost wrote an article like that. It’s a blatant misuse of draftpick valuation models. He knows better than that. It’s almost enough to get me all tinfoil-hatty.

  • MaxPower417

    Yost deserves the criticism. I’ll add this. Imagine trading a guy who plays “X” number of games for 6 guys who play the same number of games total. That means that the 6 guys play an average of 13 games a season each, while the one player plays 80. It’s rediculous.
    In addition, we’re still playing 5 skaters per side aren’t we? You can’t just total up 6 or 3 or 2 guys stats to equal just one. They can’t collectively play one guy’s minutes all at the same tine.

  • FlareKnight

    Thank god someone wrote up an article ripping into the stupidity of Yost. Looking at numbers to the point of being just plain dumb.

    It was just a series of stupid trade scenarios. In no world is Auston Matthews worth less than the guy going 7th overall and 20th. You lose that trade every single time.

    I can’t believe that guy is actually paid money to write something that stupid. I write dumb comments all the time and don’t get paid that kind of money!

    Just well done in countering that dumb article. It really is suggesting the Leafs give up the winning lotto ticket for more tickets to the crapshoot portion….

    It is important to not get too attached to the midround picks and be willing to move back for more picks since at that point….it’s all a crapshoot. But, the top of the draft isn’t that. These are guys that have insanely more value for good reason.

  • Mitch92

    Yost’s article is getting torn apart as it should be. The mistake is that the number of games played does not directly correlate to a player’s points per game average and so both must be evaluated and compared to draw any accurate conclusions as to player value/ranking.

  • Gary Empey

    While it’s intentions may be good, this is just another analytics-focused tool that is complete rubbish. Not only does it not work at the top of the draft it does not work anywhere.

    There are 30 – 2nd round picks. 22 of them will never play in the NHL. They have zero value. You can not take the value of the eight that will play in the NHL and spread their value over the other 22 and come up with an specific value for each player.

    Even if you hold all 30 – 2nd round picks you could only put a value on the whole 30 picks combined, not each individual player.