Trading up vs. trading down

[Kings Rule]

There were some comments on the draft coverage about some teams, like Carolina, believing that because of the lack of scouting opportunities the talent will be spread out more than usual. If this is the case, then lower picks will have more value this year than normal. I was curious about the value of different picks so found a "Proposed NHL DRAFT Value Pick Chart" by Michael Schuckers and used that to value the trading up done by the kings in this draft.

Here are the picks the Kings acquired by trading up and their values:
Pick Value
42 203
59 175
84 147

Total 525

Here are the picks the Kings gave up and their values:
49 186
72 169
89 134
109 94
136 84
168 74

Total 741

Points Lost 216 Equivalent in value to a high second round pick.

If you also consider that the lower picks might have higher than normal value this year, that might be the equivalent of a low to mid first round pick value that was given up. And if you believe that your scouting is superior, then even more value was given up.

Of course, every team thinks they are better at judging talent. Every year you hear how each team's picks were so much better than everyone else rated them, but just based on these raw numbers it appears the Kings made a terrible decision in trying to trade up this year.

What are your thoughts?

Click to read more...

 

[Bandit]

Sounds like a lot of overthinking to me. Do the best you can at the time with the info you have. Trying to be the smartest guy usually blows up in your face. Kings got 4 guys they are very happy with, not sure having a couple more picks at the ass end of the draft and hoping they're the next Dasyuk is really a strategy.

 

[Gjwrams]

There is a time for both. For the Kings, they have a ton of prospect, so much so, there is going to be possible issues playing them all. I loved this year's strategy. Trade up to higher quality assets. 4 picks this year is fine. Do you want to do it every year, probably NOT, but you can't be afraid to go after the dude you want. I loved this years draft. Aside from nit being able to find that partner to get Wallstedt, it was steller.

 

[King'sPawn]

If the Kings got the quality they think they got, then losing the equivalent of an extra second rounder means jack.

And since we're talking about value, what would you rather have? An extra 2nd rounder, or a 4th, 5th, and 6th? If you choose the second round pick, then how reliable is the value equivalency?

 

[Maynard]

There were some comments on the draft coverage about some teams, like Carolina, believing that because of the lack of scouting opportunities the talent will be spread out more than usual. If this is the case, then lower picks will have more value this year than normal. I was curious about the value of different picks so found a "Proposed NHL DRAFT Value Pick Chart" by Michael Schuckers and used that to value the trading up done by the kings in this draft.

Here are the picks the Kings acquired by trading up and their values:
Pick Value
42 203
59 175
84 147

Total 525

Here are the picks the Kings gave up and their values:
49 186
72 169
89 134
109 94
136 84
168 74

Total 741

Points Lost 216 Equivalent in value to a high second round pick.

If you also consider that the lower picks might have higher than normal value this year, that might be the equivalent of a low to mid first round pick value that was given up. And if you believe that your scouting is superior, then even more value was given up.

Of course, every team thinks they are better at judging talent. Every year you hear how each team's picks were so much better than everyone else rated them, but just based on these raw numbers it appears the Kings made a terrible decision in trying to trade up this year.

What are your thoughts?

Seems like a silly exercise to assign number values to draft picks because too much context is missing for it to be useful. It doesn’t account for all the different conditions and scenarios that define each draft. A high second rounder has different value year to year depending on the draft class right?

 

[bland]

Nerds!

 

[Mats26]

I think there is some truth to this. I mean every time you move up it cost you an asset or 2 and you always need to see what you are giving up to move up.

But every case is different.
Some good players simply fall in the draft and their value points are lower than they should be , those are the ones you trade up for regardless if the value pts you are giving up is more.

 

[Kings Rule]

Seems like a silly exercise to assign number values to draft picks because too much context is missing for it to be useful. It doesn’t account for all the different conditions and scenarios that define each draft. A high second rounder has different value year to year depending on the draft class right?

The analysis on the draft pick value is very sophisticated and uses ten years of data for the entire NHL. I think its a statistically valid analysis done at a PhD level. You can find it here: https://myslu.stlawu.edu/~msch/sports/Schuckers_NHL_Draft.pdf

 

[BringTheReign]

There’s also the contract limit to consider. We’ve had a ton of picks recently and we don’t want to put ourselves in a situation where we’re losing or forced to trade a player due to not having room to sign them. I’m happy that they used their picks to get guys they wanted, especially in an uncertain year like this one.

 

[Steve Zissou]

Nerds!

 

[Statto]

The analysis on the draft pick value is very sophisticated and uses ten years of data for the entire NHL. I think its a statistically valid analysis done at a PhD level. You can find it here: https://myslu.stlawu.edu/~msch/sports/Schuckers_NHL_Draft.pdf

Some of the most invalid correlations are done at ‘PHD level’, often using data sets that are too small to matter but just big enough for whatever paper is being written. That’s why we spend so much time thinking ‘random food item’ is bad for us and years later someone proves the exact opposite with meaningful data. 10 years of draft data is nowhere near enough to be statistically significant as it doesn’t even cover a generation (it should be multiple). However given the variables change and evolve so much over time, I just don’t see it as a workable model even with 50 years of data.

 

[Go Rams]

Kings have soooooo many prospects I think they wanted to draft four "sure things" rather than throw darts.

 

[kilowatt]

I'd be interested in seeing some kind of aggregate prediction analysis compared to real production and relate that to guys the consensus are high on that fall. Off the top of my head, guys line Caufield and Kaliyev look to be worth more than their draft slots would suggest, especially when you consider that most "experts" had them rated/ranked higher than they ultimately were drafted.

All of this to say I wonder if the actual value of the 50th pick for the 60th and 100th picks is skewed if the player should have been, say, the 25th pick.

I feel like that was a lot of word vomit though.

 

[Reclamation Project]

All depends on the team and their position. Kings have the deepest prospect pool in the entire NHL, I don't care to have 10-12 selections in a draft like other teams. At this point you need to start narrowing it down and focusing your assets on high probability of success game breakers or later round gambles. I'd rather get 3-4 guys that were at the top of the scouts' list instead of getting 7-8 guys that were reaches. I trust the Kings scouts and Rob Blake. Look at the pipeline they've built in the last couple years - nearly universally regarded as the BEST pipeline in the entire league. Just trust the pro-cess!

 

[lumbergh]

Some of the most invalid correlations are done at ‘PHD level’, often using data sets that are too small to matter but just big enough for whatever paper is being written. That’s why we spend so much time thinking ‘random food item’ is bad for us and years later someone proves the exact opposite with meaningful data. 10 years of draft data is nowhere near enough to be statistically significant as it doesn’t even cover a generation (it should be multiple). However given the variables change and evolve so much over time, I just don’t see it as a workable model even with 50 years of data.

And some of the most invalid posts are made at the “HFBoards level”. I mean, you’re just going to ignore the numbers out in front of you and malign a bunch of people who are trained to analyze, right?

Let me make a sports analogy here. This is like saying, “some of the worst errors I’ve ever seen are made by professional baseball players.” That’s of course ignoring the millions of awful plays on softball diamonds around the country in beer leagues all day, every day. And you don’t sit around and watch any of those beer league games.

I’m sure you’re a smart person, and you’re free to disagree with the analysis in the opening post. You’re smart enough to fling around “statistically significant” even when you know it wouldn’t apply here. Just be glad someone did the analysis so we have something to talk about and take it with the grain of salt it deserves.

 

[YP44]

I like both. If you have a guy in your cross hairs that you want go up and get him. If you have a guy you are targeting that you think will be available later trade down, get your guy and get another asset.

 

[Fishhead]

This kind of analysis would be much more useful if all teams developed draft picks equally, but alas, they do not. It is very difficult to compare this many different entities and it requires a ton of assumptions to do so. For example, a defenseman or goalie picked in the middle rounds by the Kings is more likely to be a valuable pick than a forward based on their development systems strength.

Assigning numerical values to picks would work really well for the first two rounds, but like the draft itself would be a crapshoot after that. The closer you get to the top of the draft the stronger the consensus is for obvious reasons, but outside of the top 50 picks or so it falls off rapidly. And this is not even taking into account trying to assign a numerical value to a certain pick. For example, what kind of value would be attributed to a player like Dwight King? He was an average player, but decent value for a 4th rounder. His numbers don't jump off the page, but his value is much higher than his production because he had two excellent playoff series at key moments - Phoenix in '12 and Chicago in '14. Or what about Trevor Lewis? His traditional metrics aren't good at all for his draft position, yet we all know how big of a cog he was on those cup teams. His penalty killing was key to many a playoff series, and his backchecking in game 5 against Vancouver help set the stage for that run. None of that is taken into account and it's a huge component of value.

Regardless of all of that, the player sample used in that paper is from the late 80s to the late '90s, when the internet was still in its infancy and draft prep would be considered stone age compared to today. Video was shared by sending VHS tapes around rather than clicking a mouse, and drafting abilities of different teams varied wildly depending on the money they could spend towards that. Today, everyone essentially has the same information at their disposal. Given the wide disparity of cash flow among teams back then, it pretty much renders the data useless for comparison outside of the first 40-50 players or so that everyone knew.

 

[Brodeur]

Draft pick value chart works a bit better for football (which has a deeper pool of players) and even then some execs have said that the early 90's value charts lost some currency in recent years.

The tough thing with draft value charts is the assumption that every draft is the same. I have no qualms with the Kings moving up to grab Pinelli who could have gone around pick #25 and nobody would have thought it was unusual. It really does depend on the names involved.

The late 90's / early 00's Dallas Stars usually had a late first round pick and they liked to trade down for multiple picks. Using the draft pick value chart in the original post:

1998: Traded #27 for #39 and #57 - 291 points lost vs. 384 points gained (209 + 175)
1999: Traded #28 for #32 and #96 - 283 points lost vs. 360 points gained (246 + 114)
2002: Traded #13 for #26, #42, and #185 - 481 points lost vs. 565 points gained (297 + 203 + 65)
2003: Traded #28 for #36 and #54 - 283 points lost vs. 399 points gained (217 + 182)
2004: Traded #20 and #153 for #28, #52, #88, #91 -429 points lost (350 + 79) vs. 732 points gained (283 + 184 + 137 + 128)

So by the draft pick value chart, Dallas "won" all of those trades. Let's now add in the names.

1998: #27 (Scott Gomez) for #39 (John Erskine) and #57 (Tyler Bouck)
1999: #28 (Kristian Kudroc) for #32 (Michael Ryan) and #96 (Mathias Tjarnqvist)
2002: #13 (Alexander Semin) for #26 (Martin Vagner), #42 (Marius Holtet), and #185 (Francis Walthier)
2003: #28 (Corey Perry) for #36 (Vojtech Polak) and #54 (B.J. Crombeen)
2004: #20 (Travis Zajac) and #153 (Steven Zalewski) for #28 (Marc Fistric), #52 (Raymond Sawada), #88 (Clayton Barthel), and a future 3rd (Richard Clune)

Doesn't look like Dallas "won" once you consider the names, granted they obviously wouldn't necessarily have taken Gomez/Semin/Perry/Zajac had they kept the pick. Craig Button told a story about how they begged Bob Gainey to trade up for Jiri Fischer in 1998 but it didn't work out.

 

[Brodeur]

Another fun one to look back on is the Ducks at the 2008 Draft. They had depleted their farm to win the Cup, so they were looking to restock the cupboard. Meanwhile their scouting staff apparently really liked Erik Karlsson. So they figured they could trade back from #12 to #17 and still get him. Unfortunately for them, Ottawa traded up to #15 to take Karlsson.

#12 (507 points) vs. #17 (395), #35 (222), #39 (209). By the draft pick value chart, Anaheim got the same "value" as a mid-first round pick for moving back 5 spots. But instead of Erik Karlsson, they got Jake Gardiner, Nic Deschamps, and Eric O'Dell (who they didn't bother to sign).

At least with higher picks, teams typically trade back with the mentality that they can still land their intended target rather than wanting to add extra picks. Like in 2007, St. Louis would have taken Lars Eller with the 9th pick but figured they could safely trade back to #13 and still get him. So it was worth adding a couple extra picks. Looking at that one:

#9 (596) vs. #13 (481), #44 (196), and a future 3rd rounder, that looks like a win for St. Louis on paper. But they ended up with Aaron Palushaj and Ian Schultz with the extra picks. Meanwhile San Jose traded up to #9 because they had intel that Florida was going to take Logan Couture at #10.

At the same draft, St. Louis offered #9 and #24 to Edmonton for #6. According to the value chart, St. Louis was crazy to make that offer. They badly wanted Jakub Voracek who was probably more worthwhile than Eller and Ian Cole who they ended up with instead.