Login

icewindoo · 02-02-2025, 12:20 AM

A little while ago @LALLAREN_1 some fucking asshole posted an analysis of how GMs value their picks for the SHL Entry Draft. It was a look deep into the minds of the GMs, telling us what GMs actually think of the relative value of the different pick positions in the draft. Obviously, a higher draft pick is "worth" more than a lower draft pick, and it was great to see some numbers teasing out how that valuation breaks down across the length of the draft.

That got to me thinking: how good are GMs at recognizing the value of prospects? It's one thing to do whatever it takes to get a high draft pick. It's another thing entirely to use that high pick to select a player who will actually provide the maximum value throughout their SHL career. Do GMs selecting higher in the draft actually end up picking better players? Is the higher value that they are assigning to higher picks actually justified? I decided to gather some numbers to investigate the relationship between the pick position and the value of the player chosen.

Methodology

First we need to consider how to measure the value of the selected players. Obviously you can't just tally up and compare total points achieved by each player, since there are many kinds of players, and not all of them are offensively-minded. Not surprisingly for anyone aware of my media, I've chosen to look at each player's "Player Contribution" (PC) as a judge of how much value that player provided in their career. I think PC is a decent way to compare players, including different types of players.

Why look at PC and not something like TPE? That's just me, I guess, and I have the data already. I also prefer to look at in-game impact, not just some theoretical number like TPE; not all TPE values are the same.

There are a bunch of limitations in my data gathering. The biggest issue is that my PC calculations rely on FHM data that is only available from S66 onwards; so this means that I'm only looking at player performances from S66 thru S80 (15 seasons). Since the length of players' careers can vary so much (and I wanted to limit the effect of that variance), I decided to limit consideration of each player's career to their "first seven" SHL seasons. I put "first seven" in quotes because that's not quite correct: it's actually the seven seasons from Draft+3 through Draft+9, in order to avoid including any seasons of a player that got called up early. I've also excluded goalies entirely from this analysis, because, well, they're goalies.

So for all players who were drafted between S57 (that is, starting with those whose D+9 season was S66) and S77 (that is, ending with those whose D+3 season was S80), I added up the PC values from the "early" seasons of their SHL career, and then calculated the average early career PC across all players who were drafted at each pick position.

Results and Discussion

The graph below shows the raw results of how the early career PC values compare against graph position.

[Image: wqJqb6V.png]

For example, the graph shows that for the 21 skaters who were drafted 5th overall from S57 through S77, they averaged a total early career PC of 162. Two points of reference to help understand the relative size of these numbers: the highest total early career PC that I measured was 436 for Lias Ekholm-Gunnarsson (S66 thru S72); and S79's MVP Chris Valentine amassed a total early career PC of 349 (S71 thru S77).

Overall, the graph shows roughly what you would hope to see: players picked earlier in the draft generally have higher total contribution ("value") than players picked later in the draft, though having a top five pick is certainly no guarantee of picking a superstar. The one thing that jumps out though are those spikes on the right side of the graph. This means that some high-value players were picked very late in the draft. However, each of those large spikes is calculated from just two players; only two of the drafts went longer than 66 picks.

The graph below shows the number of picks actually made at each position in the draft.

[Image: V7BPzVF.png]

The dips in number of picks for some positions (e.g., 10 to 12) represent goalies who were drafted at those positions and excluded from this analysis. If we eliminate the outsized influence of the draft positions with very few picks (i.e., those fewer than 10), then the results of total early career PC versus draft position can be seen here:

[Image: DaKbKLY.png]

That looks a lot "nicer", and I've added an exponential trendline to show the rough trend. For example, players picked late in the first round (#18) contribute about twice as much over their career as players picked in the middle of the second round (#33).

Since I know that most people do (reasonably) put a lot of stock in TPE, here's what the average peak TPE looks like across the draft positions:

[Image: io2ye28.png]

The TPE graph looks fairly similar, though it's a bit "flatter" than the PC graph. This suggests to me that in later rounds players with decent peak TPE can still be found, but perhaps their actual in-game contribution doesn't quite match up with the TPE. It also tells me that my own player's expected peak value (12th overall draft) is about 1200; I gonna do my best to beat that.

It should be noted that these are very small sample sizes: 21 data points for each draft position is a very small number. For example, it's hard to tell if that obvious dip within the first three picks is meaningful, or if it's just a function of having few data points.

But overall I'd say that GMs do a pretty good job of identifying future talent at draft time.

LALLAREN_1 · 02-02-2025, 01:36 PM

This is great, I’m honored I managed to inspire someone to go into the thick of stats and data like I did. Pay is kind of lackluster for these kinds of articles comparing time spent to $$$ return so anyone who undertakes these types of pieces is doing it for the love of data which I love to see and fully support!

I really enjoy the use of dual charts, making sure we see both TPE and PC. I’d personally lean more toward TPE as it is independent of any team performance, roles, and game engine dice rolls. It’s purely a numerical representation of your theoretical max value. Of course, lower TPE players can be worth more to teams. Teams are greater than their combined TPE alone.

Great and Fun read!

Popol · 02-02-2025, 06:09 PM

Great read! a good tool for pick trading

***bluesfan55*** · 02-02-2025, 08:15 PM

babe wake up another icewindoo media dropped

icewindoo · 02-02-2025, 09:03 PM

02-02-2025, 01:36 PMLALLAREN_1 Wrote: Pay is kind of lackluster for these kinds of articles comparing time spent to $$$ return so anyone who undertakes these types of pieces is doing it for the love of data which I love to see and fully support!

Lol, ain't that the truth. It was a fair bit of work to pull that all together, and just before posting I finally checked the word count: "A thousand words, that's it? Dang."

But thanks for the kind words!

By-Tor · 02-02-2025, 09:03 PM

Love this. Great analysis!

Credit to enigmatic, Merica, tweedledunn, and jaypc8237 for sigs

***UptownCord*** · 02-02-2025, 10:04 PM

skaters only; sad!

LALLAREN_1 · 02-03-2025, 10:47 AM

02-02-2025, 09:03 PMicewindoo Wrote: Lol, ain't that the truth. It was a fair bit of work to pull that all together, and just before posting I finally checked the word count: "A thousand words, that's it? Dang."

But thanks for the kind words!

Haha, yeah, I know that feeling. When I was done with the first draft of the trade value chart, I was staring at it like, "Only 1k words?! ". So that's why there is a history section and a detailed step-by-step process in there to boost word count while still remaining relevant to the article itself. As it stands, easy to write, quantity-based media is the way to go, unfortunately. I will still probably do some data analysis at times but not for the money, that's for sure. When I need money history articles it is.

***Anthique*** · 02-05-2025, 01:34 PM

GMs are doing good then? What a surprise haha!

Great article, love it!

Navigation

Extra Menu

About us