In a day or two, I hope to be able to pull simple data: Who has had the most x round picks since 2000? Who has hit on the most x round picks (with a hit defined as at least one NHL game) since x year (back to 2000)? Stuff like that.Puck-Lurker wrote: Sun Feb 16, 2025 12:32 pmLegit looking forward to what you will cook up.FLPensFan wrote: Sat Feb 15, 2025 5:18 pmThanks! The Perri Pick Value seems like a good measure of the value before the pick. The other model I saw, created by a current and former NHL scout, was based more on value after the pick was made (based on games played). Going to look at a few more things, but right now these two are leading models I plan to use on my draft analysis spreadsheet.Victor wrote: Sat Feb 15, 2025 2:06 pmThere's this nice article about the model used by PuckPedia to valur draft picks: https://puckpedia.com/PerriPickValueFLPensFan wrote: Fri Feb 14, 2025 10:22 pmThere are a lot of different variables that can go into defining successful draft picks. There are things you cannot control, such as strength of a particular draft year. In general, the first scale would revolve simply around the number of picks taken, the rounds those picks were taken, and did said pick reach the NHL. Then I would have a second scale that would look at production against where the pick was taken, and what the expectations were. Took a top 10 pick that was expected to be a first line guy that ended up as a low end 3rd liner, maybe that's a C-F type of rating. I know we didn't draft him, but Puljujarvi to date would likely be an F rating by simply being a top 5 pick who hasn't lived up to expectations. On the flip side, Jake Guentzel was a 3rd round pick who was seen as a good player, but I believe has far exceeded output expecations. He would definitely get an A rating.100565 wrote: Thu Feb 13, 2025 2:11 pmThe issue with debating draft success is defining what a successful draft pick is. Playing 1 game in NHL?? Playing 100?? How about goalies?
Overall, in most years, mid second round picks and below have a very small probability of being a legit NHL player (defining as playing 4-6 years in the league). The Pens are certainly due for a hit but I don’t think their drafting is too much below what is expected league wide - given their draft positions and amount of picks.
If you good which NHL team is best at drafting, you'll get a few different articles back, several that were done last year. One by the Athletic ranked Pittsburgh at 15th, and over the time period they reviewed, Pittsburgh had the 2nd fewest picks (I believe Columbus had the fewest). Another article by Sportsnet last year ranked the Penguins as 5th best over the past 10 years. There are a few others out there.
This may actually be a project that I take on sometime between now and the NHL draft: How successful has each team been at drafting since the year 2000 or 2001 (which accounts for all teams except Vegas and Seattle). I found another article from about 5 years ago that assigned values to overall pick ranges (example, picks 1 through 5 may be an expected return(ER) of 1, picks 6-14 an ER of .75, etc, etc all the way to a range that covers up to round 7 (or 9 if going back to 2000)). If I do this the way I want to, then I could make it adjustable....look at only a certain round, only a certain year range, etc.
We shall see if I actually follow through.
It was developed by the former Director of Analytics for the Arizona Coyotes. It also references an article by Eric Tulsky from 2013.
From what I understand of the Perri Pick model, is that it critically prioritizes draft order from a theoretical standpoint. Meaning that the best 32 players are the first 32 players drafted, kinda. But we know how often players will drop, despite having high value at the draft.
Lets say you trade a single 1st round pick (17-32 overall), for two 2nd rounders and two 3rd rounders (using the average pick each time). I think that would be calculated as being fairly even as a trade based on Perri Pick. I'd take the four picks and run, myself. A higher chance to hit on one, or a lower chance to hit on between one and four?
I'm not a gambler though
Also not a top 2-3 coach of anything
Ultimately, what I'd like to achieve, is a way to see who got the most value out of their pick, with something like this:
---Perri Pick shows value of the draft position, irregardless of player. As you said, more about how much that pick is worth.
---There's an Expected Return model that assigns a value based on what that pick should return in terms of a player (based on games played)
---I'm looking to see there are existing models or something I can create on my own (and I'm not a stats guy, or an amateur one at best) that looks at pick value, expected return, value based on games played, and a value based on production....meld those 4 together to get one grade. So like, a 1st overall pick that plays 1000 NHL games and averages a point per game (a Sidney Crosby) is the top of the model.
Simple stuff like I mentioned hopefully available in a few days. The model related pieces will come a bit later.
