The Devils deadline moves are indicative of a team preparing for the future. That means, it’s time to do a bit of a diagnostic on how the moves are as a means to that end. There are a lot of components to decisions/transactions like these, but I’m mostly concerned with two things 1) How is the tank process going? 2) How valuable is out set of draft picks?
How Goes the Tank?
The Devils are not doing well. Exactly how not well? According to The Athletic’s Dom Luszcyszcyszcsyzcysn, who has year-end projections for every team, the Devils are projected to finish with 73 points, which would be 4th worst in the league. They are in a group of 4 teams (with Detroit, LA, and Anaheim) that are battling to be the second worst team in the NHL ahead of Ottawa who, ironically and pathetically, is the only one of these teams without their first round selection.
They have been plummeting recently due to a wave of injuries that would be comical if not for how depressing it was. If we continue to play a forward corps composed almost entirely of the roster of a last-place AHL team, we can expect to be truly competitive with even the worst of NHL rosters in the our journey to the basement of the standings. Also, there is largely consensus on this spot for the Devils. Micah McCurdy’s HockeyViz has the same bottom 5 as Dom, as does MoneyPuck.
Dom Loosechinson was kind enough to provide me with the probabilities for each NHL team finishing in each standings position.
The Devils have over a 50% chance of finishing either 3rd or 4th, and just below a 50% chance of finishing somewhere in the bottom 3. In a simple league with simple rules, this would be the end of the story. But a lot of other things are in play ... so there are a bunch of extra calculations to do to figure out the values of the potential picks we have.
We have to account for a lot of moving parts to determine the value of the Devils 2019 draft picks. We have to add in the traded picks, find the probabilities for standings positions of those teams, then choose a value structure for the draft slots to attribute to the probability distribution of all the teams’ picks we have, adjust the 1st round values for lottery probability, and finally sum up the values by team.
So let’s get started!
Value of the Devils Picks
Question 1: What Picks do the Devils have?
According to CapFriendly, the Devils have the following picks: their own 1st, 2nd, 4th, 5th, 6th, and 7th (Traded the 3rd for Pat Maroon last year); Boston’s 2nd (received in Mojo trade); the Predators’ 2nd (acquired in exchange for Boyle); the Ducks’ 3rd (awarded on condition of the Henrique-Vatanen trade); and, lastly, the Dallas 3rd (came with Carrick in the deadline deal for Lovejoy).
The two 2nd rounders are likely to be fairly low in the 2nd as the Predators and Bruins are two of the best teams in the NHL, but the Stars’ and, especially, the Ducks’ picks will be early in their round. The Devils could have 6 selections in the top 75 or so selections. Can we get a little more specific though? Of course.
Question 2: Where will the teams whose picks we have finish?
Again, thanks to Dom Lushihtzu for providing these probabilities. Now that we have all of the probabilities for each pick and standing position, we need to figure out how important each selection is.
3. How much is a selection worth?
This is an area that is really widely studied, and there have been a wide variety of techniques. Michael Schuckers built a model using GP as the target variable — how many GP did each draft spot play in their career on average. Dawson Sprigings (now with Colorado) created a model that improved upon this by using point shares as the target variable — how many standings points did a team add over the drafted players career in that spot on average. And in a separate approach entirely, Eric Tulsky (now with Carolina) used trades to determine how GMs value their picks.
There are a lot of interesting conversations to be had about the best way to design a pick model — and if you have thoughts please elaborate in the comments section — but for simplicity, we’re going to use the DTM (Sprigings) model here. It’s my personal preference because GP and trade value models are both slaves to the bias of the deployment whereas the PS is at least an attempt at true player performance.
Take note that these units are arbitrary and have meaning only in proportion to the other draft selections. If you’d like to know how they translate, using totals from the first 217 draft picks from every year with at least that many draftees — one unit is worth approximately 0.83 games played and 0.05 point shares. So the top overall pick, with 1100 of value, is a player we’d expect to play ~913 games (1100*0.83) and contribute ~55 standings points (1100*0.05) to his team over his career. As a sanity check, the median GP total of all pre-lockout 1st overall draft selections was the Capitals Rick Green who played 843 games and had 45.5 Point Shares, so this is certainly a passable conversion factor.
4. How does the lottery impact the selection values?
The lottery means that the worst team is not actually guaranteed the top pick. So it’s not as easy as just saying that Ottawa is probably finishing last so Colorado (lol) is going to have a pick of 1100 value, because they might not win the lottery. Since the lottery impacts the picks rather than the teams, I decided to make lotto-adjustments directly to DTM’s numbers. I pulled the lotto probabilities from Tankathon. Replace the team column with whomever finishes 1st, 2nd etc.
As an example, the team slated to select 1st overall (currently Colorado with Ottawa’s pick) has an 18.5% chance of selecting 1st, 16.5% chance of selecting 2nd, and 14.4% chance of selecting 3rd, and 50.6% chance of losing on all 3 lotto picks and picking 4th. So the actual value of their position will be the weighted average of each of their potential selections. The top 4 picks on DTM’s model have values of 1100, 960, 870, and 820 respectively. So the value of the #1 position, pre-lottery is 1100(0.185) + 960(0.165) + 870(0.144) + 820(0.506) = 902.1
If you do this to every lottery-susceptible pick, you get the following adjustments:
5) What are the Devils 2019 selections worth?
We now have all of the necessary data to calculate the value of each Devils pick. For each round, we do the following:
- Find the probability distribution of Devils possible positions (see: Devils End of Year Standings Probabilities bar graph in the intro)
- Find the probability distribution of possible positions of other teams draft picks (see Standings Probabilities graph in “Question 2”)
- Multiply the probabilities by the lotto-weighted values (calculated in “Question 4” from numbers in “Question 3”) for each pick
For instance, the Devils have the following probabilities of finishing 22nd through 31st — 0.001, 0.004, 0.009, 0.027, 0.064, 0.152, 0.261, 0.263, 0.189, 0.03 — and each of those gets multiplied by the lotto-weighted value of each position calculated above — for the first round, that’s 662.7, 692.9, 716.91, 736.65, 761.9, 786.7, 811.28, 841.04, 868.2, 902.1 — and summed together to get the total value of we get the expected value of the Devils 1st round pick: 823.1. Using the conversion factors from above, this is equivalent to a player that played about 680 games and was worth about 38 standings points. Here are some players with comparable career numbers.
This math mapped onto all of the Devils selections (using the original team’s standings prediction distributions) gives us the chart below.
The numbers in each bar show how many selections in that round the team has. The Devils 3 picks in the 2nd round gives them the highest total value in that round at 914 — requisite value to a high end 1st round pick. The Devils have a nice combination of top 100 picks and depth (a pick in every round).
If we look at the value of the picks acquired in trades, we see that Shero acquired picks with values of 282.4, 260.2, 236.5, and 207.8 — a total of 986.9 units. That is value higher than any pick before the lottery is run, and it is lower only than the 1st overall pick after the lotto is decided.
Now, the Devils lost a 3rd round pick in the same trade they got the Ducks one so that’s a bit misleading, but the fact that Shero’s been able to accrue value requisite to that of a top-end first rounder by trading away almost exclusively temporary pieces is a testament to his team-building.
Lastly, if we use the conversion units I suggested above, the Devils total value of 2525 units is equivalent to about 2100 man-games played and about 120 standings points. The closest comparable to that standings point total was the 1994 Montreal draft featuring Tomas Vokoun and Jose Theodore. The GP total would be nested between the 1994 Devils (Souray, Elias, Souray) and he 1988 Canucks (basically just Trevor Linden).
It’s worth pointing out that some have criticized models like this for not sufficiently separating the top of the draft (top 3 picks) from the rest of it. Since hockey is a “strong link” game — the franchise altering players at the top of the draft have value that is very difficult to accumulate later in the draft. The application of this to the draft is not thoroughly studied, but is noteworthy when finding aggregate values as in this article.
Conclusion and Your Thoughts
Shero has pieced together one of the strongest draft pick arsenals of any team in the league. Combine that with the injury-induced free-fall down the standings, and this team should expect to have at least a could pillars of the rebuild moving forward.
Pair the imminent draftees at 17-years old, with pieces like Ty Smith (18), Jesper Bratt (20), and Nico Hischier (20) and you have a really bright future planned. That is of course if Shero doesn’t attempt another Palmieri or Johansson type deal.
Which brings me to you guys. What do you think the plan should be with these picks? How do you think we should quantify value for picks, and what do you think of the disparity between the top 3 picks and the rest of the draft?
Thanks as always for reading and leave your thoughts below!