After an interminable wait, the 2020 NHL draft has come and gone, and is now several weeks behind us. Players know which teams they’re headed to, leaving us to evaluate each team’s performance. To wrap up a long year of draft coverage at Silver Seven, this post (plus another one tomorrow) will take a by-the-numbers look at the draft to answer a few key questions: which teams performed the best? Who didn’t do as well? Which players were the biggest steals, and which were the biggest reaches?
Part one will be divided into four sections:
- An explanation of my methodology and recent tweaks
- The biggest steals and the biggest reaches
- Some undrafted players to watch
- And an assessment of the accuracy of various lists
To tackle these questions I’ll be basing this post on the consolidated rankings, a grouping of 50 reliable public sources, which aims to synthesize an opinion of which players are most highly valued at the day of the draft. You can read the full methodology of how that’s accomplished here, although the two terms you’ll need to know for this post are expected pick value and expected player value.
Expected pick value is taken from research done by Michael Schuckers, which assigns an expected value to each draft pick based on how likely a player drafted in that slot is to become a useful NHLer. Naturally, higher picks are worth significantly more than late round picks, with a steep drop off at the beginning of the first round and a plateau around the middle of the second round.
A note for those who’ve been following all year: the analysis in this post switches from Schuckers’ 2011 values to his 2016 values, which means player values are slightly different than what’s shown in the public consolidated rankings. They’re more up to date, and also follow a much smoother curve than 2011 which has a few quirks (draft value slightly rises at the beginning of the third round, for example). The overall trend is still very similar.
Expected player value is where the consolidated rankings come in, averaging each player’s placement from the group of sources. Because the average is done on the same scale as Schuckers’ pick value chart, it makes the two values comparable. Marco Rossi, for example, had an expected player value of 642, typically corresponding to a pick range between 4th and 7th overall. But he was taken at 9th overall by Minnesota, which has an expected pick value of just 456. The player value is greater than the pick value by a pretty significant margin, meaning the pick is classified as a steal in the eyes of public draft analysts.
The comparison isn’t quite be as cut-and-dry as that, however. There are still a couple of outstanding issues to address:
- How do we account for players that aren’t ranked by all 50 sources?
- How do we deal with drafted players who weren’t ranked by any of the 50 sources?
- How do we account for different variations in public rankings?
The first issue is the most pressing issue in my opinion, as it represents the biggest deficiency in using an average rather than something like a Kemeny-Young algorithm. Only a small handful of sources choose to rank 217 players (for good reason) and some lists can be as short as only 31 players. It leads to someone like Jaydon Dureau looking like a steal, for example, when his average value of 81 is far greater than his pick value 46 (he was drafted 146th by Tampa Bay). The problem is that Dureau was ranked by only one of the 50 sources, who placed him 73rd on their list, while everyone else left him unranked. Intuition would tell us that still isn’t a very good value pick; even though he went higher than the one source who ranked him might have predicted, there’s 49 other sources who left him completely unranked!
The solution I’ve chosen isn’t the most elegant, but it ends up altering both the player and pick values. The player value is changed to reflect how often the player is ranked, while the pick value is adjusted based on how often it was assigned to a player in the consolidated rankings.
Keeping with Dureau, his original value of 81 turns into 81 * (1 source / 50) = 1.62. But the 146th pick was only included in 11 of the 50 lists, so it wouldn’t be fair to compare the new player value to the original pick value of 46, since other lists may have included him had they been extended. So the pick value becomes 46 * (11 / 50) = 10.12. The new difference is nowhere near enough to classify him as one of the draft’s biggest reaches, and instead understandably represents a value loss (-8.50) in the eyes of the public consensus.
To answer the last question of accounting for variation, while this is much better captured by looking at each player’s expected pick range (one standard deviation from their average draft value), this can’t be applied to players ranked by one or zero sources who have a standard deviation of zero. Thus, it’s unfortunately omitted from this analysis.
As a last bit of housekeeping for the rest of the post, you may notice that some picks are listed as one draft slot earlier than officially listed by the NHL, due to the Coyotes’ 2nd round pick being revoked. For example, Yegor Sokolov was selected with the 61st pick in the draft, but because he was the 60th player selected he’ll just be referred to as pick #60.
Now, let’s get to the results!
Biggest Steals & Reaches
The first task is simple enough: take the difference between the adjusted player and pick values. Here an interactive chart for this too if you want to look at every pick. Here are the top 20:
In a draft filled with high-end talent, the biggest of the big steals happened right at the top when Marco Rossi went 9th overall to the Minnesota Wild. Cole Perfetti falling to Winnipeg at the subsequent pick takes third place, with Anton Lundell at 12th not too far behind.
It’s also worth having a look at Carolina, who my model considers to have the second biggest steal in Noel Gunler, and two others in the top fifteen with Zion Nybeck and Alexander Pashin in the later rounds. Toronto ties them with three placements in the top 20.
Let’s move to the bottom 20:
18 of the top 20 biggest reaches occurred in the first two rounds, where there’s much more room for mistakes to be made. To nobody’s surprise, the draft’s biggest surprise pick of Yegor Chinakhov takes the top spot with Columbus pouncing on him at 21st overall. But the next two aren’t far behind, with New Jersey’s preceding pick of Shakir Mukhamadullin also being a head-scratcher to the broader scouting community. Ottawa’s reach on Jake Sanderson in the top five rounds out the top three.
The Senators were the kings of reaching in this draft according to my model, making up a full quarter of the top 20 value losses in the draft. Montreal and Chicago are the only other teams with multiple appearances.
Another curiosity is the positional breakdown of each list — notice how the steals are mostly made up of forwards, while the reaches skew more heavily towards defencemen. Let’s compare the average value gained/lost per pick, split by position:
Forwards: +3.37 (119 picks)
Defence: -13.35 (55 picks)
Goalies: -13.74 (16 picks)
I’m of two minds on how to decipher this, but I think there’s a middle ground between NHL teams overvaluing defencemen and goalies and the public undervaluing them. I lean more towards the opinion of NHL teams being the side overvaluing players, since they’re affected more by their perceived scarcity. I suspect that number’s also inflated on the negative side for goalies who were completely omitted from three of the 50 lists, affecting them more when the values become adjusted. Nonetheless, I’d be curious to see a larger scale results-based analysis to see if NHL teams really tend to overvalue goalies and defencemen more than forwards.
Undrafted Players to Watch
This analysis treats the NHL draft as a negative-sum event, meaning there was still unclaimed value even after the 217th pick was made by Tampa Bay. Some undrafted players will eventually go on to have an impact as either re-entries or free agents, and while it may be a bit early to be looking at overager candidates for 2021, I think it’s worth acknowledging the highest-value players who were still left on the board.
1: Théo Rochette (C) — xRange of 67-122, 28/50 sources — comparable value to Cross Hanas (DET), Oskar Magnusson (WSH), Maxim Groshev (TBL)
“Rochette was considered a potential first round prospect before the season, but his play throughout the year showed otherwise. After a disappointing Hlinka Gretzky Cup, the center had a very slow start to the season, in part due to his mononucleosis diagnosis. Rochette never really picked up the pace throughout the season and his progression stagnated.”
2: Oliver Suni (RW) — xRange of 73-116, 21/50 sources — comparable value to Rory Kerins (CGY), Carson Bantle (ARI), Blake Biondi (MTL)
“Elite third line players today need to have three main qualities: speed, size, and IQ. Oliver Suni has all three. His skating improved considerably over the course of his first OHL season, showing a consistent ability to beat defenders wide or to retrievals. He also is big and strong enough to win most battles and is especially good at gaining inside leverage on opposing players.”
3: Ruben Rafkin (RD) — xRange of 73-125, 20/50 sources — comparable value to Samuel Knazko (CBJ), Jacob Truscott (VAN), Thimo Nickl (ANA)
“Rafkin is a poised three-zone defenceman who is comfortable with the puck on his stick, strong on his feet, physical in man-on-man battles, heavy for his size and calculated in his approach. He won’t wow you but he plays a modern, puck-possession game that should translate well at the NHL level. Though his game lacks a truly dynamic quality, he’s more well-rounded than many of his peers. He could have late-round value long term.”
4: Juuso Maenpaa (C) — xRange of 67-147, 18/50 sources — comparable value to Carson Bantle (ARI), Yegor Sokolov (OTT), Bogdan Trineyev (WSH)
“The Rookie of the Year winner in the Finnish U20 league, Mäenpää is one of the league’s speedier players who can fly all over the ice. He’s electric in transition, seems to be all over the place in the offensive zone, and as a natural centre did a consistent job involving himself in his team’s backcheck. He’s seemingly fearless with the puck on his stick, with solid puck skills to make him versatile as a playmaker from the boards or even a one-timer option in the slot.”
5: Ethan Cardwell (C) — xRange of 76-137, 20/50 sources — comparable value to Carson Bantle (ARI), Yegor Sokolov (OTT), Bogdan Trineyev (WSH)
“Whatever the challenge — Cardwell was up to the task. He really helped solidify [Barrie’s] top-six forwards group after joining them in a mid-season trade from the Saginaw Spirit. Sometimes that meant taking a beating, but the 5-foot-10, 157-pound Cardwell never shied away from a challenge. Whether it was playing the net front on the power play or plying his trade on the penalty kill; he ably, and dutifully performed his task to the utmost of his ability.”
Others: James Hardie, Dmitri Rashevsky, Jacob Dion, Pavel Tyutnev, Nick Malik, Simon Kubicek, Charlie DesRoches, Ivan Didkovsky, Hugo Styf, Pavel Gogolev, Lleyton Moore, Samuel Hlavaj, Christoffer Sedoff, Brady Burns, Simon Knak
A few themes appear: players who suffered from mid-season injuries, players who generated hype based on a couple small hot streaks, and players who are vertically challenged. Maybe we’ll see some of these names re-appearing for 2021.
While predicting the outcome of the draft isn’t necessarily the goal of the lists that are included in the consolidated rankings, coming close to what happens can sometimes be a neat little bonus. I’ll be using the exact same method as last year to determine which lists came closest to what happened on draft day, so head over there if you’re curious about the details. In short my calculation of list accuracy combines two metrics: % of players drafted within the range of the rankings, and the average value difference (switched to Schuckers’ 2011 values). Maximum score of 10 for leading both categories.
Here were the top performers:
2020 NHL Draft List Accuracy
|Rank||Source||Players Ranked||Drafted Accuracy||Positional Accuracy||Score|
|Rank||Source||Players Ranked||Drafted Accuracy||Positional Accuracy||Score|
|9||Draft Prospects Hockey||217||71.43%||36.30||7.0|
|10||The Prospect Network||100||77.00%||48.68||6.9|
In a shocking surprise, Bob McKenzie takes the top spot for the second year in a row. This is actually totally expected — McKenzie’s list differs from the rest since it’s a consolidation of opinions from NHL scouts, rather than public opinion. Recrutes also makes sense for second place, as its draft guide is mostly comprised of quotes from unnamed NHL sources.
A few spots down is the consolidated rankings, which uses the arbitrary list of players ranked by at least 50% of the sources. It falls a couple spots after placing second in 2019, although it’s still proven to be one of the more reliable sources for draft accuracy given that it’s made up by a large group of voices.
A couple other sources repeat in the top ten for a second straight year: Chris Peters (8th in 2019) and Derek Neumeier (4th in 2019). The full list can be found here.
Looking a bit closer at the consolidated rankings, the main feature of using the value-based system is forming an expected range for each player, which you can find the details on here. In this sense, the rankings performed even better this year with 94 players falling within their expected range (44%) compared to only 60 players in 2019 (28%). There was improvement pretty much all across the board — the first round hits went up from 52% to 61%, while rounds 4-7 were way less chaotic in 2020 moving from only 11% in 2019 to 39% this year. The wide ranges with few sources makes me suspect that there’s a lot of room for variation with these metrics, and on its own these accuracies should be taken with a grain of salt. But even in the earlier rounds, it seems the uncertainty captured by expected range did a pretty decent job this year.
Come back tomorrow for part two: a power ranking of all 31 teams’ performances at the draft!