The Value of a Draft Pick

Analyzing the value of every NHL draft pick since 1992.

6m30s. January, 2018. Data correct as of October, 2017.

The NHL Entry Draft takes place every June, when 31 general managers get together and take turns selecting the rights to young hockey players. Some of them work out, and some of them don’t. But every year, also in June, somebody always brings u the question of how we value draft picks. When Elliotte Friedman, a journalist for Rogers Sportsnet and Hockey Night in Canada, brought it up this past summer, it got me thinking. How do we put a value on a draft pick? Friedman’s article linked to another report, this one by Stephen Burtch, containing a few graphs and insights that confirmed it: first round picks generally (read: almost always) produce better results. He also pointed out that teams exchange draft picks with “alarming frequency”, something I had never really thought about before. Which of these trades worked out? Which didn’t? How can we tell?

While the data to support these insights is available in abundance, it generally comes in tables full of statistics, or spreadsheets packed with numbers and percentages. This might get my heart rate up a few notches, I can imagine it has the opposite effect for many. So I set about turning it into something visual, a tool we can have fun with and maybe gain some insights from. I combined data from with the magic of Python and R Studio to analyze every draft pick since 1992 and compare them across the board.

While some metrics lend themselves quite well to this sort of visual, others really don’t. Dom Luszczyszyn’s Game Score, for instance, lets us compare players across all positions quickly and simple. Other stats, like goals or assists, aren’t so simple (i.e. a defenceman is likely going to have score fewer goals than a left winger, so he’ll look like a less valuable player). I’ve converted these into something I call stat share, meaning a player’s share of each stat (e.g. goals or assists) relative to his position, and that’s what’s presented here.

Feel free to jump right in, or continue to scroll and I’ll tell you all about it.

Analyzing NHL Draft Picks

Below are all 6265 players drafted into the NHL since 1992, when the league expanded to 24 teams and the draft moved from 12 rounds to 11. Select a VIEW to see players sorted by year and round, by team rank, or by value. Each circle represents a player, sized by the selected metric. Players who have never played a game in the NHL are represented by dots at 25% opacity. Hover or tap on a player to see their stats, both over the entire career and through the first five years after their draft year. Click or tap on a ROUND to expand.


Size by
Based on

The Metrics

To really see how valuable a player is, we can’t just look at the usual stats. Goals, assists, points, PIMs — it’s all there, but to really measure how well a player performed, we need more. Just because a centreman scored twice last night, doesn’t mean he had a better game than his defenceman that blocked eight shots and was on the ice for twelve of his team’s shots. It might, but we can’t reasonably compare those guys because they play different games. It’s like apples and oranges. To see what I mean, use the dropdown above to change the statistic we’re sizing by from Average Game Score to something else.

This is where Dom Luszczyszyn’s Game Score comes in. Based on what was originally a baseball metric and then a basketball stat, Game Score looks at nine different stats for skaters and two for goalies. You can read more about it here. This is what we’re looking at here, where each circle represents a player, sized by their average game score over their entire career.

That means we can compare guys like Alex Ovechkin (the highest avg. game score, at 1.46) to Sergei Gonchar (a defenceman, at .69) to Roberto Luongo (an often overlooked goalie, at .92), getting a quick comparison in their value, even though they are all very different kinds of players.

First Five

When we’re analyzing a draft pick’s value, we also have to consider how quickly he can break into the league. Some will tell you that if a player’s not in the chel within a few years, he wasn’t worth it. Or maybe it’s a question of how many games they need to play, a threshold Jamie Fitzpatrick sets at 200.

Either way, the amount of time it takes a draft pick to be a legitimate force in the NHL should have an effect on his value. Some guys start strong and slump after a few years, like Patrick Lalime, who went from an AGS of almost .8 in his first five years to .48 over his entire career; others just keep getting better, like Alex Ovechkin, going from an AGS of under .8 to almost 1.5.

Success Rate

How many draft picks actually crack the big league? Less than half. Between 1992 and 2017, there were 6265 players drafted. Of that, 2484 have played an NHL game. That’s 39%. That means there’s 3781 players that just never made it, including 2001’s first round pick, Igor Knyazev. Nice one Carolina. The Boston Bruins, on the other hand, are doing pretty well when it comes to drafting. 48% of their draft picks have made it to the NHL. That’s in stark contrast to the Red Wings’ 31% success rate. Of course, the jury’s still out on the most recent picks, so we’ll see.

The (Un)Expected

First round picks, particularly the top five, are going to work out to be some of the most valuable players in the game, right? Yeah, pretty much. Sidney Crosby, Alex Ovechkin, and Conor McDavid are proof. Each was easily the best in their draft year and they went first overall, as you’d expect.

Take a look at this visualization on a larger screen to look into the deeper rounds, where players like Pavel Datsyuk stand out. He’s an incredible player, but many seem to forget that he was picked near the end of the sixth round, 171st overall. His circle is not only the largest in the round, but it’s the largest for the whole year. Detroit saw something nobody else did and it paid off. In fact, it’s comparable to most first rounders, no question. His career AGS sits at 1.1, and McDavid is around 1.2.

But what happened in 1996? Pavel Datsyuk happened. It’s pretty well known that Datsyuk is an incredible player, but many seem to forget that he was picked near the end of the sixth round, 171st overall. His circle is not only the largest in the round, but it’s the largest for the whole year. Detroit saw something nobody else did and it paid off. In fact, it’s comparable to most first rounders, no question. His career AGS sits at 1.1, and McDavid is around 1.2.


Play around with the controls above to change the metric, position, or player shown, and see what you find!

Here we see every pick laid out by the team that chose them, sized by their share of the points scored by all players at their position. Clearly Vegas has some work to do. Either that, or this is their first year in the league.

The longer the “tail”, the more picks the team has made, and the longer the “fat part of the tail”, the more draft value the team has come away with.

Value versus Expectation

When we compare pick number to the selected stat, in this case average game score, it looks like our successful picks are more abundant to the left, or the earlier rounds. This is true, but it also looks like there are more early picks than there are later, which is not the case. In reality, the further to the right, the more dots are placed along the bottom axis, meaning a huge chunk of later picks never played in the NHL.

This view also allows us to chart expectations, in this case, specifically for right wingers. For instance, we can expect a right winger drafted early in the first round (top five), like Patrick Kane, to score 70 goals in the five years after he was drafted. He scored 126. The line in blue shows the expected values, or averages, for each position in the draft. As expected, it gets closer to 0 the deeper we get.

Play around with the controls above to change the metric, view, time period, position, or player shown.

A note on the data:

Data is current as of October, 2017, meaning that some stats for the current season are included, but not for everybody. There are more than 6000 players included, so it’s more than possible that I missed something. A cleaner, more complete, more efficient dataset is on its way, but in the meantime, let me know of any problems or missing bits.

Data comes from individual player pages at I compared basic and advanced stats within positions, meaning that players are compared to similar types of players (i.e. goalies to goalies, defence to defence) to give every player a share of each statistic, as well as a rank amongst every draft pick their team has made since 1992.