Draft Theory: On Risk and Reward

Garret Hohl
May 12 2015 12:30PM

As the 2015 NHL Entry Draft quickly approaches us, we look more into draft theory and discuss processes that allow one to maximize their success.

Most draft services open to the public will discuss two factors in a prospect: risk and reward.

The higher a player's ceiling, the more potential reward a prospect has. While sometimes this varies, for the most part public third party scouting services agree on upside with each other, the numbers, and how players historically turn out.

There does seem to be an inefficiency though, and that is in determining risk. The more one immerses themselves in draft analytics the more one discovers that their is a flaw in risk assessment.

Here are two players familiar to Winnipeg Jets fans to represent this difference.

In 2013, Kevin Cheveldayoff and the Winnipeg Jets drafted Joshua Morrissey 13th overall. The left-hand shot defender had been ranked by most public draft services in the mid-first round. The scouting notes described Morrissey as a dynamic skater, who struggles defensively and his size will be a hurdle he would have to try and overcome. Morrissey's size was listed between 5'11 and 6'0 depending on the location.

Morrissey was a strong scorer, who had already posted 85 points in 138 games in the WHL prior to being drafted.

Four months later the Jets picked up Keaton Ellerby on waivers from the Los Angeles Kings.

For those that do not know, Ellerby was once considered a high end draft prospect. Final draft rankings had Ellerby as one the top defensemen in the draft. CS had him #1 for NA defensemen. ISS had him ranked #2 for defensemen, as did The Hockey News. McKeen's final ranking had the worst ranking, placing Ellerby as the 3rd best defensman of the 2007 draft. In the end Ellerby was drafted 10th overall.

The main reason was size. Ellerby was a 6'4 defender who could skate very well for his size. His point production was not terrible by any measure, but 33 points in 137 WHL games prior to the draft places him at a clear level below Morrissey.

Now Ellerby is not a failed draft pick. He has already played 212 games in the NHL, which is just over the 200 NHL game threshold many use to evaluate those that did succeed in making the NHL.

Still, Ellerby and Morrissey make good examples of where the flaw is with evaluating risk of draft picks.

Going into the draft, many thought that Ellerby had an impressive ceiling, but many were unsure if he would ever reach it. He was well thought of however as a low-risk draft selection due to his size.

This is the common thought process. Bigger players can play lower depth roles if they do not reach top-4 defender or top-6/9 forward potential. Smaller players though do not have that safety. This means there are more potential spots for a larger player, which means less risk.

This is partly true. The issue is that it is only partly true.

Now obviously the situation varies individually from person to person, but there is a trend that the better overall player is the safer player.

Using the newly developed The Projection Project application, we can look at drafted players who scored similarly to Morrissey and Ellerby and were within two inches in height. I also filtered for draft picks in the first three rounds.

Screen shot 2015-05-12 at 10.06.09 AM

As you can see, the numbers seem to fit the idea that a player with Keaton Ellerby's size and scoring can fill a depth role better than a player with similar size and numbers to Joshua Morrissey. However, this does not make Ellerby-types the safer pick.

Expanding outside of exclusively the first three rounds and we see a stark difference in success rates.

Joshua Morrissey Draft Cohorts

Screen shot 2015-05-12 at 9.50.58 AM

Keaton Ellerby Draft Cohorts

Screen shot 2015-05-12 at 9.51.38 AM

Charts are courtesy of The Projection Project.

This is not news to the analytical community. Rhys Jessop, now of Canucks Army fame, long ago showed that higher scorers succeed in making the NHL more often even for defensemen.

Now take caution in making the wrong assumption. This does not mean that a higher scoring small player is automatically safer pick than a larger player who does not score as effectively.

Size matters in hockey. There is a positive relationship in successfully making the NHL and height (although not really much for weight). If it were not for Ellerby's 6'4 frame, he probably would not have been effective enough to have the career he had.

It's a more complex reality than just simply more scoring equates to a better prospect.

Many have seen 2015 draft eligible Oliver Kylington falter through the draft ranks, with scouts questioning if Kylington is a boom or bust draft pick. Some are even placing Thomas Chabot (who is a really good prospect) as the less risky player due to his size and ability to play a depth role if he doesn't hit his ceiling.

Still, Josh Weissbock and Money Puck's draft cohort system has found over 50 percent success rates for players like Kylington, while Chabot-type players sit around 21 percent success rate.

It's not simply due to scoring levels. The short in stature Mitchel Vande Sompel cohorts only succeed at about a 20 percent rate, even though Vande Sompel has nearly doubled Chabot's scoring.

Traditionally determining ceiling and upside has been a highly effective process. The draft market has been relatively efficient in that area. However, it still seems that without statistics it becomes much more difficult to get an accurate handle on how much risk a player presents.

A96da2dbde4a6e338d7c1a0c0287249d
I like hockey. I like numbers. I like the Jets. I have written for the Winnipeg Free Press, Jets Nations, Hockey-Graphs, Arctic Ice Hockey, and Hockey Prospectus. Currently I am managing Jets Nation and you can catch me tweeting @GarretHohl.
Avatar
#1 Anthony
May 12 2015, 12:55PM
Trash it!
0
trashes
Props
1
props

How do you measure effective non point producing defenders? Example Ryan McDonagh produces offence, but is equally effective away from the puck. Is that measured in this analysis?

Thanks, very interesting work...

Avatar
#2 t_bison
May 12 2015, 01:23PM
Trash it!
0
trashes
Props
1
props

@Anthony

I wonder if as we move along if possession will give a much better view of NHL prospect potential. The only issue is that technology has not filtered down that far. It might be worthwhile to examine a 2PS% WOWY? I'm not sure.

Avatar
#3 Anthony
May 12 2015, 01:39PM
Trash it!
0
trashes
Props
1
props

I think this is amazing info, really makes you think, only fear that I have is that things that cannot be captured or measured are given zero value. I could be completely wrong thou, hence the request for followup.

Avatar
#5 Anthony
May 12 2015, 08:27PM
Trash it!
0
trashes
Props
3
props

@Garrett

I appreciate that, just trying to find out where they're may be inequities and ways to qualify them.

General question, what value to do put in measurables vs. non measurables? I think Ive read Dean Lombardi put it at a 1/3 ratio, measureables vs other aspects of team building (probably why they target bigger players in drafts) but curious where you see it.

Also, why Morrissey? I thought as late birthdays are always looked down in stats arguments, they're more advanced then others? Also tough to call him a success, he's yet to play a game outside of junior? Ive liked what ive seen via the WJC, but figured you would use a more established player from the same draft as Ellerby as a true parallel. Please don't take it as a negative, I'm a fan of the work you do, every day Im learning more about the measureables of players and what makes them successful.

Avatar
#6 Ziggy
May 13 2015, 05:32AM
Trash it!
0
trashes
Props
0
props

@Anthony

You can use Subban or Weber drafted by the Canadians later in back to back rounds, both have closer heights to Morriessy then Ellerby.

Avatar
#7 Anthony
May 13 2015, 02:17PM
Trash it!
0
trashes
Props
2
props

100 percent right with Subban, stats always support that one.

Weber thou, he had 18 points in 70 games in his draft year?

If we where to use these metrics at the time, he'd be ripped for even being considered a top 30 pick. That's why I'm asking these questions, where does the stats end and the other aspects start? Hindsight isn't available at the draft table.

Avatar
#11 TrueBlue
May 16 2015, 02:08PM
Trash it!
0
trashes
Props
1
props
Garret Hohl wrote:

Hey Anthony.

Why Morrissey? Because he's a familiar player for Jets fans. Morrissey's birthdate does not change anything, as the comparison was not Morrissey vs Ellerby, but players who have statistically performed like Morrissey and Ellerby while being similar heights.

We have far more advanced ways to look at a players scoring, with adjusting for age, league, era, and other factors.

Again, we are talking about making more informed decisions, not making decisions. Making an arbitrary line of measured to non measured factors IMO is looking at it the wrong way.

I did some work previously introducing some ideas on draft theory here: http://hockey-graphs.com/2014/11/17/draft-develop-how-analytics-can-be-combined-with-qualitative-scouting/

However, I should mention that you'd be surprised how well numbers have performed in predicting successful players. Our sister site Canuck's Army has created a model that already beats out many NHL teams.

Brother site :)

Gotta say, you guys have a thoughtful comments section over here.

Comments are closed for this article.