Potato Picks: Introduction

When you have an NHL team that is not very good, it is natural for its fans to turn to the youth. The questions facing a team's management have less and less to do with day-to-day operations of the club and more to do with setting it up for the future. As a team toils amidst the bottom of the standings, conversations tend to continuously come back to the same topic: the NHL entry draft.

I have always found the NHL draft to be a fascinating topic. It is of course an exercise in predictions. Picking which teenagers are going to develop into NHL players is a brutally difficult task, and despite every team pouring tons of money into a scouting department, teams continue to be pretty much completely terrible anyway.

But are some teams less terrible than others? Do some GM's really have an "eye for talent" that sets them apart from the rest? Most fans will say yes. Some fans will even say so with a high amount of confidence, but there is almost nothing to back this up. Drafting is difficult, and even evaluating the performance of a draft is arguably even more difficult. I have seen many approaches to it over the years but in my opinion they all lack something. They all lack a baseline.

In 2001, the brilliant baseball analyst known only as "tangotiger" created what he called "Marcel the Monkey" The purpose of the Marcel was to establish a basic way of forecasting the performance of baseball players. To quote Tango from the above link:

[Marcel is] is the most basic forecasting system you can have, that uses as little intelligence as possible. So, that's the allusion to the monkey. It uses 3 years of MLB data, with the most recent data weighted heavier. It regresses towards the mean. And it has an age factor.

The reason I loved Marcel is because it established a baseline expectation for what a forecasting system should be. The idea, to me, was that if you were going to pour hours and hours (and hours) into a hyper-advanced forecasting system that uses all kinds of fancy data and algorithms, you need to first be able to establish you can do better than the basic one. And, the funny thing was, many of them could not consistently beat Marcel by a significant margin, at least at the time.

The idea behind the Potato Picks is to establish a similar kind of baseline for NHL drafting. I have constructed the laziest method possible for picking hockey players. If the GM of your team was a potato, this is what it would have to rely on. The potato does not have access to dozens of grizzled scouts who travel the world and fill out reports, it does not even have access to highlights or scouting packages on YouTube, nor does it have access to Corsi metrics, Zone Entry data or any kind of proprietary system.

The potato's method is purposefully simplistic to the point of ridicule:

  1. Take a players points-per-game in the most recent season before the draft.
  2. Apply a positional adjustment.
  3. Apply a draft-age adjustment.
  4. Apply a height adjustment.

(Goalies are excluded. The potato can't be bothered with Goalies.)

I will go into more details on these adjustments in a future post, but that is essentially it. No even-strength metrics, primary points, or even granular age adjustments. This is the laziest way of doing it for someone who already has a full-time job, and it should serve us to set a basic expectation that any more advanced models, and, indeed, any NHL team, should be able to consistently exceed.

Using nothing more than traditional stats and biographical info, we can now "replay" a draft as any team and select the top-ranked player still available. We can then evaluate the benefit to the team of their GM and scouting staff as opposed to how they would have done using the dumbest way of picking possible.

To illustrate, I will do the 2013 Canucks. The following table will show who the Canucks selected with their picks, and who the potato would have selected, given which players will still available at the time of the selection.

Bo Horvat Alexander Wennberg
Hunter Shinkaruk Nic Petan
Cole Cassels Brendan Harms
Jordan Subban Lucas Wallmark
Anton Cederholm Jaedon Descheneau
Mike Williamson Conner Rankin
Miles Liberati Emil Galimov

While the potato went "off the board" in selecting a few players who were not even drafted in 2013, it honestly did not end up that much worse off. Wennberg is no slouch, and Petan has played more NHL games than Shinkaruk. Wallmark would go undrafted and get selected in the 2014 draft; he has scored 51 points in 43 AHL games this year with a few NHL stints in Carolina. If you were to give the edge to the Canucks in this draft, it would not be by as much as an advantage over a potato probably should be.

One draft does not mean anything. Over the coming days and weeks I will be doing every draft from 2013-2016 for each team in the NHL. I will also attempt to calculate a score for each one, so that each team can be compared to the potato to calculate the added value of their scouts. This method eliminates the problem of having to account for draft-position. At the end of this project we will be able to fairly compare how each team is doing in their recent drafting by evaluating their selections against the "potato picks" from the same positions.

Finally, when discussing our team's future we will have a basis by which we can marvel at the prescience of our team's leaders, and have some insight into who is getting their money's worth, and who can barely even outperform a potato.

comments powered by Disqus