It’s here, the moment you’ve been waiting for: Our fantasy hockey cheat sheet is ready. It’s time to start prepping for your draft.
For those who aren’t new here, you know the drill: Skip down to the big, bold link and press download. No need to read what you already read last year.
For those who are new, here’s the deal: We have a fancy spreadsheet that creates fully customizable rankings tailored to whatever weird way your league is set up. There’s no real standardized hockey league and the best way to get ahead is with rankings that take into account your league’s specific settings. Your league is not the same league as the next person’s and that’s a very important point when it comes to research and rankings. In fantasy hockey, it’s not one-size-fits-all and depending on any ranking — no matter how smart the person writing it is — might already be your first mistake.
Advertisement
Other rankings are great for standard leagues or for research purposes, but they may not apply directly to your own league. That’s why this spreadsheet exists, to create an easy way for any fantasy hockey manager to have their own rankings that actually fit their league’s scoring system and format. (And trust me, it pays to have projections for every stat you need, not just points.)
You can find download links to the sheets below (in Excel and Numbers/Mac formats). Underneath that, you can find instructions on how the sheet works as well as answers to some frequently asked questions. (Note: All patches and updates will be updated at the bottom of this story.)
One final note: Yahoo has lost its mind with its position designations this year (Shayna wrote a great story about it here). We’re going to keep our eyes on that and update the sheet once that hopefully changes for the better. With that being said, we have a second sheet for those in Fantrax leagues where the positions aren’t outrageously designated.
The first thing you need to do is to go to the “Settings” tab to start customizing. On the left side is where you’ll enter your league’s scoring system. If you’re in a points-based league, you only need to worry about the points column. If you’re in a categories or roto league, only worry about the categories column.
For points leagues, simply type in the value next to each stat. If a goal is worth three points, type “3.” That’s it.
For category leagues, I use standard deviations to score players in each category, so simply type “1” next to whatever category your league uses. This weights each category equally. If you would like to add more weight to certain categories (goals, assists, shots) over others (PIM, hits, blocks) based on scarcity, you can make the former categories worth more than one and the latter categories worth less.
Advertisement
Next, you are going to focus on the right side of the page.
At the top, fill in how many players are in the starting lineup. If you’re in a league that doesn’t separate forwards by position, just make sure all three positions add up to the total amount of forward positions. At the bottom under “number of teams,” type in how many teams are in your league.
For ADP data, type in which fantasy service you use.
For league type, type in “points” or “categories” depending on whichever your league uses. This year there’s only one page for rankings as opposed to separate pages for both so this step is crucial as it makes sure the correct total fantasy point calculation is used.
For forward positions, type in “C/LW/RW” if it’s separated or “forwards” if it’s not. I realize all of this is pretty self-explanatory when looking at the spreadsheet, but for those less fluent, I figured I would spell it out for you.
Then there’s “games played projections.” “No” means that every skater is projected to play all 82 games, while “yes” uses my games played estimates. If you’re of the mind that injuries are random, it might be better to go by a player’s projected per-game output. Just make sure you go through the list to mark the players who already have known injuries, like Brandon Montour or Jack Quinn.
There’s one more setting, but that one requires further explanation.
There’s a reason you’re asked to fill out league size and starting lineup size above, and that’s so the spreadsheet can calculate each player’s value over replacement — the key to the rankings.
Essentially, each player has a projected fantasy point rate, but that number needs the added context of what position he plays and what other players at the position are expected to do. Generally, this is to control for the fact that there’s an abundance of strong scoring centers and not as many capable wingers or defenders.
Advertisement
For example, if Dougie Hamilton is expected to score 400 fantasy points and Tage Thompson is expected to score 500, it’s the latter that seems more valuable. But it all depends on that baseline, where you can generally get a 400-point center for your bench just as easily as a 300-point defender. That makes them equally valuable.
Now, here’s where things get tricky — and I give a lot of credit to some commenters last year for their thoughtful discourse on the subject: How do you actually establish the baseline?
Previously, it was an automatic calculation based on league size and starting lineup combined with how players are normally drafted within the top 100. That generally made defenders more valuable, but some were saying it didn’t go far enough — that the need for four defencemen in a standard league makes them even more valuable than that method suggested. They posited that positional requirement was the way to go, that the baseline should be starting lineup multiplied by team size — essentially, the true definition of replacement.
Using the draft to dictate the process helps account for the fact that people don’t normally draft defencemen very high (even if they should), but perhaps it’s not cognizant enough of the reality of defencemen needs. With most teams now opting for a 4F1D power play, finding a good defender is more difficult than it’s ever been.
For that reason, there’s now an option for how to establish baseline: draft (the usual way based on top 100 selections), position (based on starting lineup and league size) and “blend,” which mixes both together. I made blend the default because I’m not sure which method is better, but the option is now there for you based on your own beliefs. As long as you’re not reaching for targets and drafting along ADP, you should be fine either way.
All of that work sets up your rankings, but once you go to “The List” page you will see that everything is probably out of order. All you need to do now is sort. In Apple Numbers, just double-click the top of the rank column and press “Sort Ascending.” In Excel, click the arrow at the top of the rank column and then click “Sort Ascending.” The rank is based on each player’s VORP, so alternatively you can sort by that as well. If you’re in a league that uses a salary cap, the “/$” column compares VORP to a player’s salary, so alternatively you can also sort by that.
For keeper leagues, use the “keep” column to denote any player that is a keeper. You can then filter them out using the arrows above in Excel or the filter option under “Organize” in the sidebar in Numbers.
Advertisement
The final step is disagreeing with us. I know you probably will and that’s OK! In fact, last year we made that process even easier with a specific column for adjusting called “boom or bust.”
The gist here is that we know a model, no matter how good, is not infallible. Human intuition is important and we wanted to include some adjustments based on our own feelings and disagreements with the model’s output. This is especially helpful for younger players who look ready to break out or players who switched teams — two instances the model has historically struggled with.
Shayna Goldman and I went through each projection and debated whether a player deserved to be a little higher or lower. On the actual list page you can see a “+” or “-“ denoting exactly which players we tinkered with, making it easy to see exactly where our subjective judgment came into play.
This is also where your own disagreements come into play. On the sheet there’s a column labelled “ADJ” that’s already filled with our adjustments, but you can change those or add your own. It’s as simple as adding a “+” or “-“ for small adjustments, or “+ +” and “- -“ for bigger adjustments (not recommended for goalies). Don’t be afraid to use your gut instinct.
The final step is actually drafting your team. Here’s what I usually do: Once a player is drafted, I use the “keep” column as a way to denote which players have been taken already. If they’ve been drafted I might type a “Y” in that column, and then you can use the filters above the column to automatically hide players that are selected as it happens.
It’s also worth keeping track of your team while you draft and that’s where the “Team Comparison” tab comes in. Type in the players you draft as it happens and the table will auto-populate their projections. If you’ve got time you can also type in the other teams as well for an instant comparison of where your team stands.
First and foremost, it starts with Evolving Hockey. I don’t know where I would be without that site, but all the data that goes into creating the projections comes from there.
After that, it’s a relatively simple Marcel projection that uses the past three years of data for each stat weighted by recency, regressed to the mean, and age-adjusted. That’s done on a per-minute level and applied to time-on-ice projections based on each player’s ice time last season, adjusted for where they’ll likely fit on the depth chart this year (with the help of the NHL beat writing staff here).
Advertisement
This is always a very important question. While the projections themselves look absolute, the reality is that there are margins of error. A player projected to hit 80 points really means “somewhere between 70 and 90 points.” Worst case he’s on the low end, best case he’s on the high end — with some exceptions to the rule that end up way higher or lower. I didn’t go through last year’s projections, but the general rule of thumb is points are usually off by nine-to-10 points while shots are usually off by 25.
It’s worth noting that our subjective adjustments improved the margin of error for points by one and were in the right direction 60 percent of the time. If you consider the projections a standard over/under, that’s a pretty good hit rate.
Not exactly, and it goes back to the point made above. Think of a projection as a probability and the figure shown as an average of that probability. The likeliest point total for McDavid next year is 142 points in 82 games, but as we saw above, the average error for projections is about nine points in a normal season. So really, that’s saying he’ll likely score between 133 and 152 points in a season, give or take. It’s a ballpark figure where the goal is to be less wrong. A modelled projection will do that better than just guessing, and a good projection will do better than a bad one.
No, and it comes down to the range discussed above. If David Pastrnak is at 542 fantasy points and Nikita Kucherov is at 540 points, both players are well within each other’s error bounds. Pastrnak is probably the better player, but if you want to argue Kucherov is safer because of his consistency, pedigree and the fact Tampa Bay didn’t lose its top two centers from last year — those are important debates to have while making selections.
That type of context won’t be explicitly included in a projection like this which features zero subjectivity whatsoever, aside from the post-hoc boom/bust adjustments. That’s why it will be helpful to read all the other fantasy hockey draft content coming down the pipeline to become more informed about the players who might under or overperform their projections.
In general, it’s best to follow the rankings, but it’s fair to squabble when players are very close to each other. In those cases, you’re usually splitting hairs.
Fair enough! For points leagues, start volume is the key for goaltenders and often when you see a goaltender ranked higher than expected here, it’s for that reason alone. In a league that’s veering more and more toward tandem situations, it pays to have a legitimate starter likely to play in 70 percent of their games as those are few and far between. That’s even true if the goalie isn’t anything special, like John Gibson in Anaheim or Darcy Kuemper in Washington.
GO DEEPER
Re-drafting the NHL class of 2020: Stützle goes No. 1, Nikishin rises dramatically
(Top photo: Steph Chambers/Getty Images)
Subscribe to The Athletic for in-depth coverage of your favorite players, teams, leagues and clubs. Try a week on us.
