Loading scores...
Draft Guide

Analytics in Fantasy Hockey?

by Ryan Dadoun
Updated On: October 4, 2018, 4:04 pm ET

Ryan Dadoun: If we went back even five years, analytics in hockey would have been viewed as fringe statistics with questionable value.  Now most teams look at fancy stats and they've become accepted to the point where NHL.com is going to start publishing them.  Even still, there relevance is still a topic being debated in some circle and their role when it comes to fantasy hockey is undefined.

My question to the group is do you see value in analytics when assessing players and are there advanced statistical categories that you would be interested in using in a future fantasy league?
Michael Finewax: Analytics shmanalytics (now I know how Cliff Fletcher feels).  I don't see the value of Corsi etc. in fantasy hockey.  Most stats that people play are straightforward and it really does not matter to me if someone is on the ice for more shots on net or the play is in the other teams end more often.  A goal is a goal and an assist is an assist. The only advanced stat that makes sense in fantasy is plus/minus and that has been a stat for decades.

Kevin BrownEven though stats like Corsi and Fenwick don't have any direct correlation to fantasy, there are other novel numbers that I have found useful in player evaluations. Zone starts, for example, provide a glimpse into how coaches deploy the players on their roster. I have often identified players whose zone starts have become more weighted towards the offensive end as breakout candidates before their traditional offensive stats have shown this.
Corey Abbott: I see value in using some advanced stats categories to assess players and predict what they are capable of.  Obviously, nothing is an exact science, but I find zone starts, PDO (combination of a team's save percentage and shooting percentage at even strength when a player is on the ice) and on-ice shooting percentage (used to see how well a team scores when a certain player is on the ice) to be interesting tools.  

Adjusted save percentage is a different take on the goaltending position that I like as well.  It attempts to create a more even playing field for goalies by giving a save percentage if they faced a league average amount of shots from each of the three shooting zones (high, medium, and low probability of success).  It was introduced to account for teams that give up more and less high-quality shots.  Thanks to war-on-ice.com for the glossary help.  

I don't really know which advanced stats would work in fantasy leagues, but I do think it's possible that tracking Corsi and/or Fenwick will lead to the elimination of plus-minus ratings and I'm OK with that.  I think I'd rather have a plus-minus represented by shots than by goals because it steers the conversation more to the individual and less to the team. 

Ryan Dadoun: A goal is a goal, but I think there's value for digging a little deeper for two reasons.  One is to attempt to determine if a player is playing over his head (which we can attempt to with PDO, although as Corey said it's not an exact science) and is thus over or undervalued.  The other is as an alternative to traditional stats.
Just like some people prefer using Hits as a category over PIM because they see hits as a useful player quality while PIM is more controversial, I think there's an argument to be made, as Corey mentioned, for replacing plus/minus with either F+/- or FF%Rel.  The first one is just the Fenwick version of plus/minus but the second one is more interesting because it measures how your team does while you're on the ice, compared to how it does when your off of it.  More often than not, your plus/minus is a reflection of your team as much as it is of the individual, but with FF%Rel, you're forced to evaluate the player based on his value to the team - both offensively and defensively.  Suddenly a good player on a bad team might have value over a great player on an elite team - at least as far as that statistic is concerned.  It's different enough from the other statistics typically tracked to potentially add another layer of strategy to drafting.
Ultimately though, the statistics tracked by War-on-Ice might just be scratching the surface of what's to come.  The NHL is stepping up its game when it comes to tracking players on ice movements.  It's not unrealistic to believe that puck possession itself will be a category that we can accurately track (which is what Corsi/Fenwick are trying to approximate now).  We might also be able to track passing (the percentage of successful ones executed, the percentage intercepted, etc), and maybe even some form of drives (the distance you successfully move the puck away from your net/zone).  Is there an aspect of the game, where if it was quantified, you would see value in it either in terms of determining a players fantasy worth or for use as a category in fantasy leagues?

Finewax: Ok guys.  Prove it to me. I still think that while the advanced stats are a useful tool when actually playing the game, I cannot see the value of it in fantasy hockey.

Dadoun: Are you talking about the value of it when assessing players or for use as categories?  If it's the prior, then the easiest answer is that a lot of what we award in fantasy hockey is the final result of something larger.  A lot can go into the production of a goal and analytics can give us a fuller picture which can be used to give us more insight towards future trends.  A guy that gets 20 goals one season might get just 10 the following campaign and analytics can give us a better shot at predicting that.

Even if you don't buy into that argument, the fact that teams are putting more weight into is noteworthy by itself.  In some cases lines have been evaluated and formed in part due to analytics, so knowing which players look good in that regard can give you insight into what players are likely to get more/less minutes.

Here's the shorter version of that: At the end of the day, fantasy hockey is about predicting future value.  Given that, any tools that are useful in assessing value are useful to fantasy owners.

If the question is why use them as categories, then that's more subjective.  If you like the current statistics counted then that's fine, but some of these categories (or categories that could emerge in the future with better player/puck tracking) can give us a different spin on things and challenge people to evaluate players in different ways, like I was talking about with FF%Rel.

One thing I'd personally like to see is an evaluation of current/future statistics with an eye towards better quantifying the work of defensive defensemen, because in my mind that's the group that gets ignored most by fantasy owners.  Hits/Blocked Shots help, but I got to believe there's more out there.

Finewax: The bad thing about hits/blocked shots is that it is so subjective because a hit in one rink is not necessarily a hit in another rink depending on who is keeping those stats in an arena. I remember growing up that there used to be phantom assists given to some star players (the Rangers come to mind) so players on other teams would have to be even better than that because their stats guys were above board. That has changed with the NHL really taking charge of scoring. 

A lot of the analytics are subjective, and although some are not, I still have a hard time thinking they make a difference in fantasy hockey.

Brown: When it comes to using advanced stats as scoring categories in fantasy, I think we’re a long way from seeing that sort of transition in hockey. I’m as big a proponent of these measures as anyone, but I see how difficult it has been for the fantasy baseball community to accept on base percentage or OPS as standard categories and I envision a similar struggle in hockey. The resistance towards a change of this nature comes from the fact that people like what they are familiar with and the better, more useful statistics are often pushed away because they’re not easy for the general public to understand.
In response to Michael’s request for us to prove the usefulness of advanced stats, I’d like to point to a simple advanced measure that we haven’t discussed yet. Websites like war-on-ice.com typically break down a player’s assists to demonstrate which were first assists and which were second. I think there’s a lot to be gained by identifying players who rack up a high percentage of first assists since those players are driving play and creating more scoring opportunities than the average, even if their raw point totals don’t show it. 
What say you to that, Mike? Do you see value in evaluating a stat like this for the purpose of getting ahead of your competition?
Finewax: That's  a useful stat but that stat has been available for a long time. Their point totals would show it because they are still getting an assist. Usually players who have a lot of first assists are good playmakers anyway and are usually highly regarded. How many players have 20 first assists and just a handful of secondary assists? Besides, a player could make the play, pass off to a teammate who takes a shot and someone else gets the rebound. The secondary assist is far more noteworthy than the first assist.
Dadoun: The point Michael raised is worth discussing further because I think it's at the heart of this argument: Context is critical.  A goal scored due to a goaltending mistake isn't as much of an indicator of a player's skill as a goal scored by outmanuevering two defenders and then beating the goaltender one-on-one.  A goal scored off of a players back might not be as good of an indicator of skill as a rocket one-timer that slams into the back of the net.  Why does that even matter when all four of those scenarios count as a goal?  Because we ultimately don't care about the goal already scored.  That one's already in the books so it either counted for us or against us, but either way, there's nothing we can do about it.  What we care about is the next goal, the future performance of a player because that's at the foundation of what we do.  We attempt to predict what's to come because that's what people have an interest in.
So context matters because it's how we weed out the misleading statistics.  To use a simplified example, back in November I wrote an article at a time when Joe Thornton had seven goals in 20 games.  I mentioned in it that three of his seven goals were empty netters, so his pace should be taken with a grain of salt and sure enough, he only has 11 goals in 50 contests now.  Now that's an obvious example of gaining context through statistics, but it's only obvious because empty net goals is a statistic we've used forever.  Some of these new statistics can be as helpful in determining that all important context and other statistics to come could be even better.  Right now, I think the potential for fancy statistics to help individual fantasy owners is heightened precisely because they're new and not commonly accepted.  It means you have a potential edge over your competition if you understand analytics and consider them to an appropriate extent because odds are your opponent won't be doing the same.
But back to the four goals I discussed, fancy statistics isn't as good as looking at each goal for yourself as they don't tell the full story.  However, if you don't have the time to watch every NHL goal, then analytics will at least provide you with a richer picture than you would have gotten by simply looking at the base statistics.  Ultimately, the more information you have, the better position you're in to make the call as to whether one player with 50 points will be more or less valuable in the future than another player with 50 points.
Abbott: I think right now advanced stats are still too unconventional for most of the hockey-watching public, but that is going to change when NHL.com undergoes its big revamp later this month. They are adding Corsi, Fenwick, PDO, Zone starts, average shot distance, goals and assists per 20 and 60 minutes, penalties drawn and taken per 20 and 60 minutes among others.  That will go a long way to changing the perception and making fancy stats more mainstream.  

As I said earlier, I would like to see better alternatives to the current plus-minus rating and which players drive possession could be it.  Plus-minus as it is is incredibly flawed and I don't like it as a category in fantasy leagues.  It ends up being more of a line stat or a team stat rather than the individual and it isn't a very telling one because it is ripe for inaccuracies.

When sites like Yahoo begin providing different options, I think we will get some replacements for plus-minus.  Most leagues I'm in currently have already phased out penalty minutes for hits and/or blocks and I won't miss plus-minus ratings.  

Brown: Corey, I share your dislike of plus/minus since I don’t think it really tells us anything about what a player has accomplished so it has little value in fantasy leagues or in real-life. Average shot distance is a very interesting stat since it helps us try to separate goaltender performance from team context, which is a mystery that has always been extremely difficult to parse out. After watching Ilya Bryzgalov go from superstar to also-ran after leaving the Coyotes and seeing the Predators spend the better part of a decade turning average goaltenders into stars (Pekka Rinne notwithstanding), the fantasy community understands that certain situations are more favorable for netminders, but we have never been able to attach a number to that.
All told, I think we will eventually find that the prevalence of fancy stats will lead to more useful player analysis and projection methods, even if the way in which that happens is not apparent yet.

Abbott: I'm looking forward to seeing what analytics can bring to the goaltending position as well.  Adjusted save percentage and average shot distance can help place netminders on a much more even playing field.  Every shot a goalie faces is not the same and some teams give up more high-quality chances than others, but these stats among others can provide us with a more detailed picture.  

That can be said about skaters as well and not just goaltenders. That's why I have warmed up to advanced stats after being apprehensive at first.  They can be used to supplement the numbers we've had for years and in some cases delve deeper into them.  

I think it's great that the NHL is adding fancy stats to their website.  Advanced stats and the player/puck tracking that was unveiled at the NHL All-Star Game bring a lot to the table for teams and fans.  It feels like the NHL is finally catching up.

Ryan Dadoun
Ryan Dadoun is an Associate Editor for Hockey on Rotoworld. Feel free to follow him on Twitter or check out his blog.