Let’s say you play in a no-movement fantasy league that does scoring by either standard rotisserie or points. In the most unbelievable feat ever achieved in the history of hockey, Connor McDavid scores 30 goals and 80 assists in the first game of the season – somehow managing to setup or score a goal almost twice every minute. He then, having broken the game of hockey, promptly retires.
I give you this surreal scenario to make a point. In your standard rotisserie or points league, you are completely satisfied with this outcome. McDavid got you the points you pencilled him in for when you drafted him and they count just the same in one game as they do if he had spread them out over 82 games (a much more likely scenario, if you ask us).
But in a head-to-head league, you might not be as pleased. Sure, you pretty much sewed up a Week 1 victory, but now your No. 1 draft pick isn’t going to help you in Week 2 or beyond because of his retirement.
The outlandish example above is just to frame the discussion here about standard deviation and coefficient of variation we are about to have. In other words, this is all about consistency and that matters a lot more in head-to-head leagues.
Every Friday last season, as part of putting together the fantasy forecaster for ESPN.com, I saved a snapshot of the current statistics in the league. Using that collection of sheets of data, we can take a deeper dive into the consistency of a player’s production from week to week. Did a lot of his points come in bunches? Or did he spread out the scoring throughout the season?
We’ll get into some specific players below, but for this exercise we are looking at each player’s fantasy points per 60 minutes (FPP60) in each week of the season from the start to May 7 as a Friday-to-Thursday sample. The last “week” of the season is also included, but its the May 8 to May 20 session that was mostly just the Vancouver Canucks playing out games postponed by their COVID outbreak. Due to the restraints of the data, players who were traded midseason are not being looked at (its too onerous to manually adjust the amount of raw data to get them to match) and I excluded any players with fewer than 10 games last season.
A quick primer on standard deviation: It’s a look at how much spread there is in a set of data compared to the mean average. So, Connor McDavid had 9.41 FPP60 last season, and we are able to calculate his individual FPP60 for each week of the season. Some were much higher than 9.41 FPP60 and some were much lower. We can calculate how much he strayed from his average throughout the weeks – or his standard deviation – as 4.32 FPP60.
OK… So what? Is that high? Low? Good? Bad? Well, it’s actually relative. It’s high or low relative to his own overall FPP60 and also good or bad relative to how others players fared. But generally speaking, a higher standard deviation means there were more extremes in the data. Which also means, less consistency. A lower standard deviation means the values in the data were closer to the mean. In other words, more consistency.
As far as being good or bad, you can argue it both ways. If you are in a head-to-head league, you obviously want more players with a low standard of deviation in order to have them help you win each week. Or maybe you don’t? You could be in a league with an active trade market and want to use players who pop off on occasion to trade up in value.
In terms of making it relative, we can find the coefficient of variation or the relative standard deviation, which is just the standard deviation expressed as a percentage of the mean average. We need to do this because not all of the player’s are created equal. McDavid’s standard deviation is basically the same as Carl Grundstrom’s, but they mean very different things relative to the player.
In other words, was the spread of McDavid’s FPP60 through the season high relative to his overall FPP60? The coefficient of variation in this case is 45.9 percent (deviation divided by mean). That’s one of the relatives, but the other is looking at other players.
Mitch Marner’s FPP60 was 6.49 last season, so not as good as McDavid, clearly. But his coefficient of variation (CF) was 27.3 percent. That means Marner had more weeks last season – many more weeks – in which his output was closer to his overall average than McDavid. Marner was more consistent in his fantasy output throughout the season.
We can see this looking at the weekly FPP60 of both players. While McDavid had four weeks of the season with an FPP60 over 12.0, he also had three weeks of the season with an FPP60 below 4.0. Marner never had a week above 12.0 FPP60 and only had one week in which he was just below 4.0 FPP60.
So let’s have a look at some of the players who popped on either end of the spectrum for coefficient of variation (CF).
Consistency Kings
Mitchell Marner, F, Toronto Maple Leafs (CF 27.3%): I didn’t pluck Marner from thin air to use him in the example above. He was, by this measure, the most consistent fantasy player in the NHL last season. Even though the top of the list for low CF is overwhelmingly defensemen, Marner still rises to the top. Actually, among forwards for low CF, John Tavares (CF 32.8%) ranks third and Auston Matthews (36.2%) ranks ninth. The Leafs put up fantasy points and they do it game in and game out.
I actually think there might be enough here to argue that Matthews should be the No. 1 overall pick in head-to-head leagues over McDavid. The overall output is projected to be more from McDavid, but Matthews is right behind him and, by this measure, much more consistent. Remember when I mentioned McDavid had three weeks last season with an FPP60 lower than 4.0? Matthews only had one. It’s certainly something to chew on if you have the first pick.
Ivan Provorov, D, Philadelphia Flyers (CF 29.8%): As the defenseman among the top 20 for fantasy points with the lowest CF, Provorov is simply our poster boy to point out that defensemen are much, much more consistent than forwards. Of the top 100 skaters with the lowest CF percentage, 59 of them are defensemen. Not a single defensemen among the top 50 for fantasy points last season posted a CF higher than 54.5 percent. How the position collects fantasy points – power-play scoring, blocked shots, hits – means there isn’t nearly as much variation as there is among forwards.
Mark Scheifele, F, Winnipeg Jets (CF 34.0%): Yet another metric of fantasy value in which Scheifele shines. Without miss, I find myself banging the drum for Scheifele’s virtues ahead of every season. He had the fourth-lowest CF among all forwards (and that’s third if you take out Garnet Hathaway and his fantasy-irrelevant 12 minutes per game). Scheifele finished 12th among forwards for total fantasy points last season, 15th the season prior and 16th the season before that – yet he is ranked 46th in the default ESPN rankings with a current average draft position of 52.0. … Every year it’s like this.
Who else is consistent? Ryan Pulock (CF 28.4%), Brady Tkachuk (CF 34.4%), Dougie Hamilton (CF 35.2%), Jake Guentzel (CF 35.5%), Adam Larsson (CF 36.5%), Adam Fox (36.7%), Victor Hedman (36.7%).
Streakers
Jonathan Huberdeau, F, Florida Panthers (CF 66.7%): I was already extremely bearish on Huberdeau based on his draft position before going through this exercise. Now I’m even less interested in him as a first-round pick (which is where he landed in the ESPN rankings). Huberdeau played all but one game last season, yet still had five weeks in which he fell below 4.0 FPP60. That’s just shy of a third of the matchups in head-to-head leagues when Huberdeau let you down. That is not good enough for an early pick.
Luke Kunin, F, Nashville Predators (CF 105.5%): There was some missed time to injury that helped inflate his variability, but there were also some wide outcomes to Kunin’s contributions. During six separate weeks last season, Kunin crested past 8.0 FPP60 – which is an elite mark. But those weeks were largely spaced out (with a few of the more successful weeks starting to cluster toward the end of the season). The end result was 6.15 FPP60, but missed time pushed him out of the top 300 skaters for fantasy points. The point here, is that a little bit of consistency and a locked role on a scoring line would push Kunin into the top 100 forwards at his current FPP60. We’ll have to see if he can achieve either of those two prerequisites.
Mike Hoffman, F, Montreal Canadiens (CF 87.4%): Not all streaks – good or bad – are on the player; some of them have to do with coaching. Hoffman’s see-saw production last year could be chalked up to his last-minute role with the Blues. A free agent at the most unfortunate time mid-pandemic, Hoffman didn’t get offered any contracts through a protracted offseason, then had to settle for a one-year chance with the Blues – though it was clear he wasn’t a part of the plan ahead of time. It showed with his deployment, which was largely on the third line at first and changed frequently. A couple of big weeks through the season for FPP60 might make his standard deviation from last season high, but they also show he still has some offense to contribute if used properly. A new chance with the Habs on a three-year contract might be the right fit.
Viktor Arvidsson, F, Los Angeles Kings (CF 82.6%): The Preds offense had some issues last season, posting its lowest team goals per game since 2013-14, and Arvidsson was among those who suffered. Now with the Los Angeles Kings and with a chance to help redefine the line assignments, the future could be brighter again. Like Hoffman, I think Arvidsson’s variability last season shows he still has the potential to be better in the right spot.
Andre Burakovsky, F, Colorado Avalanche (CF 77.0 %): The Avs stopped using Nazem Kadri on the top power-play unit partway through last season, and it was to the great benefit of Burakovsky some of the time, and Joonas Donskoi at other times. Burakovsky had several weeks of spiked production when he was the lucky player to join Nathan MacKinnon and Co. on the man advantage, but the role was traded back and forth with Donskoi. Now that Donskoi is a member of the Kraken, maybe Burakovsky can settle into the gig?
What about goaltenders?
Talk about variance. You may have guessed that there isn’t much consistency here when compared to the forwards. A goaltender’s FPP60 from week to week is a roller-coaster ride, even for the best of them. It’s not helped by the fact that they could easily score negative fantasy points in a bad game, nor by the fact that they don’t play every game.
Marc-Andre Fleury had the best coefficient of variation among regular starters last season at 67.4 percent. While more than a few regular starters were into the hundreds for percentage due to several weeks in which they would score negative points. Joonas Korpisalo, for example, only averaged 0.43 FPP60 on the season, with wild swings in FPP60 that pushed his standard deviation to 2.63 FPP60 and his CF to 607.2 percent.
Even Cam Talbot, who no one doubts was a valuable fantasy asset last season after finshing 10th overall for fantasy points among goalies, had a standard deviation higher than 100 percent. He had six weeks (some of them injured) in which he scored zero or below zero FPP60, while also posting four weeks with an FPP60 better than 4.00 (which is very good for goaltenders). So Talbot, like many goalies, pumped in those fantasy points during stints of action, while not contributing in other weeks.
This variation in the contribution from goaltenders is why I like to carry a minimum of four goaltenders that get decent amounts of action in head-to-head leagues.
Frederik Andersen, G, Carolina Hurricanes (CF 223.0%): I don’t care about Andersen’s high variation from last season. What I do care about is Alex Nedeljkovic, Petr Mrazek and James Reimer all finishing among the top seven goaltenders with the lowest CF variation last season. That tells me that the Hurricanes are a stable place to backstop, which could do wonders for Andersen – and even Anttii Raanta – this season.
Juuse Saros, G, Nashville Predators (CF 123.3%): Saros doesn’t stand out as exceptionally consistent among goaltenders, ranking in at 24th overall for low CF last season. That said, if you take his play from only the third week of March forward, which followed a very rough start, his CF drops to 67.9 percent, which is just behind Fleury for the most consistent in the league.
For goaltenders who finished in the top 20 for fantasy points, Jacob Markstrom has the highest CF percentage (269.6%), with Kevin Lankinen (230.9%) and Vitek Vanacek (205.1%) not far behind. But again, the position is volatile.
Ready to play? Sign up for free today.