Roster building is always one of the most fascinating subjects in sports because it’s something nearly every fan thinks and dreams about – how to turn their team into a champion.
For hockey there are many blueprints that can get the job done, but they almost always start at the top. A team needs high end talent to compete and win, the proof of that is engraved on the Stanley Cup every season. It’s rare that a team wins it all without a player who’s among the league’s best and looking back through past Cup winners, I honestly can’t really recall a team that pulled it off without star power.
That effect is even larger in the NBA, a star-driven league where you’re not winning without an MVP-candidate on your team. Star power is a big topic of interest in the NBA these days in the wake of the Golden State Warriors dominance and after an off-season where a bunch of upper tier players moved to new locales. One of those stars on the move was Gordon Hayward, which led to a piece over at FiveThirtyEight.com that measured the effects of star power in the NBA and where certain players fit compared to past championship teams.
With that in mind, I wanted to recreate the analysis for the purposes of the NHL to see where the league’s best players fit in relation to each other, which teams have the most star-power, and how much it really matters. We know it’s not at the same magnitude of the NBA, but how far off is it?
First, we have to look at past champions. Five Thirty Eight looked at the three best players on each championship team to figure out how good a player needs to be to fit championship criteria. We’re going to use five players instead because NHL rosters are bigger, with players ranked by their Game Score percentile (to see how good they are relative to their position). The only issue is we can only go back 10 seasons because that’s when the NHL decided to start tracking real-time stats.
Over the past decade, the best player on each Cup-winning team has been better than 99 percent of the league at their position on average, with the worst player at 97.7 percent. That’s pretty similar to the NBA as it essentially equates to an MVP calibre season. In both sports it seems it’s hard to go the distance without at least one guy performing at that level. That’s what it takes to win, but even that alone isn’t enough. Teams need more than one guy who can carry the load as evidenced by four slots over the 90th percentile and one more very close to it.
With that info, we can establish some tiers for the league’s best and brightest. I went one standard deviation lower than the average to better capture a minimum threshold for each championship player grouping, as using the average itself would filter out half the players that would’ve qualified for a grouping.
Here are the tiers.
Tier 1 – Think year-end award candidates.
98.5% - 100.0%
Tier 2 – Think elite players, among the best at their position.
96.5% - 98.5%
Tier 3 – Think all-star calibre talent, among the best in the conference.
94.0% - 96.5%
Tier 4 – Think great players who complement players in the above tiers.
91.0% - 94.0%
Tier 5 – Think very good players who are borderline stars in the right environment.
87.5% - 91.0%
Now it’s time to see who fits where. Rather than using last year’s numbers, I used projected Game Score for 2017-18 to show what kind of top-end talent teams have going into next season. It’s what we’ve been using for playoff odds and last week’s off-season changes post, so this also works as a check and balance to see if the numbers themselves mostly make sense. While it won’t create a perfect list (no metric or human can), it should make for a reasonable estimate of the league landscape even if there are some disagreements about specific placements (even I don’t 100 percent agree with where some players land, so please don’t yell at me, thanks). It’s also important to remember this is just a projection and players can certainly play better or worse than their tier suggests. Without further adieu, here’s where each of the league’s stars ranks.
There are some surprises sprinkled throughout, but I do think this list works for the most part (fans from certain fanbases may disagree and to them I say yes, I do hate your favorite player and/or team). Most of the surprises are likely due to teammate effects which are difficult to suss out, making some players look better because of who they play with. Overall, it feels like the majority of players are ranked correctly, though I’m a little biased because it’s my model.
The biggest difference between this and the NBA version is that there are a lot more players within in each tier. That’s likely due to larger rosters and also because the gap between the best and the rest is much smaller in the NHL. That means there’s a larger group of players who can be foundational pieces to a Cup contender, which makes sense given the current state of parity in the league.
So to win a Cup, all you need is one guy from each tier and you’re set, right? Well, not really. You don’t necessarily need one from each – no team is actually built like that – just as long as you have enough star power to split the difference. For example, a team might not need a Tier 1 guy at all if they have multiple Tier 2 or 3 players, like Columbus, or they might not necessarily need a Tier 2 talent if they’ve got someone at Tier 1 and a few guys at Tier 3, like St. Louis.
There’s a lot of flexibility here and what 538 did was create another metric called star points to figure out which teams had the most high-end talent. The method is pretty arbitrary (three points for alpha, two points for beta, one point for gamma), but this is just a fun off-season exercise anyways so we’ll go along with it. For our version, each team gets five points for any Tier 1 player, four points for any Tier 2 player and so on.
A team with one player from each tier would have 15 star points, but teams have won with less. Seven of the last 10 champions had 14 or more star points with the other three teams having 11, 12 and 13. That makes 14 star points a pretty reasonable championship cutoff. Anything less and a team would have to have some great depth to compensate (which those three teams did). Having a lot of stars is no guarantee for success, though it does help a lot. Nearly half of teams with 20 or more star points make it to the conference final, and one-fifth went on to win the Cup.
It’s only 10 seasons of data, but the trend is still interesting, albeit unsurprising. The more star power a team has, the more likely they are to succeed. We probably didn’t need math for that, but it helps to see it illustrated and examined, especially for those that are skeptical that the NHL is driven by their star talent. It may not be to the degree of the NBA, but teams still need an ensemble of top tier players to go the distance. Those players aren’t always readily available either which is what makes complete rebuilds so attractive.
So how does the current NHL landscape look? Which teams have enough star-power to be a Cup contender and which teams are still missing a key piece or two of the puzzle? According to star points, there are 11 teams in the league with enough star power to win a championship with 12 other teams not too far behind the cutoff.
Star points isn’t a ranking of team strength, but it is pretty cool to see that the better teams find themselves near the top and the weaker teams near the bottom. While there’ll be some differences in depth, what clearly separates the good and bad teams in this league are the best players. Those guys are the difference makers and without them teams languish near the bottom in the hopes they can draft their own superstars and make their way up.
What’s missing from most of those teams at the bottom are Tier 1 players. It’s hard to get far without one and you can see that it’s a critical piece to most of the contending teams. Just one team is above the championship cutoff without a Tier 1 player and that’s Columbus who barely gets by with a trio of Tier 2 stars. Those players at the very top, the MVP-calibre players, those are the ones who drive the bus and it’s hard to win a Cup without one. But having one MVP-calibre player isn’t enough. The Devils aren’t going anywhere unless they add to Taylor Hall and the Islanders haven’t got very far with just John Tavares and some low-tier stars yet either.
They’re an important piece, but they’re not the only piece and that’s why this kind of analysis is important. It shows how crucial it is to build a core of great players and how many pieces are necessary to get to the point of contention. Building up that foundation is the hardest and most essential part, but that’s not where the process ends. Just like it takes more than one star to win, it takes more than just the stars to win, too.
Hockey is the ultimate team game as they say, and there are 13 players outside of a team’s best five that also have to contribute. In basketball the best players play most of the game, but that’s obviously not the case in hockey where players take shifts which is what makes depth so critical. For years Pittsburgh had the necessary star power to compete with Sidney Crosby, Evgeni Malkin and Kris Letang, but they couldn’t do much without more help around them. We’re seeing the same thing play out with Chicago as their supporting cast gets gutted leaving them without much help for their stars.
The bottom line is that this league runs on star power. Teams are defined by their cornerstone players, win or lose. But while the best and worst of this league is usually separated by a team’s best players, it’s not the whole picture. Star power is the start of a good team, and it’s what makes a great team, but they need support around them to become a championship team.
THE LATEST HOCKEY NEWS PODCAST: