# Standings points make more sense than wins for replacement level metrics in hockey

Let’s get pedantic.

Baseball analysts have been measuring player performance by comparing results to those of a hypothetical replacement level player for over a decade. The idea is to create a common baseline of comparison by measuring how much better a player performed than one that could be easily acquired as a replacement.

In hockey, the most straightforward way to think of a replacement level player is a composite of the 14th forward or 8th defender on every team’s depth chart. That player will likely be an NHL-AHL tweener who a team could acquire for little or no cost either via waivers or a low impact trade involving a low round draft pick or another player of similar value.

In baseball, the common currency used to describe a player’s value above replacement is wins. Several analysts have ported versions of that approach to hockey over the last decade with the most currently well-established version hosted at **Evolving Hockey**. Wins Above Replacement (WAR) estimates how many wins a player provides above a replacement level player.

While carrying the concept of WAR from baseball to hockey has obvious appeal, the transition isn’t as straight forward as it might seem. In baseball, sorting the standings by wins will tell you which team is the best. But that isn’t always the case in hockey. The existence of three point games complicates the interpretation of WAR. When discussing the standings in hockey, we don’t typically talk in terms of wins and losses. We talk in terms of standings points. Thus, citing WAR doesn’t conjure the kind of immediate understanding of value that it does in baseball.

### Why Standings Points Above Replacement

That’s why Evolving Hockey last year introduced Standings Points Above Replacement (SPAR). Intuitively, saying a player is worth 7 SPAR gives a more immediate sense of value than saying a player is worth 3.2 WAR. If you tell me a player takes a team from 41 to 44 wins, I know that’s useful but it doesn’t translate directly to the standings. If you tell me a player takes a team from 89 points to 96 points, I know exactly what that means. They went from being out of the playoffs to being a bubble team.

For this reason, SPAR is a better and more intuitive metric for hockey than WAR. Aside from tradition, I can’t see a reason why WAR would be preferable. The measurements are essentially the same both in technique and logic but SPAR uses a baseline and framework that makes sense for hockey while WAR leaves the reader to do additional math to figure out what that increase in wins means in the standings.

Using SPAR also makes more sense from an analysis perspective. We can do lots of useful things with it that WAR doesn’t allow as cleanly. To start, let’s figure out how many standings points a team full of replacement players would generate. We can do this using data from Evolving Hockey. By summing up all of the SPAR in a give year, subtracting that from the total number of standings points for that season, and then dividing that by the 31 (or 30) teams, we can estimate the standings points of a replacement level team.

The following chart does just that for all seasons back to 2007-2008 excluding the lockout shortened season in 2012-2013 and the currently suspended season.

On average over the last dozen years, a team of all replacement level players would generate about 55 points. That makes intuitive sense. If you filled a whole roster with players available on waivers, they wouldn’t win many games. Another way to think of this is that teams get on average about 55 free points per season.

If we want to dig a little deeper, we can do the same calculation as above at the team level by subtracting team SPAR from their actual standings points. This gives us an estimate of about how many “free” standings points a given team banked in a season. Below is the distribution of those free points.

The distribution is centered at 55, which aligns with what we found above. What’s useful about the distribution is that it shows us that if we expand one standard deviation in either direction from the center, we get a range of about 49 to 62 points. That means that front offices entering a season should expect to fall somewhere in that range if they filled their entire roster with players they could acquire for little or no cost.

From a team building perspective, a front office’s job is to add enough talent to get from those theoretical free points to the playoffs. If we use the average we found by looking at the seasons in aggregate, we start from an average of 55. The average standings points of the 16th place team is 94. So, on average, if a team wants to make the playoffs, they need to find at least 39 standings points among the 23 player roster.

This is the advantage of SPAR. We can’t do any of the above with WAR as intuitively because three teams with the same amount of wins could all have meaningfully different numbers of standings points. But using a SPAR model, a team can project their finish in the standings by starting from a baseline of 55 points and adding the value of each player on their roster until they arrive at a total. If that number lands at 84, they should be prepared to try to improve their roster if they hope to make the playoffs. If that number lands at 98, they should feel good about their playoff chances and look to add at the margins.

### The fun stuff

Aside from the team building aspect, we can also do some fun stuff with SPAR. For example, the first question that comes to my mind is how often good players are the difference between their teams making and missing the playoffs. As in, if we removed this player from the roster, would the team still be good enough to qualify for the playoffs?

The following chart attempts to dig into that idea. I’m too lazy to code all the playoff rules and tiebreakers so I settled for measuring whether a player moved their team from below 16th place in the NHL to 16th place or above.

The chart shows the top 40 seasons in terms of impact on standings order and the blue lines represent players whose teams would have fallen below the 16th place cut line if swapped with a replacement level player.

Based on those parameters, what we’re really looking at here are players who had great seasons on teams that were in a close playoff race.

If we’re going to ask the question above, human nature requires that we also ask the reverse question. Have any players caused their teams to slip below the 16th place cut line? That answer is also yes but far more infrequently than the opposite. That makes sense because as much as we in the public like to criticize front offices and coaches, they aren’t in the habit of overplaying individual skaters so much that they would single-handedly alter the course of the season. The opposite effect, playing the best players enough to carry the team to a much better finish in the standings, happens every season.

The following chart shows the reverse of the one above. The lines in orange are the ones where the player moved their team below the 16th place cut line.

While working on this idea, I floated this chart past a few people to get feedback and noted Sabres Twitter legend Kevin alerted me to the special case of Robyn Regehr. Not only does he appear here as a player who may have been the difference between the Sabres making and missing the playoffs in 2011-2012 but he has a singular moment captured on video from that season that perfectly represents his place on this chart.

In this April game against the Flyers, he gets individually burnt for a goal that pretty much cost Buffalo the playoffs.

Rarely does a singular video highlight line up so cleanly with a number like this. Apologies to Robyn Regehr but I can’t ignore serendipity like that.

## Wrap Up

The most common joke people like to make whenever anyone brings up WAR is to recite the lyrics from the old Edwin Starr song: War, what is it good for? Absolutely nothing! That lyric is certainly true in its original context but also holds some weight in the context of this silly article. WAR in hockey feels like a stat ported from another sport, which is exactly what it is. SPAR is the same stat tailored for hockey.

For that reason, I’ll be using it in places where I would have used WAR in the past. I consider this a best practice for hockey analysis and encourage others to consider doing the same. After all, when you want to know which NHL team is best, you don’t sort the standings by the win column. You sort it by the standings points column.

*Note: Thanks to **Luke and Josh from Evolving Hockey** for answering some questions while I was working on this post. Specifically, they reminded me that goalies exist, which I had forgotten and was causing some issues with the analysis. Also, thanks again to **Kevin** for the Regehr tip.*

## Comments ()