I want to step away from lockout or contract talk and simply discuss a concept in hockey I like.

When analyzing teams throughout the season, I often like to look at “goal differential” rather than points pace. I think the quality of a team is reflected in how many goals they score, and how many they give up. Over a large span of games, the amount of goals and goals given up can predict how many wins a team will earn.

This is mostly a baseball concept, but it has been applied to other sports. Primarily basketball, but the MLB.com standings page will allow you to look at predicted wins and losses based on the formula, which is as follows:

In baseball, they score runs. In hockey, you score goals (unless you’re Scott Gomez. Hey ho!). Using the same formula, let’s look at, say, the Anaheim Ducks over the last three seasons. They’ve won 128 out of 246 games, for a winning percentage of .520. They’ve doing this scoring 722 goals including shootout wins, and allowing 724 including shootout losses.

722 ^ 2 / ( ( 722 ^ 2 ) + ( 724 ^ 2 ) ) = .499

So it isn’t too far off. A winning percentage of .499 would equal roughly 123 wins, which is just five below the predicted total.

In December, back when everybody was complaining that Minnesota wouldn’t continue their success, one of the reasons was because they had a goal differential of pretty close to even despite being a handful of games above .500. This meant they were winning a disproportionate number of one-goal games.

Winning one-goal games is a talent, but it’s one that isn’t too separate from winning two-goal or three-goal games.

One of the reasons I was looking at this is because of The Hockey News’ predictions. The Edmonton Oilers and Montreal are slated to finish in 13th in their respective conferences this upcoming season. Why?

Well, for Edmonton:

We do believe the Oilers will improve from recent showings and maybe even move out of the draft lottery. Nail Yakupov is dynamic, but first-year expectations should be tempered. Defense and goaltending is still a big question mark and in a loaded Western Conference those are debilitating shortfalls.

And for Montreal:

While Montreal found a wonderful line of Erik Cole, Max Pacioretty and David Desharnais last season, the team still doesn’t score enough goals. They added grit in the form of Brandon Prust and Colby Armstrong that will beef up an undersized forward corps, but that doesn’t address the need for offense.

The funny thing with Montreal is that while they can stand to improve, they were 14th in the league in defence last season and 19th in offence. Sure, that isn’t amazing, but that doesn’t equal up being 29th in wins without some serious bad luck.

For Edmonton, they were 28th in the league in wins, but 19th in offence (tied with the ‘Habs at 212 goals) and 23rd in defence. I’m not saying they should compete for a playoff spot anytime soon, but to figure out exactly how much of a push they need, we need of figure out how far back they are.

Here is a table of expected wins based on goals for and against last season, along with a table of expected points, which just gives teams a bonus 10 points for the average number of OT losses a team sustains in a season:

Exp. W Wins Exp. Pts
Boston 52 49 115
Pittsburgh 51 51 112
St. Louis 51 49 111
Vancouver 50 51 110
Detroit 49 48 108
NY Rangers 49 51 107
Philadelphia 46 47 103
Nashville 46 48 102
New Jersey 45 48 99
San Jose 44 43 99
Los Angeles 44 40 99
Phoenix 43 42 97
Chicago 43 45 95
Ottawa 43 41 95
Washington 40 42 89
Dallas 39 42 88
Buffalo 39 39 88
Colorado 39 41 87
Montreal 38 31 87
Winnipeg 37 37 85
Florida 36 38 83
Calgary 36 37 83
Edmonton 36 32 82
Anaheim 36 34 82
Carolina 36 33 81
Toronto 36 35 81
Tampa Bay 34 38 77
NY Islanders 32 34 74
Minnesota 31 35 72
Columbus 31 29 71

This chart has Buffalo making the playoffs barely, but Florida falling out, being actually the third best team in the division rather than the best due to fewer OT losses and a -24 goal differential. The Canadiens actually have less of a hurdle than expected, with a similar goal differential to the Panthers. Imagine the expectations in Montreal if the coin fell the other way last season.

Tampa Bay finished about four wins higher than predicted, so their road back to the playoffs is a little bit longer. Ditto Minnesota. Carolina ought to have won three more games than they did last year, so maybe their road of about 10 points (or five wins) is a bit shorter.

You never know the kind of bounces you’ll get in one-goal games, but those are critical in determining some of the playoff races. Once you get into those playoff races, you can throw the stats out the window and play to win, of course, but when assessing your team’s quality in the offseason, I wonder if it may be wiser to pay attention to predicted wins rather than simply placement in the standings.

Comments (8)

  1. Someone recently published a paper on the Pythagorean W-L formula as it pertains to hockey. It’s pretty math-y, but pretty cool:


  2. Doesn’t this ignore personnel/philosophy changes though?

  3. On the Wild thing. It should be noted that on December 15th, the Wild were one of the top teams in the conference in Goal differential with a +13(Only Detroit and Vancouver had a better GD in the West). It also should be noted that December 14th was the day Koivu went out(PMB, Seto and Latendresse went out in the previous 3 or 4 games). 15 Short days later at the end of December, the Wild had an even Goal Differential.

    More evidence that the Wild’s injuries, more than anything else led to their downfall, as they were doing very well in goal differential prior to the injuries. It was only after that they couldnt score(who could missing 4 of the top 6 forwards).

    • Ha, noticed that after I posted. It was pretty grim through November, however. I expect a 7-game winning streak would put a team in plus-territory if we’re early in the season.

    • A team has to be accountable for carrying injury prone players though. I don’t think Wild fans can complain when PMB and Latendresse go down, because we all know it’ll happen eventually.

      As a Canucks fan, I always knew Salo wasn’t good for more than 60 games a year, but management usually acquired the depth to accommodate his fragility

  4. Did you remove empty-netters? Teams that tend to get narrow wins will also tend get a disproportionate number of empty-net goals for, and the same for narrow losses and empty-net goals against.

    Montreal and Edmonton were both -5 on empty-netters. Meanwhile, Florida was +4. That makes the swing between those clubs look even bigger; omitting empty-netters, Montreal’s net GD was -2, a full 21 goals better than Florida’s.

Leave a Reply

Your email address will not be published. Required fields are marked *