Panic on the Streets of HawkeyTown: A Rational Response

Following the NHL playoff race can be confusing because teams have "games in hand" and the rankings are by points instead of winning percentage as in most other sports. And while sorting the standings by Points Percentage (PPct) is a good way to help clarify things, it's definitely not the best predictor of future standings.

The more I hear people try to tell me the season is over, the more I want to show that statistics prove otherwise. I also simply refuse to believe that a team like Tampa Bay with a -5 Goal Differential can, over time, continue to win at a .649 PPct pace, while Boston who has a +36 GDiff has only a .605 PPct. Or that the Blackhawks who have a +21 GDiff should end the season tied with Columbus who has a -17 GDiff.

Over the long term can a team that consistently gives up more goals than it scores win a large majority of its games? Or vice versa? In my opinion, there is no stronger determinate of a team's strength and long term success than Goal Differential (GDiff) tempered by Strength of Schedule (SOS). So that brought me back again to one of my favorite stats, the Simple Rating System (SRS).

SRS = GDiff + SOS. It seems almost too simple to be a powerful analytical tool, but the SOS calculation is actually rather labor intensive. I have already shown in the past that SRS is a good determinate as to whether a team makes the playoffs (87% of teams with a positive SRS get in since the lockout) and I have suggested that teams with a strong SRS rating do well in the playoffs, but is there a strong correlation between SRS and a team's actual record?

In order to determine this, I took the records of all 30 NHL teams over the past 5 years and sorted the records based on SRS rating. I then broke that down into groups and charted the average SRS of each group against the average PPct of that group. Here is the expanded data showing the ridiculous amount of info that went into this study.

Here are the condensed groupings showing the average record, Pts, PPCT, SRS, etc, for each group:


And here is that data graphed out:


I had expected a relationship between SRS and PPct, but this was unexpectedly close - not just Elvis' "Kissing Cousins" close, more like The Doors' "The End" close. All the data points fell within a 5% margin of error of the linear trend line. And of the 150 individual (not grouped) season records analyzed, only 17 fell outside the margin of error. So on average about 3 or 4 teams per season fell outside the margin of error. Something interesting that I also noticed was that 15 of those 17 teams that over- or under-performed had unusually high or low SOS’s. The teams that overperformed based on their SRS had a weaker than average SOS. Those that underperformed had a stronger SOS than average. I’ll touch more on this in my next FanPost regarding SRS and playoff performance.

Here are the predicted standings based on each team's current PPct. Again, I don't believe this is a very accurate forecasting method. The teams highlighted in a lighter color are the ones that are significantly under- or over-performing based on their SRS. The projected PTS are in the Pts82 column. The rest of the data is based on the standings as of 2/17:


Next, I charted these projected records based on PPct against each team's SRS and overlaid that on my previous chart. Since I'm looking at a much smaller data set (1,722 games played so far in 2011 vs 12,300 games making up the 5 Yr analysis), I expected the results to be a bit more scattered. But based on my 5 Yr averages, if PPct is a good forecaster then I should only see 3 or 4, maybe even 5, teams significantly over- or under-perform. Instead I found 7. There's no rule that states 7 teams can't end up outside the margin of error, but in a typical season that number should be halved.

Here is that graph:


The area above and below the trend line should be about the same. There are 18 teams over the line and 12 below. Most of the overperforming teams are very close to the trend line though except for one or two. There are fewer teams below the trend line but 6 teams are underperforming by a wider margin, so that ends up equaling things out.

Forecasting the season finish by PPct isn't realistic though. Using that method shows static results; no team makes any moves in the standings and the result predicts that teams that have been beating the odds will continue to do so over the remaining 30% of the season. It's like flipping a coin 1,000 times. The chances of the coin landing on heads is 50%. Yet if you flip the coin 70 times, it might come up heads 45 times and tails only 25 times. To believe it will keep coming up heads 64% of the time in the future is irrational. The longer you flip it, the closer it will come back to statistical norms.

SRS and a team's record behave in the same way. My chart proves that if you know a team's SRS, you can accurately predict their record if you have a large enough sample size. In the short term, a team may beat the odds, but in the long run statistical norms will prevail. The problem is, and the part that keeps us on the edge of our seats, is that we don't know when that will occur. The law of averages does not know an 82 game season. It could take 100 or 1,000 games to fall in line with the norm we've established. We can assume it will eventually happen, but we don't know when.

One humorous observation from the graph above: Everyone (except apparently the Hawks) knows the Oilers suck, but according to the chart the Oilers are even underperforming against their level of suckitude. And you thought the Hawks had it bad.

So let's look at the Hawks. They have an SRS of +0.39. That translates to a PPct of .632 and a record of 33-18-6 and 72 pts after 57 games and about 104 pts at season's end. Instead, the Hawks are 29-22-6 for 64 pts. The variance from the norm simply comes down to bad luck. You can disagree with the concept of luck, but you can't successfully refute it when looking at a large sample size. You can suggest that the Hawks don't have the fight that they did last season, but that is completely a subjective statement. They have an established SRS performance level that didn't happen by mistake. They may have lost a few games because they "collapsed," but they also won some games where the competition's fans also feel their team collapsed, not because the Hawks were "clutch." Why does one team score 7 soft goals in one game but have 20 scoring chances in the next, yet get shutout? It comes down to luck.

We can't change the past, but through SRS we can suggest what might happen in the future. I took the current PPct of each team and instead projected the final standings based on the established PPct per SRS for each team (SRS PPCT). So the calculation for the Hawks looks like this:

Final Pts = Current Pts + ((82 Games - Games Played) x 2 x SRS PPct) = 64 + ((82 - 57) x 2 x .632) = 64 + 32 = 96 pts

Here is that graph:


As you can see the disbursement is now tighter. In reality, based on past seasons, it's more likely that 3 of the 6 underperforming teams will overperform based on their current SRS PPct and get back within the margin of error. That would leave about 4 teams outside the margin of error. But which ones? The Hawks have a current SRS of .39 and are right on the cusp of the stronger .41-.60 grouping. Therefore they have the highest probably of being one of the teams to overperform over the remaining 30% of the season. In addition, teams like Minnesota, Calgary, and Dallas are all outperforming their SRS PPct, especially over the past 10 games. If their SRS ratings were to remain the same, you can definitely expect a cool down from a number of these teams.

You also can't expect a team to get hot NOW just because you think it should. There are quite a few hot teams in the West right now and logically you need to have just as many cold teams. This will change. Cold teams will get hot, hot teams will cool down. That's not wishful thinking; it's based on established trends. We just don't know when it will happen - or if the season will run out before it does.

Below are the revised standings based on the graph above. There are no changes in who makes the playoffs in the East, but in the West the Hawks would be in and the Stars out. I could have tried to predict which teams should get hot based on their placement in each SRS group which suggests the Hawks are the most likely team to overperform, but I wanted to use the most conservative approach - and even that conservative approach states that "All is not fucked here." No matter what though, the race will be very tight and come down to the final games.

Projected standings based on established SRS PPct (again, focus on the Pts82 column):

Again, I used the most conservative approach here. Because the Hawks are so close to jumping into the next highest grouping, it might be more likely that they end up closer to 97-99 pts rather than 96. Likewise in the East, it is extremely likely that the Lightening finish with around 100 pts instead of between 102 and 106. Boston would then also move up due to Tampa Bay's drop off.

And that folks is the most rational response I can give to the cliff jumpers and doomsday sayers. It's not rose-colored; it's actually very black and white with a touch of gray. When statistics show that the probably of the Hawks making the playoffs are slim, I will gladly write that FanPost. But until that time I will allow rational thought and established trends to assist in making my decisions.

FYI, I started writing this post a few days ago, so this wasn't in reaction to any particular recent comments or even Sam's post game summary. Based on SRS, I just happened to know there was a 78% chance of cliff jumping and dead horse beating after yesterday's game. And I have graphs to prove it.

Log In Sign Up

Log In Sign Up

Please choose a new SB Nation username and password

As part of the new SB Nation launch, prior users will need to choose a permanent username, along with a new password.

Your username will be used to login to SB Nation going forward.

I already have a Vox Media account!

Verify Vox Media account

Please login to your Vox Media account. This account will be linked to your previously existing Eater account.

Please choose a new SB Nation username and password

As part of the new SB Nation launch, prior MT authors will need to choose a new username and password.

Your username will be used to login to SB Nation going forward.

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

By becoming a registered user, you are also agreeing to our Terms and confirming that you have read our Privacy Policy.

Join Second City Hockey

You must be a member of Second City Hockey to participate.

We have our own Community Guidelines at Second City Hockey. You should read them.

Join Second City Hockey

You must be a member of Second City Hockey to participate.

We have our own Community Guidelines at Second City Hockey. You should read them.




Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.