Saturday 24 December 2011

NHL Realignment, Part 4: To Wildcard or not to Wildcard

In the previous posts we've explored what the actual BOG approved plan will mean for each team in terms of their playoff chances.

In this instalment we'll have a look at one particular aspect of the approved plan, namely the re-seeding of teams for the semifinals. Now, I think that's a great idea, as it will create an added incentive to make it to the playoffs in a high a spot as possible (league wide).

However, in order to make that re-seeding "legit" there is an important admission that is made by the NHL. That admission is that the regular season is "good enough" to rank teams across the whole league. After all, any given two teams in the semifinals will share at least 58 games out of the 82 each season (2 games against each of the other 28 teams, plus the home and away against each other), or even 60, depending on how the chips fall.

So what does that have to do with wildcards? Well, if the regular season is "good enough" to seed the semifinalists then surely that would also apply to all the teams across the league, regardless of where they ended up. After all, that is how draft picks will be decided. So, given that we have re-seeding during the third round of the playoffs, I think the case could be made for 3 playoff spots per conference, plus 4 wildcards based on the regular season.

What would that mean for the playoff chances of each team? Such a move would first of all eliminate playoff bleed entirely. Strong 5th placed teams would make the playoffs in place of weaker 4th placed teams. It would also ensure that strong conferences are properly represented in the playoffs.

What would it mean for the playoff structure though? Well, we would no longer have the clean "in-conference" first and second rounds. We could still have the 2nd and 3rd placed teams face off in a kind of conference semifinal, but the 1st placed team could end up playing a team outside it's conference. If I had my way, I'd give the President's Trophy winners the privilege of choosing their opponent in the first round from the wildcards, the second best conference winner would get second pick, and so on.

The implication would be that a conference winners could face really long travel in the first round of the playoffs, depending on the choices those conference winners make. It might even be a benefit to finish second in your conference (if you know you'll finish as the worst placed conference winner) simply because you will probably have shorter travel that way, and the worst conference winner will likely be left with the best placed wildcard, which will be the best 4th placed team from one of the conferences, rather than the 4th placed team in your own conference.

Another implication is that we could potentially see conference rivals meeting in the Stanley Cup Final, since they might end up on different sides of the draw. Penguins vs Flyers in the final? Devils vs Rangers? Canucks vs Sharks?

In short, a wildcard system would mean that any two teams could face off for the Cup. It would give the President's Trophy winners a tool to influence the path of themselves and the other teams towards the final. It would likely be a pain for the worst placed conference winner, depending on how the wildcards fall in a given year. On the other hand, it would all but ensure that the 16 best teams (based on the regular season) make it to the playoffs.

So what do we want? The 16 best teams, or true conference showdowns in the first and second rounds?

To wildcard or not to wildcard?

Friday 16 December 2011

NHL Realignment, Part 3: Final thoughts

In Part 1 and Part 2 of our realignment series we looked at the theoretical gains and losses in terms of playoff spots, as well as the actual bleed of playoff spots between the conferences and odds for each team in the NHL. Take a look at those posts before proceeding. 

In this post we'll focus on some final thoughts about the new conference system.
The realignment map again.
Image by Tom Fulery

Winners and losers
Who are the winners then? Well, it depends on the time scale that you're looking at. Based on the last 5 seasons the major winners are Boston and Montreal who get a nice fat 20% increase in their playoff chances. Their new odds are in fact 100%., despite the fact that they haven't made the playoff each of the 5 last seasons.  So based on past performance I can already declare that Boston and Montreal will take 2 of the playoff spots from the Eastern conference.

Based on the last 10 seasons our big winners would probably be Dallas and Nashville, who despite being in an 8 team conference would still get a nice 9% bump up to 69%. Both should be looking to cash in on playoff spots next year.

Any losers? There sure are. Based on the past 5 seasons the biggest is Pittsburgh. They've been in 5 straight playoffs, so should be looking at a sure playoff spot. But, in the new Atlantic conference, their odds have taken a tumble down to 87%. Still pretty good, but no longer such a sure thing, specially with the Devils, Rangers, Flyers and Capitals hanging around, each with a 70% chance of one of the 4 playoff spots.

Based on the last 10 seasons the losers are pretty much any team in the Western conference, San Jose and Vancouver in particular. They each take about a 10% hit to their odds, down to 80% and 71% respectively.

Is Toronto one of the big winners? No, not really. They have a better chance than before, but still have to get past Boston, Montreal, Buffalo, Ottowa and Tampa Bay... all teams with better records the last 10 and 5 years.

Should Detroit want to move east? No, not really. They're in a cosy conference where they should be making the playoffs every year till global warming makes hockey impractical... They could move to the Eastern and still be nearly a sure thing in the playoffs, but a move to the Atlantic conference would be ill advised. If they did move into the Altantic, we'd have a hell of a conference totalling 53 playoff spots the last 10 seasons, but Pittsburgh and Detroit would both be looking at only a 71% chance of making the playoffs instead of being permanent fixtures (at least the last 5 years).

Final thoughts
Is the new conference setup the best one we could have come up with, given 4 conferences? It very nearly is. The only reasonable ones that I could find that yielded better numbers in terms of playoff "bleed" between conferences was to move Colorado into Central, and Columbus into Atlantic or Eastern (Eastern would be slightly better, numbers wise). Either move would reduce the bleed from 0.9 playoff spot each year to 0.8 or 0.6 playoff spots depending on where Columbus go. And either configuration would be safe against both a Phx-QC move and a Phx-KC move, i.e. the bleed would not get worse.

A Phoenix move you say?
Well... the math says that if they were to move to Quebec, then that would stop the bleed from the Western conference. We'd be down to a bleed of about 0.5 playoff spots across the whole league, down from 0.9. If they moved to Kansas City, then the best solution in terms of playoff bleed would be to push Columbus east, not Detroit. If Detroit were to ever leave the Central conference, that conference would become a playoff spot magnet, sucking in 1 whole playoff spot each year, from the other 3 conferences combined. The conference that Detroit moves to would then become the big bleeder.

The last thing I wanna do is talk about one potential problem with the new conference setup, namely which teams would not make the playoffs. If we look back at the last 5 season we can apply the new setup to the old divisional results. Now bear in mind that the game matrix is different, but still... Last year the LA Kings finished 12th in the NHL. They would not have made the playoffs, since Vancouver, San Jose, Phoenix and Anaheim would have been the 4 to go. Instead of the Kings we would have had the Dallas Stars into the playoffs. They finished 16th, so that's not too bad, but we'd also have the Rangers sneaking into the playoffs, much as they actually did last season. So we're trading 12th place for 18th. In 08-09 it would've been even worse. Carolina and the Rangers finished 11th and 12th that season, but neither would've made the new playoffs, since Washington, New Jersey, Pittsburgh and Philadelphia would have taken the 4 spots. Florida in 14th didn't make the playoffs, but would now, along with Anaheim and Buffalo in 17th and 18th. So effectively 11 and 12 wouldn't have made it, but 17 and 18 would have. That could be a concern, but the new game schedule should make this kind of situation really unlikely. If it were to happen consistently the perfect remedy would be to have 3 playoff spots per conference plus 4 wildcards. Looking back the last 10 seasons, such a system would have ensured that the top 16 make the playoffs every single year since 2000-01. Historically that would've been to the advantage of the two western conferences, particularly the new Western conference. If that system had been in effect we would have seen 48 playoff teams from the new Western conference in the last 10 years instead of "'just" 45.

Anyway, that's a whole lot of talk, and a whole lot of numbers. So I'll leave you all with this: The new 4 conference setup isn't all that bad, but we'll have to keep an eye on where the playoff spots go. It might all work out fine with the new game matrix. But if the conferences do stay unbalanced, a couple at 7 teams and a couple at 8, then yes, in time you will be more likely to make the playoffs from those conferences. But at the minute, previous performance, and the strength of the conference is a bigger factor. Hope you enjoyed all the statistics.

Wednesday 14 December 2011

NHL Realignment, Part 2: Playoff bleed

So in Part 1 we looked at the theoretical bleed of playoff spots between the conferences, based solely on the number of teams in each conference. Take a look at that post before going on.

In this post we'll take it a step further and look at the actual playoff bleed between conferences, as well as the odds of each team reaching the playoffs under the new conference system.
The realignment map.
Image by Tom Fulery

In practice
In theory, it would be true that the two 8 team conferences, Western and Central would bleed playoff spots to the two 7 team conferences, Eastern and Atlantic. However, in practice the actual bleed of playoff spots is dependent on the relative strength of the conferences, not the number of teams in those conferences.

One way to figure out the strength of the conferences is to use the number of times teams in a conference reached the playoffs the last 10 seasons (the time that we have had the current divisions). So we add up the playoff appearances of teams in each of the new conferences, generating the table below. What does this data tell us?


The relative strength of conferences and how playoff spots bleed
from some conferences to others (from- to +)
Conference Playoff spots
the last
10 seasons
Playoff spots
the last
5 seasons
Playoff bleed
based on
the last
10 seasons
Playoff bleed
based on
the last
5 seasons
Weighted
average
bleed
Atlantic 43 (-3) 23 (-3) -0.3 -0.6 -0.45
Eastern 36 (+4) 16 (+4) +0.4 +0.8 +0.6
Central 36 (+4) 19 (+1) +0.4 +0.2 +0.3
Western 45 (-5) 22 (-2) -0.5 -0.4 -0.45

Well, let's look at the Atlantic and Eastern conferences. Teams from the Atlantic conference totalled 43 playoff spots the last 10 seasons, 23 in the last 5. Teams from the Eastern conference totalled just 36 playoff spots the last 10 seasons, 16 in the last 5. With the new system both conferences will receive 40 playoff spots the next 10 seasons, 20 over the next 5. So the Atlantic conference will get 3 less than they got under the divisional system, while the Eastern conference will get 4 more than they managed under the divisional system. This will lead to a "bleed" of playoff spots from the Atlantic conference to the Eastern conference. The strength of this bleed is different depending on the time scale we look at (-3/10 and +4/10 over the last 10 seasons, or -3/5 and +4/5 over the last 5 seasons), but it averages to a bleed of 0.45 playoff spots per year from the Atlantic, while the Eastern will gain 0.6 playoff spots per year. The additional 0.15 is a bleed from the old western conference, which is caused by Atlanta's move to Winnipeg.

The old western conferences breaks up into the Central and Western conferences, and the story is repeated here, with an average bleed of -0.45 from the Western conference, with a gain of +0.3 for the Central conference. Winnipeg's move "steals" some of the gain that the Central conference would have gotten, reducing it by the same gain of 0.15 spots per year we saw above for the Eastern conference.


The weighted average is the average of the bleed based on the last 10 and the last 5 seasons. In other words, the last 5 seasons are counted twice and deemed more important than the 5 seasons preceding them.

So it turns out that it isn't the two 7 team conferences that get the sweet deal out of this realignment, it is the Eastern and Central conferences, because they are the "weaker" conferences based on the playoff records of the teams within those two conferences.

What about my team?
Well, we now know that if your team is in the Eastern or Central conference, they you're in luck. On average, you will see your favourite team more often then you would based on their performance under the old divisional system. If your team is in the Atlantic or Western conference, then tough luck, the odds of your team reaching the playoffs have just gone down, since it resides in a strong conference.

But what are the actual odds of a given team reaching the playoffs after the realignment? It depends on the strength of that team relative to the strength of the conference they are in. Luckily we already have all the data we need to figure this out, namely the number of playoff appearances for each team the last 10 seasons.

Let's look at Washington and Tampa Bay and see how the math works out.

Washington has 6 playoff appearances the last 10 seasons, 4 of them in the last 5 seasons. Based on that Washington should have a 60% or 80% chance of reaching the playoffs, depending on whether you look at the last 10 seasons, or limit it to the last 5 seasons. However, Washington is in the strong Atlantic conference with teams totalling 43 playoff appearances the last 10 seasons, 23 the last 5 seasons. Because there are more than 40 and 20 appearances your odds will be reduced from the expected 60% or 80%. The easiest way to visualise this is that 43 teams will now be squeezed into 40 spots, or 23 teams squeezed into 20 spots. This means that Washington's odds for the 2012-13 season are reduced from 60% to 56% (60*40/43=56) based on the last 10 seasons, or reduced from 80% to 70% (80*20/23=70) based on the last 5 seasons. This means a reduction by 4% or 10% depending on the time scale, as seen below.

Tampa Bay finds itself in the less competitive Eastern conference, and has 5 playoff appearances the last 10 seasons, 2 of them in the last 5 seasons. So Tampa Bay should have a 50% or 40% chance of reaching the playoffs. However, because the Eastern conference only totals 36 playoff spots the last 10 seasons, 16 in the last 5, we will see an increase in Tampa Bay's odds. Where we used to have 36 teams in the playoffs we will now get 40, where we used to have 16 we will now get 20, and this surplus of playoff spots accounts for the increasing odds. For the 2012-13 season Tampa Bay's odds are increased from 50% to 56% (50*40/36=56) based on the last 10 seasons, or increased from 40% to 50% (40*20/16=50) based on the last 5 seasons. This means an increase of 6% and 10% respectively, as seen below.

So browse the table below for the teams you are interested in, and comment if you have any questions.
The meaning of the table headers are as follows:
P10 = Playoff appearances the last 10 seasons.
P5 = Playoff appearances the last 5 seasons.
N%10 = Chance of making the playoffs with the new conference setup, based on the last 10 seasons.
N%5 = Chance of making the playoffs with the new conference setup, based on the last 5 seasons.
%C10 = Percentage points of change between the old divisional system and the new conference setup, based on the last 10 seasons.
%C5 = Percentage points of change between the old divisional system and the new conference setup, based on the last 5 seasons.

Atlantic P10 P5 N%10 N%5 %C10 %C5
Eastern P10 P5 N%10 N%5 %C10 %C5
Carolina 4 1 37% 17% -3% -3%
Boston 7 4 78% 100% +8% +20%
New Jersey 9 4 84% 70% -6% -10%
Buffalo 5 3 56% 75% +6% +15%
NY Islanders 4 1 37% 17% -3% -3%
Florida 0 0 0% 0% 0% 0%
NY Rangers 5 4 47% 70% -3% -10%
Montreal 7 4 78% 100% +8% +20%
Philadelphia 9 4 84% 70% -6% -10%
Ottowa 8 3 89% 75% +9% +15%
Pittsburgh 6 5 56% 87% -4% -13%
Tampa Bay 5 2 56% 50% +6% +10%
Washington 6 4 56% 70% -4% -10%
Toronto 4 0 44% 0% +4% 0%
SUM 43 23




SUM 36 16




Chance of your team making the playoffs under the new conference setup, found in the fourth and fifth colums.
Central P10 P5 N%10 N%5 %C10 %C5
Western P10 P5 N%10 N%5 %C10 %C5
Chicago4 3 46% 64% +6% +4%
Anaheim 6 4 53% 73% -7% -7%
Columbus 1 1 12% 21% +2% +1%
Calgary 5 3 44% 55% -6% -5%
Dallas 6 2 69% 43% +9% +3%
Colorado 7 2 62% 36% -8% -4%
Detroit 10 5 100% 100% 0% 0%
Edmonton 3 0 27% 0% -3% 0%
Minnesota 3 2 35% 43% +5% +3%
Los Angeles 4 2 36% 36% -4% -4%
Nashville 6 4 69% 86% +9% +6%
Phoenix 3 2 27% 36% -3% -4%
Saint Louis5 1 58% 21% +8% +1%
San Jose 9 5 80% 91% -10% -9%
Winnipeg 1 1 12% 21% +2% +1%
Vancouver 8 4 71% 73% -9% -7%
SUM 36 19




SUM 45 22




So there you have it folks, the playoff chances of each team under the new conference system. This is all based on the past performance of the teams under the old divisional system, so we need to issue a  disclaimer. One is that the game matrix will now change. That will influence the ability of teams in the same conference to collect points, since they play a lot of the games against their conference rivals.

One other note: We could have used more sensitive data to generate the tables above. When I get the time I intend to the position that each team finished the regular season in, rather then just a binary "did/did not qualify for the playoffs". That in turn will allow us to look at conference strength in a new light and hence the chances for each team in those conferences. The playoff bleed will stay the same however.


In the next post, Part 3, we'll discuss who the winners and losers are with the implementation of this realignment, and sum up with some final thoughts.

Monday 12 December 2011

NHL Realignment, Part 1: The new map

The new map
Check out Part 2 for the real statistical meat where we calculate playoff bleed between conferences. In this post we'll go through the theoretical odds and some other preliminary stuff.


So, the mighty BOG has voted on the next format for the NHL, and it looks like we'll be seeing the much discussed 4 conference layout for the 2012-13 NHL season. It's an exciting new twist to the game, and one that I think fans will enjoy. But already some fans have pointed out that it looks unfair that some teams (the old west) have to compete for 4 playoff spots in 8 team conferences, while others (the old east) only have 7 team conferences.

So with that in mind, I thought we should explore what the new conferences actually mean in terms of playoff chances for the various teams. Below you can see a graphic representation of the new conferences, as seen on the Tom Fulery NHL Realignment Project. The long list of proposed realignments posted over there is well worth a look. While the names of the new conferences hasn't been confirmed yet, for the next few posts, we will be referring to the conferences with names indicated below.

Image by Tom Fulery

The theory
So with two conferences having 7 teams and the other two conferences having 8 teams it seems obvious that it'll be easier to reach the playoffs from a 7 team conference. In theory, that is correct. With the old divisional system, any teams odds of reaching the playoffs were 8/15, or 53.3%. In the new conference system, teams in the Western and Central conferences will have to contend with a 50-50 chance (4/8) of a playoff spot, while teams in the Eastern and Atlantic conferences will enjoy a 57.1% chance (4/7) of reaching the playoffs. 

All teams in the old west (you too Winnipeg) take a 3.3% cut to their playoff hopes. What this means is that the 8 teams in the Western conference combine to "bleed" 0.27 playoff spots each year (0.033*8). Or put a different way, the Western conference "bleeds" 1 playoff spot every 3.75 years (1/0.27). The same goes for the new Central conference.

This "bleed" results in the Eastern and Atlantic conferences "gaining" 1 playoff spot each every 3.75 years. That's the theory anyway.

However, it's important to remember that this bleed is only relative to the old division system, and the new game matrix is quite different. With the new game matrix this bleed might be lessened or it could get worse. Essentially, strong conferences will bleed playoff spots to weaker conferences now that everyone sends 4 teams.... relative to the divisional system. I'd also like to remind everyone that this bleed was present in the old divisional system too, in the form of a weak 8th team from the East or West taking the spot of a 9th or even 10th placed team from the other conference. Looking back the last 10 seasons this bleed has been from the old west to the old east at a rate of 0.4 spots per year.

Also, I've heard the concern that: "Omg, this means we could see the 27th best team in the league make the playoffs!" While that is true, I wanna point out that with the old divisional system, you could in theory see the 26th best team in the league make the playoffs... so stop complaining.


In the next post, Part 2, we'll have a look at the actual playoff chances for each team in the league, and the actual bleed of playoff spots between conferences.