Probability of a repeat champion each year from 99-00 through 15-16

[SnowblindNYR]

So with the Lightning becoming just the second team to repeat as champions since the 97-98 Detroit Red Wings, I thought about the gap between those Red Wing teams and the Penguins teams in 15-16 and 16-17. So I wanted to quantify every defending champ's probability of repeating. I originally wanted to quantify it in the aggregate of none of them repeating in 16 seasons, but then realized the math wouldn't work. So I have it each year. You'll notice a wide range of probabilities, that's because I treated teams winning rounds as 100% for that round. So the 99-00 Stars that won 3 rounds and lost in the Stanley Cup Final, their probability of winning the cup was just the probability of beating the 99-00 Devils. It wouldn't make sense saying that their probability of winning the cup was say 10%, knowing that they won the first 3 rounds. So a teams' probability increases from not making the playoffs (0%) to winning 3 rounds (prob of winning the Stanley Cup matchup) and everything in between. It's the probability they win the cup given the rounds they already won. For the final, I just chose the actual finalist since that's set in stone. I do realize this is a hybrid method of what actually happened and a forecast, so it's not completely perfect and a bit subjective.

My methodology was using the below formula on regulation wins and losses at home and away (there is no 3 on 3 or 4 on 4 OT or shootout in the playoffs):

The rest was using probability and combinatorics to determine the odds of a 4 game sweep, 5 game win, 6 game win, and 7 game win. And I modeled out each of the possible scenarios. The math seems to work because if you choose a specific season in a specific conference and add up the probability of all of the seeds in that conference winning 3 rounds (or winning the conference), that probability comes out to 100%.

Here's my final result:

Here's my blog explaining my methodology in more detail:

Probability of Defending Champs repeating as Cup winners from 99-00 through 15-16

Please be kind because this was a ton of work!

Click to read more...

 

[SnowblindNYR]

BTW, I wish this forum was more active, not enough of us nerds on HF!

 

[hamzarocks]

So with the Lightning becoming just the second team to repeat as champions since the 97-98 Detroit Red Wings, I thought about the gap between those Red Wing teams and the Penguins teams in 15-16 and 16-17. So I wanted to quantify every defending champ's probability of repeating. I originally wanted to quantify it in the aggregate of none of them repeating in 16 seasons, but then realized the math wouldn't work. So I have it each year. You'll notice a wide range of probabilities, that's because I treated teams winning rounds as 100% for that round. So the 99-00 Stars that won 3 rounds and lost in the Stanley Cup Final, their probability of winning the cup was just the probability of beating the 99-00 Devils. It wouldn't make sense saying that their probability of winning the cup was say 10%, knowing that they won the first 3 rounds. So a teams' probability increases from not making the playoffs (0%) to winning 3 rounds (prob of winning the Stanley Cup matchup) and everything in between. It's the probability they win the cup given the rounds they already won. For the final, I just chose the actual finalist since that's set in stone. I do realize this is a hybrid method of what actually happened and a forecast, so it's not completely perfect and a bit subjective.

My methodology was using the below formula on regulation wins and losses at home and away (there is no 3 on 3 or 4 on 4 OT or shootout in the playoffs):

The rest was using probability and combinatorics to determine the odds of a 4 game sweep, 5 game win, 6 game win, and 7 game win. And I modeled out each of the possible scenarios. The math seems to work because if you choose a specific season in a specific conference and add up the probability of all of the seeds in that conference winning 3 rounds (or winning the conference), that probability comes out to 100%.

Here's my final result:

View attachment 453641

Here's my blog explaining my methodology in more detail:

Probability of Defending Champs repeating as Cup winners from 99-00 through 15-16

Please be kind because this was a ton of work!

Interesting article. Never thought I'd be interested in the use of probability/combination theory in hockey. Learned this stuff mostly in grade 12 and first-year university but never thought to apply it to the NHL.

Your article was well written, and you explained your rationale and assumptions well.

 

[SnowblindNYR]

Interesting article. Never thought I'd be interested in the use of probability/combination theory in hockey. Learned this stuff mostly in grade 12 and first-year university but never thought to apply it to the NHL.

Your article was well written, and you explained your rationale and assumptions well.

Thank you that means so much to me! I spent a week on it, but it was like all I did other than work. It was exhausting. I think there's limited utility because I'm kind of connecting two methods. The actuals (if you won a round that counts as 100%) and forecast. So I don't know if I can do that. When I'm ready to do another one (might be a while), I'll create a model that gives the probability of each team winning the cup. It'll be the same as this except I'll model out the new playoff format (already started this) and will have to add a model for the Final instead of using the team we know that made it.

It's funny when I studied for the GMAT I found probability and combinatorics to be the hardest part. But I was really interested in and learned it. I still find myself tripped up by it at times.

 

[JaegerDice]

Well this certainly didn't make me feel any better about the Blackhawks losing to the Kings in the 2014 WCF....

 

[SnowblindNYR]

Well this certainly didn't make me feel any better about the Blackhawks losing to the Kings in the 2014 WCF....

So most people probably think that the Hawks were extremely close to getting to the cup final and then winning the cup because they lost in game 7 OT. But if it makes you feel better I used the probability that they'll beat the Kings in a series prior to the series starting. That probability was 76%, once it got into OT of game 7 that probability dropped to around 50% (unless you want to bump up the probability due to home ice but I don't think home ice matters in that type of situation). With it being 50% the probability of winning the cup drops to 37%.

Prob of beating the Kings before series began = 76%
Prob of beating the Rangers = 74%
Prob of winning in game 7 OT = 50%

So 76%*74%=57%

vs.

50%*74%=37%

Edit: Although, I can't quantify it one COULD argue that 76% over 4-7 games a different story than one OT in one game. Because you get probabilities of teams making the playoffs very early that gets much more reliable the closer you get to the final game. I wish I knew how to calculate that difference between 7 games and OT.

 

[Hockey Outsider]

So most people probably think that the Hawks were extremely close to getting to the cup final and then winning the cup because they lost in game 7 OT. But if it makes you feel better I used the probability that they'll beat the Kings in a series prior to the series starting. That probability was 76%, once it got into OT of game 7 that probability dropped to around 50% (unless you want to bump up the probability due to home ice but I don't think home ice matters in that type of situation). With it being 50% the probability of winning the cup drops to 37%.

Prob of beating the Kings before series began = 76%
Prob of beating the Rangers = 74%
Prob of winning in game 7 OT = 50%

So 76%*74%=57%

vs.

50%*74%=37%

Edit: Although, I can't quantify it one COULD argue that 76% over 4-7 games a different story than one OT in one game. Because you get probabilities of teams making the playoffs very early that gets much more reliable the closer you get to the final game. I wish I knew how to calculate that difference between 7 games and OT.

Thanks for posting this - interesting article.

I agree that the probability in OT between two close teams should be roughly 50/50. It's essentially a coinflip. You can use some statistical models to get a more accurate estimate, but it's a lot of work to move from 50/50 maybe something like 52/48. (Of course, it's a different story if we're talking about overtime between the 1976 Habs and Capitals).

The probability of the Blackhawks winning the series, at 76%, seems quite high. The two teams finished with an identical number of points. I know Chicago had somewhat better GF/GA data (a ratio of 1.21:1 vs 1.18:1 for the Kings), and they had home ice advance, but that seems pretty one-sided for what was generally expected to be a close series, no?

 

[SnowblindNYR]

Thanks for posting this - interesting article.

I agree that the probability in OT between two close teams should be roughly 50/50. It's essentially a coinflip. You can use some statistical models to get a more accurate estimate, but it's a lot of work to move from 50/50 maybe something like 52/48. (Of course, it's a different story if we're talking about overtime between the 1976 Habs and Capitals).

The probability of the Blackhawks winning the series, at 76%, seems quite high. The two teams finished with an identical number of points. I know Chicago had somewhat better GF/GA data (a ratio of 1.21:1 vs 1.18:1 for the Kings), and they had home ice advance, but that seems pretty one-sided for what was generally expected to be a close series, no?

In 2013-14 the Blackhawks finished with 107 points and the Kings with 100 points.

2013-2014 NHL Hockey Standings

The Kings also finished with 12 OT/SO wins and Hawks finished with 7. I only used regulation games because there's no 3 on 3, 4 on 4, or shootouts in the playoffs.

 

[SnowblindNYR]

I think the perception of them being even teams stems from them winning a bunch of cups since 2010 but obviously that's not something that would be reflected by analysis because how do you quantify that?