The last 35 NBA Champions’ Probability of Winning (based on series-by-series betting odds)

[lessthanjake]

I recently discovered that Basketball Reference has historical betting odds for each playoff series, going back to 1989. See here: https://www.basketball-reference.com/playoffs/series.html

Given this, I thought it might be interesting to look at the implied probability each champion had of winning all four playoff series, based on those betting odds. Obviously there’s caveats that betting odds are just contemporaneous estimations (that slightly overestimate both teams’ chances, so the oddsmaker makes money), and also that injuries and other things can happen during a series, which betting odds wouldn’t account for. But it still seemed like an interesting exercise, which generally gets at how big a surprise each champion was contemporaneously.

Just as a quick estimation of the methodology, here’s how the calculations work: First of all, let’s look at how to get implied probability for one series. If the odds are a + number, then you add 100 to that number and divide 100 by the resulting number. So, for instance, if a team has +250 odds, you take 100/(250+100) = 0.286. So that’d imply a 28.6% chance of winning. If the odds are a - number, then you divide that number by itself plus 100. For example, if a team has -200 odds, you take 200/(200+100) = 0.667. That implies a 66.7% chance of winning. Now that we know how to determine the implied odds for a single series, how am I getting the implied odds of winning all four series? Well, that’s just by multiplying the probability for all four series. So if betting odds give a team implied odds of 90%, 80%, 70%, and 60% in the four rounds, then the implied probability of winning the title would be 0.9*0.8*0.7*0.6 = 0.302. So that’s a 30.2% overall probability of winning the title, based on series-by-series betting odds.

So what was each of the last 35 champions’ implied probability of winning the title, based on series-by-series betting odds?

NBA Champions’ Implied Probability of Winning Title, since 1989 (using series-by-series betting odds)

1. 1997 Bulls: 73.5%
2. 1996 Bulls: 70.5%
3. 2017 Warriors: 65.5%
4. 2000 Lakers: 62.6%
5. 2013 Heat: 59.3%
6. 1992 Bulls: 59.0%
7. 2015 Warriors: 54.4%
8. 1990 Pistons: 51.0%
9. 2018 Warriors: 50.1%
10. 2020 Lakers: 47.1%
11. 2009 Lakers: 45.9%
12. 2002 Lakers: 45.7%
13. 1998 Bulls: 43.6%
14. 1999 Spurs: 38.7%
15. 1991 Bulls: 37.4%
16. 2010 Lakers: 36.4%
17. 1989 Pistons: 36.1%
18. 2001 Lakers: 35.3%
19. 2005 Spurs: 33.8%
20. 1993 Bulls: 32.6%
21. 2014 Spurs: 31.1%
22. 2012 Heat: 26.7%
23. 2016 Cavaliers: 26.1%
24. 2007 Spurs: 25.6%
25. 2003 Spurs: 24.5%
26. 2023 Nuggets: 21.6%
27. 2022 Warriors: 21.4%
28. 1994 Rockets: 20.0%
29. 2008 Celtics: 19.7%
30. 2021 Bucks: 9.4%
31. 2006 Heat: 8.4%
32. 2019 Raptors: 6.3%
33. 2011 Mavericks: 4.9%
34. 2004 Pistons: 4.4%
35. 1995 Rockets: 1.3%

Anyways, I am not presenting any particular conclusion here. I know there are people who like to exult the star player on the types of teams near the bottom of this list, with the idea being that they rose in the playoffs and carried a team to the title. There’s some validity to that. But, of course, the betting odds we’re talking about here are internalizing how good people think the team is, including the best player on the team. So the teams with the highest probability here have the highest probability in significant part because the greatness of their best players were priced into the odds. Which is to say that I don’t think this list really exposes any particular team or player. It’s more just interesting information. And, as someone who followed basketball throughout this time period, it is generally consistent with my recollection of how surprised I was by each champion, bearing in mind an assessment of the team’s quality and the quality of their opponents.

[zimpy27]

The asterisk Lakers in 2020 were as favored to win as the KD warriors in 2018.

Can we put the bubble ring to rest now knowing that outsized favourite won?

[ChipotleWest]

2011 Mavericks 4.9%

But many Lebron fans using hindsight try to claim Mavs should have been the favorites in those Finals. May as well say everyone who won should have been the favorites before the series starts.

[ChipotleWest]

The asterisk Lakers in 2020 were as favored to win as the KD warriors in 2018.

Can we put the bubble ring to rest now knowing that outsized favourite won?

Nope, doesn't change the fact it was still played in the bubble and players didn't want to be there and some complained that they couldn't sleep.

[zimpy27]

The asterisk Lakers in 2020 were as favored to win as the KD warriors in 2018.

Can we put the bubble ring to rest now knowing that outsized favourite won?

Nope, doesn't change the fact it was still played in the bubble and players didn't want to be there and some complained that they couldn't sleep.

They couldn't sleep the poor babies.

I mean sure but everyone was under the same conditions

[ChipotleWest]

The asterisk Lakers in 2020 were as favored to win as the KD warriors in 2018.

Can we put the bubble ring to rest now knowing that outsized favourite won?

Nope, doesn't change the fact it was still played in the bubble and players didn't want to be there and some complained that they couldn't sleep.

They couldn't sleep the poor babies.

I mean sure but everyone was under the same conditions

Whatever you try it's not going to change the fact it was a bubble ring, wasn't even a crowd. The season should have been cancelled I believe it was Lebron that insisted they play. Most people don't take that ring serious, but you're free to interpret it however you'd like.

[zimpy27]

Nope, doesn't change the fact it was still played in the bubble and players didn't want to be there and some complained that they couldn't sleep.

They couldn't sleep the poor babies.

I mean sure but everyone was under the same conditions

Whatever you try it's not going to change the fact it was a bubble ring, wasn't even a crowd. The season should have been cancelled I believe it was Lebron that insisted they play. Most people don't take that ring serious, but you're free to interpret it however you'd like.

All you argued was that the conditions were unusually difficult. Under these difficult conditions the expected winner actually won.

I'm not arguing with you, you're agreeing with facts. You're just choosing to ignore the achievement beyond the facts

[ChipotleWest]

In a regular 82 game season, Lebron has two rings. GOAT?

[PistolPeteJR]

Nope, doesn't change the fact it was still played in the bubble and players didn't want to be there and some complained that they couldn't sleep.

They couldn't sleep the poor babies.

I mean sure but everyone was under the same conditions

Whatever you try it's not going to change the fact it was a bubble ring, wasn't even a crowd. The season should have been cancelled I believe it was Lebron that insisted they play. Most people don't take that ring serious, but you're free to interpret it however you'd like.

Lol, the Lakers were the biggest losers of no crowd. And actually the Lakers were for sitting out. You can find that if you search online.

[rapstarter]

The asterisk Lakers in 2020 were as favored to win as the KD warriors in 2018.

Can we put the bubble ring to rest now knowing that outsized favourite won?

Two things: 1) that the favorites won doesn't disprove the oddness of the bubble, and frankly I don't know how they are related at all. It's clearly evident that non-bubble seasons had winners of all ranges of betting odds, and the bubble season having a favorite win doesn't prove or disprove anything in that sense. 2) this methodology is highly dependent on the matchups that ONE team faced and doesn't always show who the actual favorites for the title were (e.g., see Boston always having great betting odds largely due to the EC being weak, and things flipping once the WC champion is decided).

In 2020, other outsized favorites (including the Bucks) all getting knocked out by a very mediocre Miami team changed the math completely. Even in the west, Clippers (-1300) losing to Denver was one of the biggest betting odds upsets ever. One could argue (and I'm not arguing that here) that was the bubble's doing.

OP is an interesting exercise, and quite valuable when ONLY looking at the wining team's progression, but it's flawed in telling the story of the entire playoffs. For reasons above, I think it's even more flawed if trying to make a statement about the bubble (especially given the big and uncommon upsets that did happen).

[Jasen777]

I recently discovered that Basketball Reference has historical betting odds for each playoff series, going back to 1989. See here: https://www.basketball-reference.com/playoffs/series.html

Given this, I thought it might be interesting to look at the implied probability each champion had of winning all four playoff series, based on those betting odds. Obviously there’s caveats that betting odds are just contemporaneous estimations (that slightly overestimate both teams’ chances, so the oddsmaker makes money), and also that injuries and other things can happen during a series, which betting odds wouldn’t account for. But it still seemed like an interesting exercise, which generally gets at how big a surprise each champion was contemporaneously.

Interesting work. Did you remove vig first? Otherwise all the odds are slightly too high.

[lessthanjake]

I recently discovered that Basketball Reference has historical betting odds for each playoff series, going back to 1989. See here: https://www.basketball-reference.com/playoffs/series.html

Given this, I thought it might be interesting to look at the implied probability each champion had of winning all four playoff series, based on those betting odds. Obviously there’s caveats that betting odds are just contemporaneous estimations (that slightly overestimate both teams’ chances, so the oddsmaker makes money), and also that injuries and other things can happen during a series, which betting odds wouldn’t account for. But it still seemed like an interesting exercise, which generally gets at how big a surprise each champion was contemporaneously.

Interesting work. Did you remove vig first? Otherwise all the odds are slightly too high.

No I didn’t. So yeah, all the implied odds are slightly too high (I think I alluded to this in my OP), but it shouldn’t really affect the relative rankings much, if at all.

[lessthanjake]

The asterisk Lakers in 2020 were as favored to win as the KD warriors in 2018.

Can we put the bubble ring to rest now knowing that outsized favourite won?

I think there’s some important points worth noting on this:

Method in OP is solely based on actual playoff matchups, so upsets end up increasing a winning team’s probability

The method I used in the OP is based on probability in each of the given matchups a team had. So it’s not a measure of pre-playoffs title odds, and a team definitely has their overall implied probability go up if there’s upsets that result in them facing teams that are perceived to be weaker than other possible opponents.

How this affected the 2020 Lakers implied title probability

When it comes to the 2020 playoffs, the Lakers did not actually have the best pre-playoffs title odds. The Clippers did (with the Lakers 2nd). And the Clippers still had the best title odds after the first round (with the Lakers 3rd). It is quite likely that, if they faced each other, the Clippers would’ve been the very slight favorites over the Lakers—which would’ve made the Lakers’ implied title probability plummet towards the bottom of this list (assuming the Lakers beat the Clippers and then won the title, such that they’d even be on the list). But, of course, the Clippers got upset by the Nuggets before they could meet the Lakers, so the Lakers never faced them and instead faced a team that they were overwhelming favorites against. Similarly, the Bucks were seen as very similarly good as the Lakers—with the Lakers having slightly better title odds than the Bucks before the playoffs started, while the Bucks had slightly better title odds than the Lakers after the first round. If the Lakers and Bucks had met in the finals, then the betting odds for that series would’ve likely been essentially even, which again would’ve tumbled the Lakers’ overall implied title probability. Of course, the Bucks got upset by the Heat, so that matchup never happened, and the Lakers instead played a team in the finals that they were huge favorites over.

In other words, I think the 2020 Lakers being this high is more a reflection of other top teams being upset than it is the Lakers actually being outsized favorites in general. What really happened is that the Lakers were essentially co-favorites alongside the Clippers and Bucks (with the Clippers being slightly ahead of the other two)—with none of those teams being particularly big favorites overall, because the other two existed. But then those other co-favorites got upset so the Lakers only faced teams they were highly favored against, resulting in high implied title probability when using series-by-series odds.

How the Bubble fits in

And that’s where the bubble potentially comes in. I’m firmly of the view (and I think you’ll agree with this) that the bubble championship is just as valid as any other championship, because while the conditions were different than normal they were the same for everyone, so it was a level playing field. And honestly, the conditions in the NBA in general have always been changing—I don’t see people saying there should be an asterisk on titles won when teams didn’t fly on private jets or anything like that. Conditions change, and if there’s a level playing field then the achievement is equally impressive. That said, there is of course a question of whether the reason for those upsets of the Clippers and Bucks was in part due to the effect of the bubble. Was that the case? I don’t think we’ll ever know. Upsets happen in normal seasons too! Maybe Jokic and Murray just were really good even back then and would’ve beaten the Clippers either way. And same with the Heat beating the Bucks—it’s not like the Heat proved to be a total fluke. Or maybe the bubble resulted in the the Clippers and Bucks playing worse, and things would’ve been tougher for the Lakers if COVID hadn’t happened. We don’t really know. To me, it doesn’t matter a whole lot, though, because my view is that if the Clippers and Bucks were made worse by the bubble then they just were a worse team under the circumstances of that season.

How title odds looked in 2020 before the Bubble (i.e. Lakers were one of the favorites pre-Bubble)

The final thing I’ll note about this is that the Lakers clearly were going to be one of the favorites even without COVID/the bubble. On March 1, the Bucks were the title favorites at +225, the Lakers were second favorites at +285, and the Clippers were a close third at +310. And then no one else was really even close to those three. So it’s not like the team that won in the bubble came out of nowhere and wasn’t a favorite before that. Prior to the COVID affecting the season, the Lakers were one of three teams that people saw as actually plausible title contenders, and then COVID happened and they went in the bubble and won the title. It’s certainly very plausible to me that they could have won without COVID/the bubble too. Maybe it would’ve been tougher (i.e. if the Clippers and/or Bucks would’ve performed better without the bubble), but they’d still have been essentially a co-favorite.

[dockingsched]

Title and premise are misleading. Betting odds are not “probability of winning”

[lessthanjake]

Title and premise are misleading. Betting odds are not “probability of winning”

Right, and I think I covered that in the OP when I said “Obviously there’s caveats that betting odds are just contemporaneous estimations . . . But it still seemed like an interesting exercise, which generally gets at how big a surprise each champion was contemporaneously.”

If we wanted to determine the actual probability of winning, we’d need to be able to rewind time and run each of these seasons and/or playoffs a huge number of times and see how many times a given team actually won. Of course, we can’t do that. And I’m not asserting that the method in the OP is equivalent to something like that. Rather, it gets at what people *contemporaneously thought* teams’ probability of winning was (holding constant the teams they faced in each round), and uses series-by-series betting odds to get at that.

[HartfordWhalers]

I recently discovered that Basketball Reference has historical betting odds for each playoff series, going back to 1989. See here: https://www.basketball-reference.com/playoffs/series.html

Given this, I thought it might be interesting to look at the implied probability each champion had of winning all four playoff series, based on those betting odds. Obviously there’s caveats that betting odds are just contemporaneous estimations (that slightly overestimate both teams’ chances, so the oddsmaker makes money), and also that injuries and other things can happen during a series, which betting odds wouldn’t account for. But it still seemed like an interesting exercise, which generally gets at how big a surprise each champion was contemporaneously.

Just as a quick estimation of the methodology, here’s how the calculations work: First of all, let’s look at how to get implied probability for one series. If the odds are a + number, then you add 100 to that number and divide 100 by the resulting number. So, for instance, if a team has +250 odds, you take 100/(250+100) = 0.286. So that’d imply a 28.6% chance of winning. If the odds are a - number, then you divide that number by itself plus 100. For example, if a team has -200 odds, you take 200/(200+100) = 0.667. That implies a 66.7% chance of winning. Now that we know how to determine the implied odds for a single series, how am I getting the implied odds of winning all four series? Well, that’s just by multiplying the probability for all four series. So if betting odds give a team implied odds of 90%, 80%, 70%, and 60% in the four rounds, then the implied probability of winning the title would be 0.9*0.8*0.7*0.6 = 0.302. So that’s a 30.2% overall probability of winning the title, based on series-by-series betting odds.

So what was each of the last 35 champions’ implied probability of winning the title, based on series-by-series betting odds?

NBA Champions’ Implied Probability of Winning Title, since 1989 (using series-by-series betting odds)

1. 1997 Bulls: 73.5%
2. 1996 Bulls: 70.5%
3. 2017 Warriors: 65.5%
4. 2000 Lakers: 62.6%
5. 2013 Heat: 59.3%
6. 1992 Bulls: 59.0%
7. 2015 Warriors: 54.4%
8. 1990 Pistons: 51.0%
9. 2018 Warriors: 50.1%
10. 2020 Lakers: 47.1%
11. 2009 Lakers: 45.9%
12. 2002 Lakers: 45.7%
13. 1998 Bulls: 43.6%
14. 1999 Spurs: 38.7%
15. 1991 Bulls: 37.4%
16. 2010 Lakers: 36.4%
17. 1989 Pistons: 36.1%
18. 2001 Lakers: 35.3%
19. 2005 Spurs: 33.8%
20. 1993 Bulls: 32.6%
21. 2014 Spurs: 31.1%
22. 2012 Heat: 26.7%
23. 2016 Cavaliers: 26.1%
24. 2007 Spurs: 25.6%
25. 2003 Spurs: 24.5%
26. 2023 Nuggets: 21.6%
27. 2022 Warriors: 21.4%
28. 1994 Rockets: 20.0%
29. 2008 Celtics: 19.7%
30. 2021 Bucks: 9.4%
31. 2006 Heat: 8.4%
32. 2019 Raptors: 6.3%
33. 2011 Mavericks: 4.9%
34. 2004 Pistons: 4.4%
35. 1995 Rockets: 1.3%

Anyways, I am not presenting any particular conclusion here. I know there are people who like to exult the star player on the types of teams near the bottom of this list, with the idea being that they rose in the playoffs and carried a team to the title. There’s some validity to that. But, of course, the betting odds we’re talking about here are internalizing how good people think the team is, including the best player on the team. So the teams with the highest probability here have the highest probability in significant part because the greatness of their best players were priced into the odds. Which is to say that I don’t think this list really exposes any particular team or player. It’s more just interesting information. And, as someone who followed basketball throughout this time period, it is generally consistent with my recollection of how surprised I was by each champion, bearing in mind an assessment of the team’s quality and the quality of their opponents.

I'm going to argue that this is a really bad methodology. It certainly isn't championship odds. It is luck and happenstance adjusted odds given the opponents that occur..

Just using conference as an example but it obviously generalizes. Suppose there are 2 strong teams each equal in a conference, and 6 awful teams. Each strong team has a ~47% chance before the playoffs start to reach the finals, with the remaining 6 teams each at 1%.

First round comes and the two favorites are 99% odds. One of the two fully unexpected stumbles, lets call it the #1 seed just to make the point. That would mean that in each remaining series, the #2 seed would be favored at about 99% and would get by your system as a 97% or so champion despite having an exactly equal team that just quirky lost beforehand.

Or put simply, only grading the matchups that happen ignore that those matchups themselves were the subject of odds.

I cannot speak to if BR has them, but the actual futures odds of each team advancing to the championship. You can see the odds before any games start and get what the real odds (as per Vegas and all those disclaimers) were.

[lessthanjake]

I'm going to argue that this is a really bad methodology. It certainly isn't championship odds. It is luck and happenstance adjusted odds given the opponents that occur..

Just using conference as an example but it obviously generalizes. Suppose there are 2 strong teams each equal in a conference, and 6 awful teams. Each strong team has a ~47% chance before the playoffs start to reach the finals, with the remaining 6 teams each at 1%.

First round comes and the two favorites are 99% odds. One of the two fully unexpected stumbles, lets call it the #1 seed just to make the point. That would mean that in each remaining series, the #2 seed would be favored at about 99% and would get by your system as a 97% or so champion despite having an exactly equal team that just quirky lost beforehand.

Or put simply, only grading the matchups that happen ignore that those matchups themselves were the subject of odds.

I cannot speak to if BR has them, but the actual futures odds of each team advancing to the championship. You can see the odds before any games start and get what the real odds (as per Vegas and all those disclaimers) were.

Yeah, I get all that, but I’m actually specifically wanting to get at what the probability was when holding constant the teams they faced (which, I’ll note, is something out of the title team’s control, so is not unreasonable to hold constant). If we didn’t want to do that, we could look at pre-playoff title odds, which are definitely available (and actually go further back than 1989).

Does this methodology get at probability of winning (or rather, since it’s betting odds, contemporaneous perception of the probability of winning) accounting for all possible results of other series? No, of course not. It’s inherently holding the specific matchups constant. But it does get at what the contemporaneous perception of a team’s chances of winning was, given the opponents they actually faced. And, again, that’s what I’m interested in showing here. It gets at how big a surprise a title winner was at the time, given the road that that title winner faced.

So I think this is perhaps an example of an objection to the shorthand used in the title of the thread. I thought a lot about how to title the thread, and ultimately I didn’t really see a way to actually distill exactly what the thread is getting at without a title that would be stupidly long.

[Jasen777]

I'm going to argue that this is a really bad methodology. It certainly isn't championship odds. It is luck and happenstance adjusted odds given the opponents that occur..

Just using conference as an example but it obviously generalizes. Suppose there are 2 strong teams each equal in a conference, and 6 awful teams. Each strong team has a ~47% chance before the playoffs start to reach the finals, with the remaining 6 teams each at 1%.

First round comes and the two favorites are 99% odds. One of the two fully unexpected stumbles, lets call it the #1 seed just to make the point. That would mean that in each remaining series, the #2 seed would be favored at about 99% and would get by your system as a 97% or so champion despite having an exactly equal team that just quirky lost beforehand.

Or put simply, only grading the matchups that happen ignore that those matchups themselves were the subject of odds.

I cannot speak to if BR has them, but the actual futures odds of each team advancing to the championship. You can see the odds before any games start and get what the real odds (as per Vegas and all those disclaimers) were.

It's not a bad methodology, it's just not measuring what you want.

[HartfordWhalers]

It gets at how big a surprise a title winner was at the time, given the road that that title winner faced.

So I think this is perhaps an example of an objection to the shorthand used in the title of the thread. I thought a lot about how to title the thread, and ultimately I didn’t really see a way to actually distill exactly what the thread is getting at without a title that would be stupidly long.

Yeah, I think the bold is a great shorthand description. And it is an interesting concept, just different than how I read the title and how I felt people were taking things like the bubble Lakers discussion.

Maybe "Opponent adjusted title odds" as a title?

[lessthanjake]

By the way, just in case people are curious, here’s what the implied probability of winning would be if we instead looked at pre-playoff title odds (which means this is not accounting for the actual specific matchups a team actually got, nor is it as precisely accounting for in-playoff injuries or other changes in playoff form, as series-by-series odds do). For this, I am using data found here: https://www.sportsoddshistory.com/nba-odds/.

NBA Champions’ Implied Probability of Winning Title, since 1989 (using pre-playoffs title odds)

1. 1996 Bulls: 80.0%
2. 1997 Bulls: 66.7%
3. 1992 Bulls: 66.7%
4. 2017 Warriors: 63.4%
5. 2000 Lakers: 55.6%
6. 2013 Heat: 55.6%
7. 2018 Warriors: 42.2%
8. 1998 Bulls: 41.7%
9. 2002 Lakers: 41.7%
10. 1989 Pistons: 38.5%
11. 2009 Lakers: 38.5%
12. 2015 Warriors: 36.4%
13. 2008 Celtics: 35.7%
14. 1990 Pistons: 33.3%
15. 2012 Heat: 33.3%
16. 2010 Lakers: 30.8%
17. 1991 Bulls: 28.6%
18. 2005 Spurs: 28.6%
19. 2003 Spurs: 28.6%
20. 2020 Lakers: 26.3%
21. 1993 Bulls: 25.0%
22. 2001 Lakers: 25.0%
23. 2014 Spurs: 23.3%
24. 1999 Spurs: 22.2%
25. 2007 Spurs: 20.0%
26. 2016 Cavaliers: 17.4%
27. 2006 Heat: 16.7%
28. 1994 Rockets: 14.3%
29. 2004 Pistons: 12.5%
30. 2021 Bucks: 10.5%
31. 2022 Warriors: 9.1%
32. 2023 Nuggets: 9.1%
33. 2019 Raptors: 8.3%
34. 1995 Rockets: 5.3%
35. 2011 Mavericks: 5.3%

Overall, this unsurprisingly looks pretty similar to the list in the OP. At some point I’ll aim to compile a list of how much these teams’ implied title probability based on series-by-series betting odds differ from their implied title probability based on pre-playoffs odds. Of course, those changes can be for a lot of different reasons. Upsets of other teams can create a significant difference, but so can the team itself or its opponents looking a lot better or worse than previously expected as the playoffs go on.

For reference, the pre-playoffs title odds go back further than the series-by-series odds do. The above only lists the teams that were in the OP, which only go back to 1989. But below are the pre-playoffs odds for teams prior to that where we have pre-playoff title odds:

- 1987 Lakers: 71.4%
- 1986 Celtics: 62.5%
- 1988 Lakers: 62.5%
- 1985 Lakers: 45.5%
- 1983 Sixers: 41.7%
- 1984 Celtics: 33.3%
- 1980 Lakers: 33.3%
- 1976 Celtics: 33.3%
- 1982 Lakers: 28.6%
- 1981 Celtics: 25.0%
- 1979 Sonics: 25.0%
- 1977 Blazers: 16.7%
- 1978 Bullets: 11.1%