MLS Announces New Playoff Format for 2019

When evaluating the new structure, keep in mind we will likely see a return of the SuperLiga soon via expansion of the Campeones Cup or otherwise. Could look something like this, which starts to make the regular season standings look more and more valuable:

1. Conference Semifinals; Campeones Cup; CONCACAF Champions League
2. Conference Quarterfinals; Campeones Cup
3. Conference Quarterfinals; Campeones Cup
4. Conference Quarterfinals
5. Conference Quarterfinals
6. Conference Quarterfinals
7. Conference Quarterfinals
8.
9.
10.
11.
12.
13.
14.
 
  • Like
Reactions: adam
When evaluating the new structure, keep in mind we will likely see a return of the SuperLiga soon via expansion of the Campeones Cup or otherwise. Could look something like this, which starts to make the regular season standings look more and more valuable:

1. Conference Semifinals; Campeones Cup; CONCACAF Champions League
2. Conference Quarterfinals; Campeones Cup
3. Conference Quarterfinals; Campeones Cup
4. Conference Quarterfinals
5. Conference Quarterfinals
6. Conference Quarterfinals
7. Conference Quarterfinals
8.
9.
10.
11.
12.
13.
14.
Have they announced an expanded Campeones Cup? I can only find reference to the single winner v winner format.
 
Have they announced an expanded Campeones Cup? I can only find reference to the single winner v winner format.

No, but:

"MLS and Liga MX have joined forces via Soccer United Marketing and this year debuted the Campeones Cup, a one-off annual match featuring the two leagues’ reigning champions, with Tigres UANL defeating Toronto FC in the inaugural edition in September.

Garber noted that the two leagues are exploring the possibility of that event growing into a tournament with multiple teams."

https://www.mlssoccer.com/post/2018...bertadores-mls-teams-campeones-cup-could-grow
 
  • Like
Reactions: adam
No, but:

"MLS and Liga MX have joined forces via Soccer United Marketing and this year debuted the Campeones Cup, a one-off annual match featuring the two leagues’ reigning champions, with Tigres UANL defeating Toronto FC in the inaugural edition in September.

Garber noted that the two leagues are exploring the possibility of that event growing into a tournament with multiple teams."

https://www.mlssoccer.com/post/2018...bertadores-mls-teams-campeones-cup-could-grow
Did that actually happen? If it did, it must have been a total flop because this is the first I've heard of it.
 
  • Like
Reactions: Christopher Jee
Also not much difference being 2 or 3 in the playoffs this season. Roughly same advantage/disadvantage.
There is still an advantage to 2, but it is less than in the past. ASA ran the math.
https://www.americansocceranalysis....e-new-mls-playoff-format-rewards-the-top-seed

I think the article understates the value of finishing second under the new system in the text, and focuses too much on the fact that the gap between 2 and 3 has gotten smaller. It has gotten smaller by a meaningful amount, but the difference also is still meaningful.
Personally I think now 1|2 and 4|5 are the most important gaps. 2|3 is no longer as great as it was, but 4|5 really will make a huge difference now.
I don't buy this. Make me an argument that 2/3 isn't a big deal when we finish 3, win our first playoff game, and then have to go on the road for game 2. 1/2, 2/3, and 4/5 are all massive gaps because they all determine home field in a guaranteed or likely matchup.

I'm replying to these posts from the offseason thread in the new playoff format thread where it seems to fit better.

First, to expand on why I think the ASA article understates the advantage of finishing 2 over 3:

My first assumption is all his calculations are correct, and I'm using his first set of numbers that assumes all teams are equally good to isolate the benefit of seeding. Under the old system, the 2 spot formerly had a 25% chance of winning the conference while the 3 spot had 16. He calls this a 9 pct. point advantage. Under the new system the 2 seed has 19% and the 3 seed has 14%. He calls this a 5 point advantage. It looks like the advantage has been cut nearly in half. But that's not the right way to look at it. If you want to determine solely the benefit of the 2 seed over the 3 seed, I think you have to measure their relative share of their combined chance of winning the conference. Under the old system, the 2 and 3 seeds had a combined 41% chance, of which 25 belonged to the 2 seed, which meant the 2 seed had 61% of their shared chance of winning. Now those same seeds have a combined 33% chance, of which 19 belongs to the 2 seed, giving it 58% of their shared chance of winning. That's not nearly as big a drop as the article makes it seem.

Now let's apply this measure to every single spot: the advantage of 1 over 2, of 2 over 3, etc. By the math, Shwafta Shwafta is right. Here is how they stack up in order.

1-2 68% of shared chance of winning belongs to higher seed
4-5 63%
2-3 58%
3-4 58%
6-7 55%
5-6 50%
 
I'm replying to these posts from the offseason thread in the new playoff format thread where it seems to fit better.

First, to expand on why I think the ASA article understates the advantage of finishing 2 over 3:

My first assumption is all his calculations are correct, and I'm using his first set of numbers that assumes all teams are equally good to isolate the benefit of seeding. Under the old system, the 2 spot formerly had a 25% chance of winning the conference while the 3 spot had 16. He calls this a 9 pct. point advantage. Under the new system the 2 seed has 19% and the 3 seed has 14%. He calls this a 5 point advantage. It looks like the advantage has been cut nearly in half. But that's not the right way to look at it. If you want to determine solely the benefit of the 2 seed over the 3 seed, I think you have to measure their relative share of their combined chance of winning the conference. Under the old system, the 2 and 3 seeds had a combined 41% chance, of which 25 belonged to the 2 seed, which meant the 2 seed had 61% of their shared chance of winning. Now those same seeds have a combined 33% chance, of which 19 belongs to the 2 seed, giving it 58% of their shared chance of winning. That's not nearly as big a drop as the article makes it seem.

Now let's apply this measure to every single spot: the advantage of 1 over 2, of 2 over 3, etc. By the math, Shwafta Shwafta is right. Here is how they stack up in order.

1-2 68% of shared chance of winning belongs to higher seed
4-5 63%
2-3 58%
3-4 58%
6-7 55%
5-6 50%
Even without fancy maths you can see there is a clear contrast between "all at home" (1st seed), "some at home" 2,3,4, and "none at home" 5,6,7, which was my point, that the cutoff between these three "tiers" are the most important
...Your math went above my head, however. How'd you get to 41% for 2,3 when you had 19+14 = 33? I'm just awful at statistics..
 
Your math went above my head, however. How'd you get to 41% for 2,3 when you had 19+14 = 33? I'm just awful at statistics..
41% was the combined chance under the old system: 25+16=41
33% is the combined chance under the new system: 19+14=33 (as you note)
You just mixed up the total for the old system with the components of the new system.
 
  • Like
Reactions: Shwafta
41% was the combined chance under the old system: 25+16=41
33% is the combined chance under the new system: 19+14=33 (as you note)
You just mixed up the total for the old system with the components of the new system.
oh, I see. and then 25 is 61% of 41, so that's how you got that number as well.
 
  • Like
Reactions: mgarbowski
I'm replying to these posts from the offseason thread in the new playoff format thread where it seems to fit better.

First, to expand on why I think the ASA article understates the advantage of finishing 2 over 3:

My first assumption is all his calculations are correct, and I'm using his first set of numbers that assumes all teams are equally good to isolate the benefit of seeding. Under the old system, the 2 spot formerly had a 25% chance of winning the conference while the 3 spot had 16. He calls this a 9 pct. point advantage. Under the new system the 2 seed has 19% and the 3 seed has 14%. He calls this a 5 point advantage. It looks like the advantage has been cut nearly in half. But that's not the right way to look at it. If you want to determine solely the benefit of the 2 seed over the 3 seed, I think you have to measure their relative share of their combined chance of winning the conference. Under the old system, the 2 and 3 seeds had a combined 41% chance, of which 25 belonged to the 2 seed, which meant the 2 seed had 61% of their shared chance of winning. Now those same seeds have a combined 33% chance, of which 19 belongs to the 2 seed, giving it 58% of their shared chance of winning. That's not nearly as big a drop as the article makes it seem.

Now let's apply this measure to every single spot: the advantage of 1 over 2, of 2 over 3, etc. By the math, Shwafta Shwafta is right. Here is how they stack up in order.

1-2 68% of shared chance of winning belongs to higher seed
4-5 63%
2-3 58%
3-4 58%
6-7 55%
5-6 50%

I don't think I agree with this. You can't isolate two seeds like that because the payoff metric isn't to do better than the other seed, but to win the whole thing. For example, using an extreme, if a 6 seed has 10x better odds of winning than the 7 seed, but still only has a 1% chance, they increased their chances of winning against the rest of the field by only 0.9%, not 1000% as your system would contend.

Does that explanation make any sense?

So I think you do have to look at marginal increase in odds between seeds, like the ASA article did.
 
  • Like
Reactions: Christopher Jee
You can't isolate two seeds like that because the payoff metric isn't to do better than the other seed, but to win the whole thing.

I quoted the portion where I believe you went astray. I'm not measuring whether the 2 can do better than the 3, I'm comparing their relative chance of winning the conference, aka the whole thing, taking that we know that the teams involved will be either a 2 seed or a 3 seed. That last clause is the key factor. Of course either team would rather be the 1 seed, and don't want to drop to 4 or lower. But if you want to measure the relative benefit of just 2 compared to 3 and vice versa, you have to exclude all those possibilities. Which currently means they have a combined 33% chance of winning the conference, and the 2 seed has 58% of that shared chance.
If I'm right - and I concede this stuff gets tricky and I can be wrong - I think it's kind of like the Monty Hall paradox. Nothing messes people up -- including most definitely me -- in measuring probabilities more than accounting for knowledge. People have trouble with the Monty Hall problem because they forget to account for the effect of Monty's knowledge of which door is which. In the current problem, I think you're failing to account for the fact that in order to measure the relative probability of 2 or 3 winning the conference compared to the other, you have to know they won't finish 1 or 4-7, and therefore the combined 67% chance of winning the conference assigned to those seeds is irrelevant.

Unless I'm wrong. I confess the knowledge issue in probabilities drives me nuts. It took me weeks to accept the truth about the Monty Hall problem.
 
I'm replying to these posts from the offseason thread in the new playoff format thread where it seems to fit better.

First, to expand on why I think the ASA article understates the advantage of finishing 2 over 3:

My first assumption is all his calculations are correct, and I'm using his first set of numbers that assumes all teams are equally good to isolate the benefit of seeding. Under the old system, the 2 spot formerly had a 25% chance of winning the conference while the 3 spot had 16. He calls this a 9 pct. point advantage. Under the new system the 2 seed has 19% and the 3 seed has 14%. He calls this a 5 point advantage. It looks like the advantage has been cut nearly in half. But that's not the right way to look at it. If you want to determine solely the benefit of the 2 seed over the 3 seed, I think you have to measure their relative share of their combined chance of winning the conference. Under the old system, the 2 and 3 seeds had a combined 41% chance, of which 25 belonged to the 2 seed, which meant the 2 seed had 61% of their shared chance of winning. Now those same seeds have a combined 33% chance, of which 19 belongs to the 2 seed, giving it 58% of their shared chance of winning. That's not nearly as big a drop as the article makes it seem.

Now let's apply this measure to every single spot: the advantage of 1 over 2, of 2 over 3, etc. By the math, Shwafta Shwafta is right. Here is how they stack up in order.

1-2 68% of shared chance of winning belongs to higher seed
4-5 63%
2-3 58%
3-4 58%
6-7 55%
5-6 50%
Regardless of the mathematical argument, I think there is a huge cultural argument. 2 vs 3 has a chance at being home or away if/when they meet in round 2. Another home playoff game, another night in the Bronx vs watching on TV (for those of us who don't generally travel with the team). Our home record vs our road record.

If 2 or 3 gets knocked out in round 1, then the seeding is irrelevant. But if they both make it to round 2, the seeding difference is massive.

Also, they may not win the conference or MLS Cup, but they might have that much better chance at getting to the next round. One more week of obsessive dreaming about the possibility of winning it all. That (to me anyway) is a big deal. In fact, that's the whole deal. Isn't that really why we are all here every day?
 
I don't think I agree with this. You can't isolate two seeds like that because the payoff metric isn't to do better than the other seed, but to win the whole thing. For example, using an extreme, if a 6 seed has 10x better odds of winning than the 7 seed, but still only has a 1% chance, they increased their chances of winning against the rest of the field by only 0.9%, not 1000% as your system would contend.

Does that explanation make any sense?

So I think you do have to look at marginal increase in odds between seeds, like the ASA article did.
I quoted the portion where I believe you went astray. I'm not measuring whether the 2 can do better than the 3, I'm comparing their relative chance of winning the conference, aka the whole thing, taking that we know that the teams involved will be either a 2 seed or a 3 seed. That last clause is the key factor. Of course either team would rather be the 1 seed, and don't want to drop to 4 or lower. But if you want to measure the relative benefit of just 2 compared to 3 and vice versa, you have to exclude all those possibilities. Which currently means they have a combined 33% chance of winning the conference, and the 2 seed has 58% of that shared chance.
If I'm right - and I concede this stuff gets tricky and I can be wrong - I think it's kind of like the Monty Hall paradox. Nothing messes people up -- including most definitely me -- in measuring probabilities more than accounting for knowledge. People have trouble with the Monty Hall problem because they forget to account for the effect of Monty's knowledge of which door is which. In the current problem, I think you're failing to account for the fact that in order to measure the relative probability of 2 or 3 winning the conference compared to the other, you have to know they won't finish 1 or 4-7, and therefore the combined 67% chance of winning the conference assigned to those seeds is irrelevant.

Unless I'm wrong. I confess the knowledge issue in probabilities drives me nuts. It took me weeks to accept the truth about the Monty Hall problem.
Regardless of the mathematical argument, I think there is a huge cultural argument. 2 vs 3 has a chance at being home or away if/when they meet in round 2. Another home playoff game, another night in the Bronx vs watching on TV (for those of us who don't generally travel with the team). Our home record vs our road record.

If 2 or 3 gets knocked out in round 1, then the seeding is irrelevant. But if they both make it to round 2, the seeding difference is massive.

Also, they may not win the conference or MLS Cup, but they might have that much better chance at getting to the next round. One more week of obsessive dreaming about the possibility of winning it all. That (to me anyway) is a big deal. In fact, that's the whole deal. Isn't that really why we are all here every day?
If you can’t get the paragraphs down to the length of a tweet, they’re impossible to read on a phone.
 
I quoted the portion where I believe you went astray. I'm not measuring whether the 2 can do better than the 3, I'm comparing their relative chance of winning the conference, aka the whole thing, taking that we know that the teams involved will be either a 2 seed or a 3 seed. That last clause is the key factor. Of course either team would rather be the 1 seed, and don't want to drop to 4 or lower. But if you want to measure the relative benefit of just 2 compared to 3 and vice versa, you have to exclude all those possibilities. Which currently means they have a combined 33% chance of winning the conference, and the 2 seed has 58% of that shared chance.
If I'm right - and I concede this stuff gets tricky and I can be wrong - I think it's kind of like the Monty Hall paradox. Nothing messes people up -- including most definitely me -- in measuring probabilities more than accounting for knowledge. People have trouble with the Monty Hall problem because they forget to account for the effect of Monty's knowledge of which door is which. In the current problem, I think you're failing to account for the fact that in order to measure the relative probability of 2 or 3 winning the conference compared to the other, you have to know they won't finish 1 or 4-7, and therefore the combined 67% chance of winning the conference assigned to those seeds is irrelevant.

Unless I'm wrong. I confess the knowledge issue in probabilities drives me nuts. It took me weeks to accept the truth about the Monty Hall problem.

I don't have time to overly strain my brain on this right now. But aren't you then comparing apples and oranges by comparing teams that necessarily will be a 2 and a 3 seed between teams that will necessarily be a 6 and a 7 seed?

To correct for that, you would have to multiple the shared chance of winning by the total chance of either of those seeds winning, which brings you back to the original numbers.

EDIT: Put another way, it doesn't matter how much better a X is than a Y if neither of them win. By comparing the seeds in pairs, you're also making an assumption that one of them win the whole thing, and under that assumption how important it is to be seed X v seed Y.
 
Last edited:
I don't have time to overly strain my brain on this right now. But aren't you then comparing apples and oranges by comparing teams that necessarily will be a 2 and a 3 seed between teams that will necessarily be a 6 and a 7 seed?

To correct for that, you would have to multiple the shared chance of winning by the total chance of either of those seeds winning, which brings you back to the original numbers.

EDIT: Put another way, it doesn't matter how much better a X is than a Y if neither of them win. By comparing the seeds in pairs, you're also making an assumption that one of them win the whole thing, and under that assumption how important it is to be seed X v seed Y.
I don't think any of this is right, but it's apparent we're not going to resolve this here. I think we're both discussing in good faith, but we're also talking past each other and one of us is missing something the other can't make him see.
 
I don't think any of this is right, but it's apparent we're not going to resolve this here. I think we're both discussing in good faith, but we're also talking past each other and one of us is missing something the other can't make him see.

I'll try again anyway.

I think the disagreement is in what we are measuring. You're measuring the isolated impact of being Seed X rather than Seed X+1. Which is fine and well, but it's not actually what we are looking for.

My contention is that your measure isn't relevant, because X and X+1 aren't competing with each other only. Isolating them removes necessary information about how they will fair against the entire field. It's like saying the fraction 3/10 is bigger than 2/5 because you'd rather have 3 of something than 2 of something.

So again, I restate, that your measure is fine if you want to look at how much better Seed X will fare over Seed X+1 only. But in the full context of the competition, that only matters if Seed X or Seed X+1 win against the rest of the field. And we know that information, so it should be included in the calculation.

We care about which step up in seed will lead to the biggest increase in championships won, no? You win a championship by beating the entire field, not just the seed below you.
 
  • Like
Reactions: mgarbowski
I'll try again anyway.

I think the disagreement is in what we are measuring. You're measuring the isolated impact of being Seed X rather than Seed X+1. Which is fine and well, but it's not actually what we are looking for.

My contention is that your measure isn't relevant, because X and X+1 aren't competing with each other only. Isolating them removes necessary information about how they will fair against the entire field. It's like saying the fraction 3/10 is bigger than 2/5 because you'd rather have 3 of something than 2 of something.

So again, I restate, that your measure is fine if you want to look at how much better Seed X will fare over Seed X+1 only. But in the full context of the competition, that only matters if Seed X or Seed X+1 win against the rest of the field. And we know that information, so it should be included in the calculation.

We care about which step up in seed will lead to the biggest increase in championships won, no? You win a championship by beating the entire field, not just the seed below you.

Ok, we're on the same page now. We seem to agree on the logic but disagree on what question matters. I get why you say the question you identify is most important, but the discussion started with one poster identifying the specific step-ups he thought were the most impactful, and another challenging his statement. [I'm not tagging them because they don't need to get dragged into this]. Those are interesting and relevant questions as well. I'm not saying they are the only issue, or even the biggest issue, but they are meaningful.
 
  • Like
Reactions: sbrylski
We care about which step up in seed will lead to the biggest increase in championships won, no? You win a championship by beating the entire field, not just the seed below you.
FWIW, I care about championships, but I also care about every week that the team is still in the playoffs. Any week of season is better than any week of off-season. So to me, every X seed (for 1/2, 2/3, 4/5), because it gives an extra home game (guaranteed or likely in rounds 1 or 2) and therefore a greater likelihood of one more week of action and dreams, is a significant step up from every X+1 seed.

Not exactly what you and mgarbowski mgarbowski were debating, but my two cents.