[Mbodell]

How much more likely is a top seeded team to win *because* of their seeding versus just because they are of a quality that deserves a top seed?

For sake of argument let's say that in next year's Vanderbilt teams that Welland is the strongest team and that they have a top 4 seed. If you could clone the team and have an exactly as strong team, call them Welland', but Welland' has no seeding points and enters as the last seeded team how much more likely is it that Welland will win the tournament versus Welland'? Would it be a matter of only 1 or 2% more likely or would it be more like 50% more likely or what?

[Echognome]

Someone can correct me, but I believe you take over the seed of your opponent in the bracket if you defeat them. So, although you might have a tougher earlier round match, you will have easier subsequent matches. That being the case, I'd say your odds decrease only slightly (and only because you have to play an additional tough team in the early rounds). Of course your chances to reach, say, the quarterfinals drop significantly. But the chances of winning outright means you are going to have to beat good teams regardless.

[Jlall]

Not much.

[fred]

Jlall, on Aug 11 2009, 12:14 AM, said:

Not much.

In my experience:

- If your goal is to win then it doesn't matter that much. You are going to have to play and beat a bunch of strong teams regardless of where you are seeded. Being seeded (say) #1 instead of (say) #16 will usually (but not always) results in a significantly easier match for you in the round-of-32, but whoever you play, your round-of-32 opponents do not rate to be anywhere close to the best team you are going to have to beat in order to win.

- If your goal is to get to a specific round then it matters more. For example, in the past two NABCs my team has been seeded in the 9 to 12 group. That means that in the round-of-16 we always had to play a higher-seeded team. That means that, all else being equal (they probably are not since I think our team is under-seeded), we are more likely than not to lose in the round-of-16.

But I think our team earned enough seeding points in the Spingold that we will probably be in the 5 to 8 group in next year's Vanderbilt. That means that, all else being equal, we will play a team in the round-of-16 that we are supposed to beat.

So if our goal is to get to the round-of-8 (it isn't in our case), the extra seeding points we just won rate to make a difference.

Of course all of this assumes is that the seeding of the field is more-or-less true.

- It was not that long ago that seeding points were considerably more important than they are now (because there were far fewer strong teams in these events and those strong teams invariably had a lot of more seeding points than everyone else).

My perception is that it used to be normal for all of the top 8 teams to survive the round-of-32 whereas nowadays it seems to be the norm that at least one of those teams is eliminated in that round.

- A curiousity that I noticed in the Spingold for the first time is that some numbers are "bad". Being #11, for example, meant (seemily randomly) that we would never have seating rights from the round-of-16 onwards. IMO seating rights are not that important, but I still found this to be amusing (and I also found it amusing that I had never thought of this before).

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[Jlall]

Agree Fred.

This tournament I was the 21 seed which was nice because before I had been 32 etc for a while meaning winning the round of 32 was very tough. In fact I think the 42 seed rates to be better than the 32 seed (slightly tougher round of 64 match, but significantly easier round of 32 match in general). So in terms of winning the round of 32, the 21 seed was gonna be significantly better than the 32 seed (although we barely avoided the very underseeded Diamond team if we made it to the round of 32).

Since OP asked about how much difference it makes in terms of winning the thing I thought the right answer was "not much" which you seem to agree with.

[fred]

Jlall, on Aug 11 2009, 01:00 AM, said:

Agree Fred.

This tournament I was the 21 seed which was nice because before I had been 32 etc for a while meaning winning the round of 32 was very tough. In fact I think the 42 seed rates to be better than the 32 seed (slightly tougher round of 64 match, but significantly easier round of 32 match in general). So in terms of winning the round of 32, the 21 seed was gonna be significantly better than the 32 seed (although we barely avoided the very underseeded Diamond team if we made it to the round of 32).

Since OP asked about how much difference it makes in terms of winning the thing I thought the right answer was "not much" which you seem to agree with.

Agree - we are on the same page.

Probably a good general rule is that your seed rates to matter most in the round that is the next higher power of 2 above your seed.

By the way, I suspect your team was seriously under-seeded as well and I am glad we did not have to play you early in the event

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[xcurt]

Re the relative strength of "top" teams vs the next tier of teams:

In the round of 32, the higher original seed averaged about +0.5 IMPs/bd

In the round of 16, the higher original seed averaged about +0.2 IMPs/bd

In the round of 8, the higher original seed averaged about +0.3 IMPs/bd

In the round of 4, both winners were originally the lower seed.

Looking at the rosters, about the top 20 seeds have players with significant international experience (plus a few outliers like Brink/Drijver, who haven't been around long enough to rack up that many SPs).

Generally this says to me that you really want to be in the top 12 so you play one of the 21-32 seeds in the round of 32.

[awm]

It seems like a top seed will generally play one fewer good team than a low seed on the way to any round. For example, suppose you want to win the event. If you're the number one seed, then you play: 64, 32, 16, 8, 4, 2 assuming higher seeds win. Probably the 64 seed and the 32 seed are fairly weak (although there is enough randomness that one or both might be a decent international team) so you probably have to go through four good teams to win the event. If you're the 64 seed, then you play: 1, 32, 16, 8, 4, 2 again assuming that the higher seed wins. So this means you had to go through five good teams to win the event. Obviously you can disagree about where the teams generally start to be good and of course it is always possible for a good team to have a low seed, but the general point holds.

Having to go through one more good team likely reduces your chances of reaching any particular round by almost half if your team is also good, or quite a bit more if your team is not so good. This is pretty significant; if you take Fred's team for example and assume that they are roughly equal to the teams in the top 16 and quite a bit better than the lower seeds, then Fred's chance of winning from a top seed (like one or two) will be perhaps 5% (he has to beat four good teams each of which is about 50-50 and avoid being upset by two not-so-good teams). If Fred's team were seeded much lower for some reason, then his chance of winning the event will be more like 3% (now he has to beat five good teams each of which is about 50-50, one not-so-good team with a small chance of upset, and also has to get through a three-way where four advance).

The absolute difference between these numbers is not very big, but this reflects the fact that there are lots of good teams that could win. The relative difference is however pretty significant -- the difference between winning one spingold in twenty and less than one spingold in thirty.

[3for3]

Isn't the math fairly simple on this? Let's assume there are 64 teams, and our hypothetical team which is tied for the best in the event, gets seeded last. In the round of 64, instead of having 96% chance of winning, they now have 50%. All else stays the same. Let's round off 96 to 100, to keep the math simple.

There chances would then be 1/2 of what they would be otherwise.

This seems more significant than Fred and Justin are suggesting. Of course there aren't exactly 64 teams, and even if there were, there is some shuffling that goes on, so our team with no points might draw the 8 seed or something in the first round. Still, I'd argue that seeding IS important, since that one extra match where they are 50-50 instead of near locks, makes a huge mathematical difference.

Danny

[fred]

3for3, on Aug 11 2009, 12:19 PM, said:

Isn't the math fairly simple on this?  Let's assume there are 64 teams, and our hypothetical team which is tied for the best in the event, gets seeded last.  In the round of 64, instead of having 96% chance of winning, they now have 50%.  All else stays the same.  Let's round off 96 to 100, to keep the math simple. 

There chances would then be 1/2 of what they would be otherwise.

This seems more significant than Fred and Justin are suggesting.  Of course there aren't exactly 64 teams, and even if there were, there is some shuffling that goes on, so our team with no points might draw the 8 seed or something in the first round.  Still, I'd argue that seeding IS important, since that one extra match where they are 50-50 instead of near locks, makes a huge mathematical difference.

Danny

I have to admit that I did not read the second paragraph of the OP (sorry).

My answer pertained to more realistic scenarios in which a given team's seed varied among values that are relatively close to what it actually "should be".

As a rough estimate, your analysis of the question posed in the second paragraph of the OP makes sense to me.

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[3for3]

And for the range you discussed, the improved chances can't be that great. Despite all of the time spent around the seeding points, it probably matters relatively little which 'shuffle' you are in.

Danny

[Tomi2]

I made some excelsheet with a scenario,

to let it be easier I made a 8-teams KO
there are teams ABCDEFGH with A beeing the strongest then B etc.
every team has rating points from 100 for team H, to 120 for G etc till 240 for team A

if two teams play against each other the chances for one to win are own rating / (own rating + their rating)

In the semifinals you calculate this for both potential opps multiplicated by the chance that this team really makes it.
its for example for Team A: p(d wins vs e)*240 / 240+ratingD + p(e wins vs d)* 240 / 240+ratingE

same for final...

Now I first seeded the stronges team 1, then team B 2 etc, so 1-8 2-6 etc are quaterfinals and 1/8 - 4/5 2/7 - 3/6 are semis
chances to win the event are:
A: 0,235365138
B: 0,196187125
C: 0,161206397
D: 0,130015443
E: 0,102599906
F: 0,078423891
G: 0,05718023
H: 0,039021871

now first I changed the seeding in favor to team A, so they play H in quaterfinals and winner of F and G in semis then they have: 0,253560726... thats relatively 7% more
now I change the seeding against team A so they have to play B in quaterfinals and winner of C and D in semis that makes 0,186518715... thats relatively 20% less
the best scenario for team H (playing G first...) wd be 0,058217903, thats 49% more than original, worst for them (playing A and then B or C) is 0,037476704

this is like somebody else mentioned before like playing 17 or 27 events to win once

I think those are quiet significant numbers, but the key is the "rating" I used. If in real Bridge the teams are closer to each others (more equal) then seeding becomes less important, if teams divergate more, then seeding becomes more important.

Other data: If Team A and B become equal, say 230 and one is seeded 1 the other is seeded 2 then its:
A:0,222028256
B:0,209307908

if for some reason the 2nd strong team becomes seeded 7th:
A:0,231064027
B:0,183228881

[mycroft]

I guess a related question is:

How much are seeding points worth in the Trials, where enough gets you byes out of the round robin or farther?

In other words, how much is being USA I one year worth in trying to become USA I next year? I realize that the last decade's 800 lb gorilla was winning a lot of other events that didn't have "bye to the quarters" on it, but I would think that the bye is more important than the match you still have to play, even if you're an almost lock.

[3for3]

The best bye you can get from doing well in the trials in year X is a bye to the round of 16 in year X+1. Deeper byes require good results in the following cycle events.