Spingold Round of 65

[paulg]

How did they reduce the 65 teams to 32 yesterday?

 

Or did they find a team happy not to play in the event?

[jdonn]

They had one 3 way with a single survivor.

 

The interesting question was which 3 teams should be in it. 1, 64, and 65 was my first thought but I soon decided that was dumb despite the mathematical symmetry (with 65 through 128 teams entered and only heads up matches with byes, 64 would play 65 and the winner would play 1). The priority should be avoiding high seeds having to play in 3 ways. So I then thought it should be 32, 33, and 65. Just add the lowest team to the 64-team-tournament match with the least-high favorite. The actual choice of 32, 64, and 65 seems weird to me. It changed every matchup since then 1 played 63, 2 played 62, etc. But I can understand that if 32 is punished by having to be in a 3 way it's only fair both teams should be rated low, so maybe what they did was right. I guess if they had 66 teams the 3 ways would be 32-64-65 and 31-63-66 with 1-62, 2-61, etc.

[Mbodell]

The IMO obviously correct way. 31 head-to-head matches and 1 3-way with 1 survive with the two lowest seeds + middle seed (32, 64, 65).

 

Our district had 17 teams for a 4 session KO in GNT B qualifications and decided to do a last round (finals) 3-way with one survive (with earlier 3-ways with 2 survive) instead which is clearly sub-optimal.

 

The 5K spingold played down to 24 teams today and will do 8 3-ways with 2 survive so both the 5K and full spingold will have 16 teams on Wednesday and should both finish Saturday. The 1.5K spingold played down to 16 teams today and should finish Friday.

[pooltuna]

The IMO obviously correct way. 31 head-to-head matches and 1 3-way with 1 survive with the two lowest seeds + middle seed (32, 64, 65).

 

Our district had 17 teams for a 4 session KO in GNT B qualifications and decided to do a last round (finals) 3-way with one survive (with earlier 3-ways with 2 survive) instead which is clearly sub-optimal.

 

The 5K spingold played down to 24 teams today and will do 8 3-ways with 2 survive so both the 5K and full spingold will have 16 teams on Wednesday and should both finish Saturday. The 1.5K spingold played down to 16 teams today and should finish Friday.

Not sure what happened here. The bulletin showed 27 teams but chart showed 28. So not sure if we dropped 11 or 12 teams. Anyway as the 27th seed we survived the first day. Probably my experience with pick up players helped as I met my partner, C. J. Jameson, a Stanford collegiate player, Sunday night.

[TimG]

The IMO obviously correct way. 31 head-to-head matches and 1 3-way with 1 survive with the two lowest seeds + middle seed (32, 64, 65).

The question becomes: would you rather be the 32 seed and face a 3-way with one survivor with the 64 and 65 seeds, or would you want to be the 31 seed and face the 33 seed in a head-to-head match?

 

My intuition tells me that the 32 seed would have a (much) easier path to round two. And, that seems counter to the objective of seeding.

[Edmunte1]

My team was 63rd seeded, but i would rather prefered to play in a 3-way one survivor than against top seeded team

[eyhung]

The IMO obviously correct way.  31 head-to-head matches and 1 3-way with 1 survive with the two lowest seeds + middle seed (32, 64, 65).

The question becomes: would you rather be the 32 seed and face a 3-way with one survivor with the 64 and 65 seeds, or would you want to be the 31 seed and face the 33 seed in a head-to-head match?

 

My intuition tells me that the 32 seed would have a (much) easier path to round two. And, that seems counter to the objective of seeding.

Not clear to me that it's better to be 32 than 31. Even if you were known to be exactly a 65% favorite to win a match against each of the other teams in your 3-way, the probability of winning both matches as a 65-35 favorite is just 42%. I don't feel a 32 is more than 65% likely to beat a 64/65 -- the mini-Spingold takes out a lot of chaff. Adding in the greater chance of winning on IMP quotient makes it better than 42%, but it hardly seems significantly better than the roughly 50% chance you have as the 31 seed vs. the 33.

 

Anyway, even if there were some inequity in being 31 vs. 32, it doesn't really matter when you consider that nobody really cares about round 1 victories, but about who wins the event. 32 plays 1 and 31 plays 2 in round 2, and both are likely to lose, so it's good to put the 3-way in a place where it's unlikely to matter. I would say that a slight inequity in having a mid-seed advance to day 2 is far better than having an inequity with a top seed advancing to day 2.

[eyhung]

Not sure what happened here. The bulletin showed 27 teams but chart showed 28. So not sure if we dropped 11 or 12 teams. Anyway as the 27th seed we survived the first day. Probably my experience with pick up players helped as I met my partner, C. J. Jameson, a Stanford collegiate player, Sunday night.

A team was added to the 0-1500 Spingold bracket after they carelessly forgot to pre-register. So 12 were eliminated.

[TimG]

The 32 seed wins both matches 42.25% of the time. They win one match 45.5% of the time. Half of those times a team will have won two matches and half of those times everyone will have won one match and quotient will be used. If you consider that the 32 seed is 65% likely to win on quotient (not sure this is a valid assumption), that means the 32 seeds advances 57% of the time.

[jdonn]

Also I think the 32 seed is more like 85% to beat the 64 or 65 seed in a match. So....

 

Plus in Tim's calculation I bet it's more like 60% to 65% since if every team wins 1 match then it's most likely the 32 seed had the best quotient.

 

I still think 32, 33, 65 is the right lineup for the 3 way.

[barmar]

Back in the days before the mini-Spingolds, most of the low seeds were intermediate/advanced players who didn't have a chance in hell of advancing. The main reason for entering was to be able to play a top-rated team for a full day. If you're going to get stuck in a 3-way with another bottom seed and a middle, what's the point?

 

Someone wrote that the top seed shouldn't be in a 3-way, but why not? Whether they play #63 or both #64 and #65, they're a heavy favorite to win. And this gives both bottom seeds an opportunity to play a great team, although only half as long, which is probably why they entered.

 

Then again, they created the minis so that the champions wouldn't have to waste time playing against the hoi poloi for the first round.

[cherdanno]

I still think 32, 33, 65 is the right lineup for the 3 way.

Why? That makes it ridiculously better to be 34-seed than to be 33-seed.

[TimG]

Then again, they created the minis so that the champions wouldn't have to waste time playing against the hoi poloi for the first round.

I doubt this was the motivation. I expect ACBL recognized that some people did not enter the Spingold because they felt like it was a losing masterpoint proposition and also recognized that these same players would enter a restricted event.

[TimG]

I still think 32, 33, 65 is the right lineup for the 3 way.

Why? That makes it ridiculously better to be 34-seed than to be 33-seed.

If Josh's guess of the 32 seed being an 85% favorite against the 65 seed is correct, what chance is there that the 65 seed will advance?

 

The 65 seed will win both matches 2.25% of the time. If they advance 15% of the time quotient is used, that only leaves them advancing 4.2% of the time. Taken evenly from the 32 and 33 seeds, that means each of those teams has a 47.9% chance of advancing. That doesn't seem "ridiculously" worse.

 

Of course if eyhung's 65% is correct, then the 33 seed has only a 40% chance of advancing. That does seem to me to qualify as a ridiculous difference.

[jdonn]

I still think 32, 33, 65 is the right lineup for the 3 way.

Why? That makes it ridiculously better to be 34-seed than to be 33-seed.

Disagree, the 65 seed advancing is very slim odds so the 33 seed is not severely impacted.

 

Also I'm thinking if there were more teams. Say there were 70. I wouldn't want it to be 27, 69, 70, then 28, 67, 68, etc. I would want it to be 32, 33, 65, then 31, 34, 66, etc down to 27, 38, 70.

[nigel_k]

If it was 32,33,65 then I would prefer to be seeded 34 and play the 31 seed straightup rather than be seeded 33 and go in a 3 way with the 32 seed. OTOH 32,64,65 is clearly wrong because it is obviously better to be 32 than 31 in that case.

 

So maybe something like Josh's suggestion but with slightly lower seeds, e.g. 35,36,65.

[TimG]

So maybe something like Josh's suggestion but with slightly lower seeds, e.g. 35,36,65.

I don't think any solution which does not allow for each of the top 32 seeds to advance is a good one.

[jdonn]

So maybe something like Josh's suggestion but with slightly lower seeds, e.g. 35,36,65.

I don't think any solution which does not allow for each of the top 32 seeds to advance is a good one.

In fairness obviously it will never be anywhere near perfect unless you eliminate the 3-way and give 63 byes in the first round.

[jdonn]

If it was 32,33,65 then I would prefer to be seeded 34 and play the 31 seed straightup rather than be seeded 33 and go in a 3 way with the 32 seed. OTOH 32,64,65 is clearly wrong because it is obviously better to be 32 than 31 in that case.

 

So maybe something like Josh's suggestion but with slightly lower seeds, e.g. 35,36,65.

Based on your first paragraph you should want it to be 1, 64, 65 then?

[eyhung]

85% seems too high for a 32 vs. a 65 in the Spingold (where the hopeless teams are the mini-spingolds) -- it might be ok in the Vanderbilt, but the last Spingold I played in, the Vegas spingold, was noted for its very strong 60-seeds.

 

Also, see the very last post by rogerclee of this thread :

 

http://forums.bridgebase.com/index.php?showtopic=37625&hl=

 

This formula results in a ~75% chance for a team with a 0.5 IMP/bd advantage to win in a 64-board match. A 1 vs a 65 seems like a 1.0 IMP/bd advantage, but I am not certain a 32 is really 0.5 better than a 65.

[jdonn]

What do the mini spingolds have to do with it? The 65 seed (and any seed in particular) will be either equal to or worse than what the same seed would have been if more teams had entered. And obviously the 60 seeds will be better when 106 teams are entered than when 65 teams are entered.

[TimG]

I would expect the bottom few seeds in the Spingold are very often eligible for one of the mini-Spingolds, but choose to enter the Spingold. There have always been concurrent events a team could enter and there have always been teams with no chance which enter the Spingold.

[nigel_k]

If it was 32,33,65 then I would prefer to be seeded 34 and play the 31 seed straightup rather than be seeded 33 and go in a 3 way with the 32 seed. OTOH 32,64,65 is clearly wrong because it is obviously better to be 32 than 31 in that case.

 

So maybe something like Josh's suggestion but with slightly lower seeds, e.g. 35,36,65.

Based on your first paragraph you should want it to be 1, 64, 65 then?

The principle I am trying to apply is that being seed 1 is better then being seed 2, being seed 2 is better then being seed 3 etc. 'Better' meaning having a draw that improves your chances of going through.

 

I wouldn't want it to be 1, 64, 65 because then I would rather be seed 2 than seed 1.

[NickRW]

Seems to me that you have 5 choices:

 

1) 1/64/65. Logical in the sense that #1 would have played the winner of 64/65 had there been a round of 128. Downside is that you put your #1 seed in a relatively high variance position.

 

2) The normal head to head matches are 1v63, 2v62.... etc ... up to 31v33, then the 3 way is 32/64/65. Some might think it is better to be 33rd seed than 32.

 

2a) Modifying 2 is that you make #33 seed or some arbitrary other seed above that point play 64/65. Not sure how you decide exactly which seed it should be, but at least none of the top 32 seeds have anything to complain about and at some point down the seeding order surely the, say, #40 seed has a better chance of progressing in the 3 way than they would have had in a head to head against the seed they would otherwise have been playing. Probably a mathematician can come up with a formula.

 

3) The normal head to head matches are 1v62, 2v61.... etc. up to 31v32, then the 3 way is 63/64/65. None of the top seeds 32 are inconvenienced and you avoid an aribtrary choice of who plays in the 3 way. One downside is that you then guarantee a very low seed a place in the round of 32.

 

4) The other way, making #1 play 64, 2v63 etc as per normal, leaves 32, 33 and 65 fighting it out and you're really then placing a top 32 seed in a bad position - even #33 would have preferred a head to head with 32. This one seems to really not be a flier to me at all. However:

 

4a) Alternatively making it say 36, 37 and 65 in the 3 way, there probably comes a point where the two higher seeds each have at least a good a chance of progressing in a 3 way with 65 as they would have had in their normal head to head match against a higher ranking team. This also needs someone to crunch some standard deviation numbers.

 

Nick

[TimG]

2a) Modifying 2 is that you make #33 seed or some arbitrary other seed above that point play 64/65.  Not sure how you decide exactly which seed it should be, but at least none of the top 32 seeds have anything to complain about and at some point down the seeding order surely the, say, #40 seed has a better chance of progressing in the 3 way than they would have had in a head to head against the seed they would otherwise have been playing.  Probably a mathematician can come up with a formula.

 

3) The normal head to head matches are 1v62, 2v61.... etc. up to 31v32, then the 3 way is 63/64/65.  None of the top seeds 32 are inconvenienced and you avoid an aribtrary choice of who plays in the 3 way.  One downside is that you then guarantee a very low seed a place in the round of 32.

Some top 32 team(s) will be inconvenienced since they can't all advance.