PDA

View Full Version : Seeding Question



Doug Soles
11-05-2012, 09:38 AM
For all of you math geniuses out there...

What would be the worst spot (in the top 16 or so) mathematically to be ranked in the current CIF SS rankings for seeding into CIF Prelims? The obvious answer is 16, but does it suck more to be #4 vs. #9, etc? Why?

I've often wondered where the worst spot to be ranked is going in trying to make finals. Any thoughts or is it moot?

Thanks!

Doug

ScottJoerger
11-05-2012, 10:07 AM
With the "Snake" seeding that is used, you would get the same average ranking among the top 16 teams = 8.5:

H1 H2 H3 H4
1 2 3 4
8 7 6 5
9 10 11 12
16 15 14 13
Average 8.5 for each column

So I think it's fair, or at least as fair as it can be (assuming the rankings are totally accurate - which we know is impossible). However, the problem I see with this method is that it only evens out after the snake has filled in two rows (or any even number of rows).

Consider the seeding for an odd number of rows - say the top 12 or top 20 teams:

Top 12 Teams:
H1 H2 H3 H4
1 2 3 4
8 7 6 5
9 10 11 12
Averages:
6 6.3 6.7 7

Top 20 Teams:
H1 H2 H3 H4
1 2 3 4
8 7 6 5
9 10 11 12
16 15 14 13
17 18 19 20
Averages:
10.2 10.4 10.6 10.8

In both cases, the earlier heats have a lower average ranking, which should indicate a higher quality of competition among those top 12 or 20 teams. So it seems to me it would be better to be in Heat 3 or Heat 4.

I'm no mathematician, but those are my thoughts. I'd love to hear what others think about it.

Scott

Chad Scott
11-05-2012, 10:42 AM
If I were a team that is "on the bubble", I would prefer to be in an earlier heat.
Heat 1: 1, 8, 9, 16
Heat 4: 4, 5, 12, 13

Statistically speaking, it would be easier to pull an upset against the 16th ranked team than the 13th ranked team. That being said, I heard that at one time Rich did rankings out to at least 24 (though only 13 get published), so the next round of teams are still being seeded as opposed to randomly assigned. If that is true, then it really doesn't matter. Can a 20th team still upset the 12th or 13th ranked team in the 4th heat? Yep, happens every year.

Rich usually says after Prelims just how accurate it was...so what percentage of the top 16 ranked teams actually qualified for Finals. I think one year it was in the mid-80s, but usually low 90s (If I am remembering correctly).

bbrierly
11-05-2012, 11:38 AM
If you actually go through and dope out the heats (from Div 1 as I have done), using team times from Mt Sac, as well as the current rankings, it does appear that the seeding goes through at least the top 20. Now from this, it would always make sense that the "safest" heats to be in would be 3 and 4 because there would appear to be a bigger separation between the top 16 teams and the remaining seeds. As always though, its always a matter of being ready on that day.

Of course this year, Div 1 is pretty darn deep this year. There really isn't any safe heat, not that I could tell anyway.

Mark Gardner
11-05-2012, 03:09 PM
Before the revised heat sheets were revealed, you could see that Rich had the seeds listed through to 27 (at least in Div. I boys..heat 3)

Keith Chann
11-08-2012, 11:13 AM
Of course there is a Div 1 and 2 bias here because the other divisions do not have 4 heats. For a 3 heat division, the 1st heat would include teams seeded 1, 6, 7, 12, 13, 18, 19. Heat #2 would have teams 2, 5, 8, 11, 14, 17 and Heat #3 would have tema seeded 3, 4, 9, 10, 15, 16.
In heat #1, the second team is closer the 3rd team than it is to the 1st team while in heat #3, the second team is closer to the top seed in its heat. The #3 seed is closer to the 4th seeded team in Heat 3 and thus closer to being eliminated.
In a 3 heat division, it would be more beneficial to be seeded as #6 or 7 versus #9 if you are concerned about how you look versus the top couple of teams but it would be more beneficial to be a #9 if you were concerned about being eliminated.
Either scenario has advantages but if you are just concerned about advancing to the next weekend, being a # 13 seed, you theoretically would need only to beat the #18 seed to advance (a seed difference of 5 spots) while a #15 seed would need to hold off a #16 seed (a seed difference of only 1 spot!) if the seeding holds true to form.

cush
11-10-2012, 02:06 PM
interesting math here. for what it's worth in div 2, my foothill boys, who placed 5th in the 3rd seeded heat (guessing we were therefore ranked 19th or 22nd?) would have been 5th in any of the heats. in an uncanny coincidence, 3 of the 4 fifth place teams, including foothill, had team times of 1:20:22, and one 6th place team had that same time as well (from the 1st seeded heat); the 2nd seeded heat had the slowest 5th place team time of 1:20:38.

the 4th place qualifiers ranged from 1:19:22 (1st seeded heat), to 1:19:54 (2nd), to 1:19:55 (3rd), to 1:20:00 (4th)--this is ever so slightly in reverse order of what the seeding finish should have been (the slowest 4th place seed should have been in the first heat, fastest 4th should have been in the 4th heat).

all-in-all, pretty impressive seeding at least for this division/gender...

a different question altogether: how are individuals divided among the heats? it would seem the fastest individuals should be seeded in the reverse order of the team seedings, but is there any rhyme or reason?

Keith Chann
11-11-2012, 03:25 PM
How are the number of heats determined? The CIF SS bulletin used to specify how many heats each division would get was based on the number of teams entered and gave a numerical basis.
On Saturday, division 1 boys had 54 teams in 4 heats, division 2 had 55 teams in 4 heats while division 3 had 54 teams in 3 heats. The D3 starting lines were ridiculous. Why were there not 4 heats for the D3 boys?

Rich Gonzalez
11-11-2012, 06:12 PM
Will try to answer some of the questions on this thread, if able.

In a four-heat setup, the best non-qualifier (the #17 seed) should emerge be in Heat 1, if everyone ran as projected. If you are one of the best teams in the division, however, the "easiest" heat (if you wanted to cruise at prelims) would be Heat #4 since you'd only need to beat the #20 seed. In some cases, there is a pretty good amount of difference between the #17 seed and the #20 seed, especially in those smaller divisions where the depth trails off after about 13-15 teams.

Both Scott and Chad have the right idea on it.

An attempt is made each year to seed the divisions as deeply as possible with "confidence". This year, a few divisions were seeded 28 deep. One division this year was only seeded 18 deep (!!), which caused me consternation. But Division V Girls clearly lacked enough quality data to confidently go deeper. Fortunately, the Top 15 seeds in the division all advanced. Ironically, it would have been 16 for 16 but the official results for a race at one meet came into question and there was some independent information to make the inquiry seem valid. However, the meet never officially changed those results asked to be reviewed. We did the best we can.

Cush, regarding the seeding for individuals, it is dependent first on team seeding, which CIF has always asked to give first priority in the process. So lets say we have a four-heat division and after the TEAM seeding is done, 6 of the top 16 INDIVIDUALS in the division are on seeded teams that landed in heat 1, 3 of the 16 are in Heat 2, 5 of the 16 are in Heat 3, and 2 are in Heat 4. We'd then look at leading individuals on unseeded teams as well as individual qualifiers from league finals, putting the first few good runners in heats 2 and 4 on the scenario here. In the end, we try to even it out, but it ends up almost being moot when certain good teams with real good individuals opt to cruise (or sit out) a key kid.

Keith, the traditional cutoff in heat determination is a factor of 17 teams maximum. So up to 51 teams, it would be a three-heat, 17-team-per-heat setup. At 52 teams, it traditionally goes up to four heats. Perhaps to balance the alternating number of heats in a session or make it a more feasible time schedule, one division went to 54 teams and kept at three heats-- which resulted in one extra team per heat.

RichEde
11-11-2012, 09:19 PM
How are the number of heats determined? The CIF SS bulletin used to specify how many heats each division would get was based on the number of teams entered and gave a numerical basis.
On Saturday, division 1 boys had 54 teams in 4 heats, division 2 had 55 teams in 4 heats while division 3 had 54 teams in 3 heats. The D3 starting lines were ridiculous. Why were there not 4 heats for the D3 boys?

Keith,

Rich is correct that the traditional cutoff is 17 teams per race. With our original entries, there were 51 D3 boys' teams (3 heats) and 57 D1 boys' teams(4 heats). Despite trying to cross-check divisions. after our initial schedule and seeding was done an on-line, we discovered 3 teams that had been flagged through the entries as D1 but which are really D3. Unfortunately, we felt that re-seeding D3 and adding an extra race (and re-scheduling) after posting would create more confusion (not to mention running the last race in the dark) than the extra team on the line. It obviously wasn't an ideal solution but seemed the lesser of two evils.