Unofficial 2019 D3 XC Thread (Official)

Calvin women beat Loras head-to-head at Oshkosh

head to head wrote:

Calvin women beat Loras head-to-head at Oshkosh

*Wrong*

Loras didn't run at Oshkosh

"more competitive"

Loras lost head to heads against almost the whole Great Lakes Region.

Loras, at Oberlin lost to:

Oberlin - 3rd GL

John Carroll - 1st GL

Baldwin Wallace - 2nd GL

Otterbein - 8th GL

Ohio Northern - 7th GL

Case Western - 9th GL

Occidental -4th in the West

Now, If head to head regional competition is a determining factor for selection explain how you can take Loras without taking all these teams? You want to take 9 teams from the Great Lakes?

You can't take a 4th or 5th place team without taking the third place team.

Don't run like garbage all year, don't get beat by teams that have run like garbage all year at Regionals.

So many questions wrote:

Rankings aside. Regardless if the Loras women got beat by other teams, they should have been selected too. They had great race in the Central Region as well as St. Thomas. Both of these teams scored below 130 points whereas teams selected (6 deep mind you) in the Great Lakes, Atlantic, and New England regions scored well over 200 points. A head-to-head comparison would show that Loras' 1-5 and St. Thomas' 1-5 would better the scorers of RIT, Middlebury, Vassar, Rochester, and Calvin.

The committee took 4 at-large teams from 3 regions and shafted the central region when teams and individuals are more competitive than those who were selected.

This is a losing argument. You can't compare the score one school got in another region to a score another team got in a completely separate race against completely different teams. What do you mean by head-to-head comparison? Comparison of what?

Looking through the teams you mentioned, Middlebury is the only one with solid out-of-region wins (really don't see what you're complaining about with them). As for the others, none of them have wins of note. The committee had absolutely nothing to go on outside of their performance at regionals. There it probably came down to the point spreads at regionals. The teams in the Central got blown out by Wartburg, a solid but not great team. The New England teams were a reasonable distance away from two podium contenders.

All that said, I have no clue why Calvin was picked. They lost to Trine at their conference meet, finished 50 points behind 5th place Allegheny in the great lakes, and beat no good teams at any point this year, other than beating in-region Hope on October 5th. They must have given way too much weight to that win.

So many questions wrote:

Rankings aside. Regardless if the Loras women got beat by other teams, they should have been selected too. They had great race in the Central Region as well as St. Thomas. Both of these teams scored below 130 points whereas teams selected (6 deep mind you) in the Great Lakes, Atlantic, and New England regions scored well over 200 points. A head-to-head comparison would show that Loras' 1-5 and St. Thomas' 1-5 would better the scorers of RIT, Middlebury, Vassar, Rochester, and Calvin.

The committee took 4 at-large teams from 3 regions and shafted the central region when teams and individuals are more competitive than those who were selected.

I think I cracked what happened to Loras. They ran very poorly at Oberlin against out-of-region teams. They finished 200 points behind Bates, who only got 4th in New England. Losing to Luther at their conference meet didn't help.

As for St. Thomas, I have no idea. They beat La Crosse a month ago, and also did well at Louisville, but the committee probably discounted that race because of the mishap.

Blarg wrote:

"more competitive"

Loras lost head to heads against almost the whole Great Lakes Region.

Loras, at Oberlin lost to:

Oberlin - 3rd GL

John Carroll - 1st GL

Baldwin Wallace - 2nd GL

Otterbein - 8th GL

Ohio Northern - 7th GL

Case Western - 9th GL

Occidental -4th in the West

Now, If head to head regional competition is a determining factor for selection explain how you can take Loras without taking all these teams? You want to take 9 teams from the Great Lakes?

You can't take a 4th or 5th place team without taking the third place team.

Don't run like garbage all year, don't get beat by teams that have run like garbage all year at Regionals.

Head to head isn’t the only determining factor to qualify for nationals. In my opinion, the committee took heavy stock into the Oberlin race/other inter-region meets rather than the regional performances. If that is what they will always do, fine, but regional point spreads/who you beat when it matters at Regionals is something I’d like to see more heavily weighted vs a meet in early/mid October.

St. Thomas was a good team that had quality out of region wins (UW-Lax.). It seems that Loras’ win over St. Thomas at Regionals wasn’t considered in the committee’s decision process because the committee won’t look at the 4th ranked team in the region when debating if the 3rd should go. I think that is an oversight on the committee’s process of selecting teams.

I agree, but for different reasons. St. Thomas should have been selected based on their performances up to this point. Loras then should have been selected purely because St. Thomas pushed them in, and not because they deserved it for any other reason other than their good regional race. If you want consistency, THAT is what they have done in the past. In 2015, WashU's men's team just blew up at regionals and got 8th. So, they took 8 teams out of the Midwest, including a couple not so deserving teams purely because they beat WashU on that day.

I think leaving St. Olaf out was clearly correct.

All of the teams I suggested on the men's side seemed to get in (St Thomas and Luther got swapped because of the regional results, which is just a rule).

According to my computer, RIT is not unjustified since they are around 30 (although they are certainly borderline). Centre, on the other hand, was down at 45th... whereas Loras, St. Olaf, and St. Thomas seemed like very solid picks.

Based on Emory's performance at UAA, I'm guessing my rankings are a bit biased against the south since Emory was pretty close to Carnegie Mellon, but I have them ranked 25 and 37 respectively. I will, however, point out that the individual scores are not unreasonably different from how things played out at UAA - Emory just has its 4,5 buried by the deeper nationals field, since some of their top 5 from UAA fell out of their top 5 significantly at the regional meet.

26 Carnegie Mellon 752 [75, 150, 172, 177, 178, 245, 295]

37 Emory 924 [48, 156, 189, 264, 267, 288, 341]

I have that the Central Region tied with the West for the slowest regional course. The South regional (for women) was the second fastest meet all season, after the Santa Clara Bronco Invitational by 8 seconds, which was almost certainly short. The difference between the two races is around 78 seconds. Does anyone have more information about whether this seems right?

It still seems very wrong to have left out the teams from the Central region and, in my opinion, Occidental from the west region. To me it is suspicious that the teams that were left out happened to be on slower regional courses (I can at least confirm that the west regional was on the slower end). I wonder if the speed of the courses run could have biased the decisions. We will find out in a week.

(from here:

https://bijanmazaheri.wordpress.com/diii-nationals-at-large-rankings-women/

)

The Starved Elephant wrote:

mhmh wrote:

holy crap, i guess! leave it to the ncaa to create a selection system that punishes peaking. they even ran faster than haverford yesterday.

Do you think they deserved it over Santa Cruz? It had to be between those two.

yes. etown had nearly the same score gap to the champion and a much closer time gap to the champion- in tougher region that is triple the size.

the top two teams in these regions are comparable (identical?) in quality, yet in the mideast you have many many more good runners.

etown was way closer to winning their meet than ucsc. they picked the wrong team.

Well the men's 8K in the Great Lakes Region was shortened due to course conditions. Was it the same for the women? The Wartburg course wasn't fast due to snow melting amongst other things. It seems silly to compare courses and times in XC. You have to give the Region reps credit. They represented the interests of their constituents well. The Central needs to install a rep before next year. It wasn't even formally posted.

Bijan, I have a couple of questions/concerns.

1.) Parley was ahead of Corcoran in your last post. Why did that flip when Parley beat Corcoran yesterday?

2.) Why do you think your system is giving WashU such a high score? Giving Lawler and Watterson 10th and 32nd seems off.

Bijan,

Maybe you explained this already, but I was hoping you could explain your course adjustments? Where do you get the info to assume the difference in courses? For example...How is Rowan said to be 27 seconds faster than Conn College. Everyone knows Conn College is a fast course...especially this year with how dry it was and the Rowan course was accurate and packed hard as well. The Rowan course was the same exact course as was run last year at the invite and the Atlantic regional meet where Grant O'Connor won in 26:00! Because it was wet and muddy. I know it's XC and it's hard to compare, but why the assumptions that something is short because people ran fast? Couldn't we make the same argument for Conn College with how fast the times were? It must have been short!

I perform least squares a list of pairwise differences between meets to get their adjustments. That is, for every two meets I take all the people who ran in both of them and average the difference in their times. For this, I get a mean and a standard deviation. I then set up a system of equations that looks something like:

Meet A - Meet B = 30 seconds

Meet B - Meet C = 20 seconds

Meet A - Meet C = 45 seconds

...etc...

Note that because we have n variables and n choose 2 equations, the system is overdetermined, so there is no exact solution for the system of equations. This is where least squares comes in. If we write these equations in a matrix, calling d the vector of differences and m the vector of meet times, we get

A m = d

Least squares gives the solution to this equation that minimizes the squares error for each equation:

m = pseudoinverse(A) d

I also end up using standard deviations for this. This corresponds to dividing each difference equation by its standard deviation. Effectively, I “change units” so that I am minimizing the squares of the standard deviation rather than the seconds. So if I have that two meets are 30 seconds apart with standard deviation 20, my penalty for a solution that gives them 10 is not that high because it’s within reason. If the standard deviation was 2, however, a solution giving 10 would be unacceptable.

I will post updated course adjustments - In my regional rankings and initial national rankings I was taking the AVERAGE of the differences in times, which had some serious outliers. For my most recent rankings I took the MEDIAN, as well as throwing out all times over 29 min for men and over 25 min for women, since there were a few huge tanks that were skewing averages.

Hopefully this is more robust. I know there was a fall at Conn, which may have created outliers making the course look slower. I currently still have Conn as 25 seconds slower than Rowan. I'm not sure your stat about O'Connor winning in 26:00 is helpful - what were the conditions then? People ran fast at Rowan this year. You cannot deny that. Two people who have not broken 30:00 in the 10k and are 14:30-40 types from last season ran 24:00. I think the difference checks out to be reasonable.

Why do we have to explain this every year. Repeat after me: teams cannot be “pushed in” in division 3 selection. Nothing St Thomas did is relevant until Loras is selected. Loras didn’t do anything to earn selection, so they were not selected, and so St Thomas was never considered. The committee only considers wins and comparisons against “teams already in the championship field” so Loras is unable to benefit from the out of region wins St Thomas had throughout the season.

Certainly many valid critiques of how the system is set up, but if you’re going to complain about the choices made by the committee you should make an effort to read the criteria and understand how they operate first. This discussion happens on this day every year and people don’t learn. If you’re mad about the way it played out, come up with a better system and propose it at convention.

Blarg wrote:

"more competitive"

Loras lost head to heads against almost the whole Great Lakes Region.

Loras, at Oberlin lost to:

Oberlin - 3rd GL

John Carroll - 1st GL

Baldwin Wallace - 2nd GL

Otterbein - 8th GL

Ohio Northern - 7th GL

Case Western - 9th GL

Occidental -4th in the West

Now, If head to head regional competition is a determining factor for selection explain how you can take Loras without taking all these teams? You want to take 9 teams from the Great Lakes?

You can't take a 4th or 5th place team without taking the third place team.

Don't run like garbage all year, don't get beat by teams that have run like garbage all year at Regionals.

So what's the point of running Regionals then if they're just gonna look at results from the regular season for their decision? Isn't the point to peak for the end of season? That's what Loras did, but clearly all that matters to you and the selection committee is what happened at the Oberlin meet

Thank you for clarifying...makes sense.

As for your own comment about them only being 14:30 guys and not breaking 30 you could say the same about Ryan...broke 1430 once. And what about Mcaneny...similar stats to Ryan (actually faster by 2 seconds over 1500) and he was 17 seconds behind those guys.

I know it's hard with it being XC...next week will show everything.

Washed-up Grad Student wrote:

People ran fast at Rowan this year. You cannot deny that. Two people who have not broken 30:00 in the 10k and are 14:30-40 types from last season ran 24:00. I think the difference checks out to be reasonable.

Patrick Watson is a legitimate individual championship contender and Pango was AA last year. I think you ran slower than Jared and then got 15th at cross in your Junior year (which is worse than his sophomore year). Watson was AA in steeplechase. These are two really good athletes that ran well and ahead of almost all of the field. McAneny (3rd at Rowan) ran faster at Paul Short than at Rowan. It doesn't say what you think it does about the course.

Actually, you're right. Looking back at results, the reason UChicago and WashU got in that year after getting 7th and 8th place was because they beat Emory handily at UAAs., and then Emory auto-qualified out of the South. Still amazing that they took 8 teams from one region. 5th place Plateville and 6th place Augustana also beat Emory at Oshkosh that year. So that year, the South region's ineptitude was responsible for 8 teams making it out of one region. No one had to be pushed.

So I guess we have our culprit. Loras doing well at regionals screwed the deserving teams behind them.