USACFC Competition Schedule and Participants List

[oiuyt]

Originally Posted by mrbiggs

Also, we're seeding by whole team and not weapon, right? I think by weapon would be a bit more accurate, but it's clearly too late for that at this point. (Otherwise the results for each weapon as well as individuals could be very skewed).

Where conferences split the results by squad that's what should be used in the seeding. Most, if not all, of the conferences mentioned include such rankings, and some generate the 3W/6W results by compilation of said rankings.

-B

edit: ACK! I keep promising myself to stay OUT of these threads....

[mrbiggs]

Originally Posted by oiuyt

Where conferences split the results by squad that's what should be used in the seeding. Most, if not all, of the conferences mentioned include such rankings, and some generate the 3W/6W results by compilation of said rankings.

Oh, ok, thanks. That's good news.

[noodle]

anyway, for reference, i did my own seeding algorithm and used 2003-2006 to try and predict the place of teams with history in the 2007 event. here are results. the "28", i.e. last place, signifies that they had no history.

School			Act.	Pred.
Arizona State		11	13.39130436
Army			2	3.525362323
Central Michigan	21	25.47391304
Chicago			10	13.51304347
Clemson			5	11.33146135
Dartmouth		8	18.02190016
Florida			3	1.743236719
Illinois		23	28
Miami of Ohio		27	28
Michigan State		7	7.232487927
Navy			24	23.01545894
Purdue			26	18.94649759
Smith			11	8.699698063
SUNY Stony Brook	20	28
Texas			13	8.751268113
Texas A&M		15	18.22028985
U Virginia		18	18.23085749
U Wisconsin		17	23.70756843
U. Indiana		6	16.58695653
U. Maryland		13	16.29106281
U. Michigan		1	7.943055554
U. Minnesota		25	28
UCLA			28	28
UMASS			9	3.222282609
UNH			4	4.270169078
Virginia Tech		16	14.92995169
William And Mary	19	28
Xavier			22	28

note that while not exactly accurate, the predicted results are in a general neighborhood of correctness. you can more or less predict a team's success within +-4, with some outliers. which, to me, is plenty enough to seed people. also, the format changed, which will jumble results unavoidably.

[Wafath]

Originally Posted by noodle

note that while not exactly accurate, the predicted results are in a general neighborhood of correctness. you can more or less predict a team's success within +-4, with some outliers. which, to me, is plenty enough to seed people. also, the format changed, which will jumble results unavoidably.

BTW, it would look like you only used the women's results for this. How does it look for the men? How about each squad? 6W?

W

[noodle]

Originally Posted by Wafath

BTW, it would look like you only used the women's results for this. How does it look for the men? How about each squad? 6W?

W

yes, only women's squad.

its not very easy to just do each one of them for fun, since i have to parse through the strings in the results to format things correctly to run it.


i'll do men's and 6w.

[noodle]

same process, men's 3w. 30 indicates last, indicates no past data. some of the higher outliers can be attributed to format change, i think.

Arizona State		12	22.7586207
Army			2	1.651785713
Central Michigan	27	24.39655172
Chicago			17	18.98706897
Clemson			10	8.984759858
Cornell			13	2.42816092
Dartmouth		10	11.65024631
Florida			5	8.17195197
Illinois		22	30
Indiana			7	19.52586207
Maryland		4	18.69868638
Miami of Ohio		29	30
Michigan		6	3.3682266
Michigan State		16	13.2281404
Minnesota		26	21.24999999
Navy			21	17.75451559
Northwestern		9	3.034174875
Purdue			23	12.77093597
SUNY Stony Brook	18	30
Texas			8	8.601447038
Texas A&M		15	13.59195402
UC Berkeley		14	30
UCLA			30	30
UMASS			3	2.226754928
UNH			19	15.75969827
Virginia		25	20.34944582
Virginia Tech		24	20.74122537
William And Mary	1	10.34482758
Wisconsin		20	15.51724137
Xavier			28	30

[noodle]

same data for 6wt:

Arizona State		12	16.53333332
Army			2	1.522370762
Central Michigan	26	27.69333334
Clemson			7	8.80875223
Cornell			23	14.6169697
Dartmouth		8	13.00747474
Florida			3	2.865427805
Illinois		22	31
Miami of Ohio		30	31
Michigan State		13	7.699714799
Navy			21	19.44545455
Northwestern		16	14.23872549
Purdue			29	14.49180927
Smith			19	22.57695187
SUNY Stony Brook	17	31
Texas			10	5.712344022
Texas A&M		15	14.06954248
U of Chicago		14	12.71
U Virginia		25	20.68434937
U Wisconsin		18	29.59090911
U. Indiana		5	15.60333334
U. Maryland		6	17.55193108
U. Michigan		1	9.323208561
U. Minnesota		28	26.04
UC Berkeley		24	31
UCLA			31	31
UMASS			4	0.826666669
UNH			11	7.220347594
Virginia Tech		20	17.24975936
William And Mary	9	17.56666668
Xavier			27	31

[KD5MDK]

Originally Posted by noodle

anyway, for reference, i did my own seeding algorithm and used 2003-2006 to try and predict the place of teams with history in the 2007 event. here are results. the "28", i.e. last place, signifies that they had no history.

School			Act.	Pred.
Arizona State		11	13.39130436
Army			2	3.525362323
Central Michigan	21	25.47391304
Chicago			10	13.51304347
Clemson			5	11.33146135
Dartmouth		8	18.02190016
Florida			3	1.743236719
Illinois		23	28
Miami of Ohio		27	28
Michigan State		7	7.232487927
Navy			24	23.01545894
Purdue			26	18.94649759
Smith			11	8.699698063
SUNY Stony Brook	20	28
Texas			13	8.751268113
Texas A&M		15	18.22028985
U Virginia		18	18.23085749
U Wisconsin		17	23.70756843
U. Indiana		6	16.58695653
U. Maryland		13	16.29106281
U. Michigan		1	7.943055554
U. Minnesota		25	28
UCLA			28	28
UMASS			9	3.222282609
UNH			4	4.270169078
Virginia Tech		16	14.92995169
William And Mary	19	28
Xavier			22	28

note that while not exactly accurate, the predicted results are in a general neighborhood of correctness. you can more or less predict a team's success within +-4, with some outliers. which, to me, is plenty enough to seed people. also, the format changed, which will jumble results unavoidably.

By my quick visual count, 14/21 ranked schools placed within 4.5 places (to account for your funny rounding) of where you predicted. 67% accuracy (assuming you accept the +-4.5 as an acceptable range) is pretty cool. Let me check the others.

[KD5MDK]

Originally Posted by noodle

same process, men's 3w. 30 indicates last, indicates no past data. some of the higher outliers can be attributed to format change, i think.

Arizona State		12	22.7586207
Army			2	1.651785713
Central Michigan	27	24.39655172
Chicago			17	18.98706897
Clemson			10	8.984759858
Cornell			13	2.42816092
Dartmouth		10	11.65024631
Florida			5	8.17195197
Illinois		22	30
Indiana			7	19.52586207
Maryland		4	18.69868638
Miami of Ohio		29	30
Michigan		6	3.3682266
Michigan State		16	13.2281404
Minnesota		26	21.24999999
Navy			21	17.75451559
Northwestern		9	3.034174875
Purdue			23	12.77093597
SUNY Stony Brook	18	30
Texas			8	8.601447038
Texas A&M		15	13.59195402
UC Berkeley		14	30
UCLA			30	30
UMASS			3	2.226754928
UNH			19	15.75969827
Virginia		25	20.34944582
Virginia Tech		24	20.74122537
William And Mary	1	10.34482758
Wisconsin		20	15.51724137
Xavier			28	30

14/24. 2 came really close but just out. 58%

Originally Posted by noodle

same data for 6wt:

Arizona State		12	16.53333332
Army			2	1.522370762
Central Michigan	26	27.69333334
Clemson			7	8.80875223
Cornell			23	14.6169697
Dartmouth		8	13.00747474
Florida			3	2.865427805
Illinois		22	31
Miami of Ohio		30	31
Michigan State		13	7.699714799
Navy			21	19.44545455
Northwestern		16	14.23872549
Purdue			29	14.49180927
Smith			19	22.57695187
SUNY Stony Brook	17	31
Texas			10	5.712344022
Texas A&M		15	14.06954248
U of Chicago		14	12.71
U Virginia		25	20.68434937
U Wisconsin		18	29.59090911
U. Indiana		5	15.60333334
U. Maryland		6	17.55193108
U. Michigan		1	9.323208561
U. Minnesota		28	26.04
UC Berkeley		24	31
UCLA			31	31
UMASS			4	0.826666669
UNH			11	7.220347594
Virginia Tech		20	17.24975936
William And Mary	9	17.56666668
Xavier			27	31

15/25 or 60%


Does anybody have rough numbers on how well other seeding forms come to predicting final results? How well did the initial seeding last year compare to final results?

[noodle]

Originally Posted by KD5MDK

By my quick visual count, 14/21 ranked schools placed within 4.5 places (to account for your funny rounding) of where you predicted. 67% accuracy (assuming you accept the +-4.5 as an acceptable range) is pretty cool. Let me check the others.

also kind of depends on how you define "acceptible". one could take a look to see how accurate the numbers are just considering brackets as cut points, top 8, top 16, top 32.

[noodle]

Originally Posted by KD5MDK

14/24. 2 came really close but just out. 58%



15/25 or 60%


Does anybody have rough numbers on how well other seeding forms come to predicting final results? How well did the initial seeding last year compare to final results?

don't forget to note the fact that i have no ranking system for new teams, i just throw them at the bottom.

and note that this is still pretty decent way of doing it, maintaining approximately the same average overall accuracy.

[noodle]

also note: i might've forgotten to add a 1 on the last two produced results :X but i dont' think it matters that much, if there's a acceptable range of something more than 2.

[oiuyt]

Originally Posted by noodle

same data for 6wt:

<SNIP>
UMASS			4	0.826666669

UMass was expected to do slightly BETTER than 1st place?

I'm not sure your algorithm isn't somewhat flawed.

Originally Posted by KD5MDK

How well did the initial seeding last year compare to final results?

Good question. It wasn't publically released last year. Prior to that the tournament was conducted as a complete round-robin and seeding therefore wasn't necessary.

The seeding was done by squad, not by 3W or 6W team. It assigned ordinal rankings to teams within their conference (side note: as previously mentioned I had MAJOR issues with how teams were assigned which conference would count -- Smith was not considered a New England school, for example). The #1 teams were roughly ordered and then distributed, then the #2 teams were ordered and distributed, then the #3 teams, etc.

The pools were announced, after having been put into alphabetical order, so as to mask the seeding used.

Recovering the seeding used would likely prove somewhat difficult. Even which conference result was used to rank each team was hidden information.

-B

edit: &(*^$@# I need an ignore thread function. I WILL stay out of this thread. (at least until tomorrow when someone posts something interesting again and sucks me back in)

[TooLoftheDeviL]

Originally Posted by oiuyt

edit: &(*^$@# I need an ignore thread function. I WILL stay out of this thread. (at least until tomorrow when someone posts something interesting again and sucks me back in)

Don't leave us, Brad

We know it's frustrating, but it'd be super awesome to come up with some stuff now to seed the thing ahead of time. Piece together some projected pools based of different methods, see what pops out and maybe see if we can't guess the end results

-

By chance, does anyone know how we might go about making a motion to actually implement some seeding algorithm for this years championship to avoid secret deliberations and such?

[noodle]

Originally Posted by oiuyt

UMass was expected to do slightly BETTER than 1st place?

I'm not sure your algorithm isn't somewhat flawed.

like i said, i think i forgot to add a 1 in the last two results

[KD5MDK]

Originally Posted by noodle

don't forget to note the fact that i have no ranking system for new teams, i just throw them at the bottom.

I removed schools with no history prediction from my calculation. Thus, in 6WT, I said 15/25, even though there were 31 teams listed, to avoid counting those schools you couldn't attempt to predict.

[noodle]

Originally Posted by KD5MDK

I removed schools with no history prediction from my calculation. Thus, in 6WT, I said 15/25, even though there were 31 teams listed, to avoid counting those schools you couldn't attempt to predict.

well, i guess i forgot to check to see if you accounted for that or not

[DJ Apostrophe]

Originally Posted by oiuyt

UMass was expected to do slightly BETTER than 1st place?

I'm not sure your algorithm isn't somewhat flawed.

You know Brad... seeing a comment like this, you could just smile and sit all smug... because the success UMASS has had at this event can be largely attributed to you... You were a big part of getting the event started working with Bruce, and you were a big part of how the team functioned, worked, and succeeded... I'm sure I'm not alone in appreciating what you did for our program.

That said, I understand what you were saying... but I felt like you needed a misty eyed thank you.

-w