An attempt to preduct ICT spots

[Thalaba the Destroyer]

My attempt to make ICT predictions is based on a system in which 100 is average performance and 200 is roughly the highest possible score. It is based on performance relative to average on correct tossups, bonuses, and total points per game. Corrections for field strength are incorporated in the PPG comparison. The results for the SCT played on DII questions are probably not great, since I didn't make much of an effort to adjust them (but they are somewhat adjusted).

Automatic qualifiers ("S-Values" in parentheses)

1. Brown A (209)
2. Minnesota A (205)
3. Illinois A (198)
4. Cal Irvine (179)
5. Harvard B (166)
6. Carleton A (164)
7. Maryland (159)
8. Toronto B (153)
9. Florida A (139)
10. Michigan (138)
11. Illinois B (129)
12. Carnegie Mellon (129)
13. Texas A (124)
14. VCU (118)
15. Chicago A [Michigan site] (108)
16. Alabama A (106)
17. Washington A (88)
18. Gonzaga (77)

Wild-card teams

19. Minnesota B (191)
20. Chicago [WUSTL site] (178)
21. Stanford A (176)
22. Iowa (163)
23. Stanford B (156)
24. Dartmouth A (154)
25. Harvard A (153)
26. Florida State A (150)
27. Princeton A (147)
28. Harvard C (144)
29. Penn (132)
30. Chicago B [Michigan site] (126)
31. Cornell A (126)
32. Dartmouth B (125)

Wait List

1. Truman State (109)
2. Brandeis A (108)
3. Missouri State (106)
4. Toronto A (106)
5. South Carolina A (104)
6. Florida B (104)
7. Northwestern A (101)
8. Grinnell A (100)
9. Lawrence (99)
10. Yale A (98)
11. Oklahoma A (96)

Top 50 teams

1. Brown A (209)
2. Minnesota A (205)
3. Illinois A (198)
4. Minnesota B (191)
5. Cal Irvine (179)
6. Chicago [WUSTL site] (178)
7. Stanford A (176)
8. Harvard B (166)
9. Iowa (163)
10. Carleton A (164)
11. Maryland (159)
12. Stanford B (156)
13. Dartmouth A (154)
14. Toronto B (153)
15. Harvard A (153)
16. Florida State A (150)
17. Princeton A (147)
18. Harvard C (144)
19. Florida A (139)
20. Michigan (138)
21. Penn (132)
22. Illinois B (129)
23. Carnegie Mellon (129)
24. Chicago B [Michigan site] (126)
25. Cornell A (126)
26. Dartmouth B (125)
27. Texas A (124)
28. VCU (118)
29. Truman State (109)
30. Chicago A [Michigan site] (108)
31. Brandeis A (108)
32. Missouri State (106)
33. Alabama A (106)
34. Toronto A (106)
35. South Carolina A (104)
36. Florida B (104)
37. Northwestern A (101)
38. Grinnell A (100)
39. Lawrence (99)
40. Yale A (98)
41. Oklahoma A (96)
42. Rutgers A (93)
43. Tulane A (89)
44. Washington A (88)
45. Harvard D (85)
46. Wisconsin (85)
47. Duke (85)
48. Cornell B (84)
49. Queen's A (84)
50. William and Mary (83)

[theMoMA]

Just posting to say that I looked up the IP of this person, and it was the same as some other hilarious anonymous posts from HSQB past; unfortunately, when I dug further, I found out that this is because all were posted from the IP hider known as "hidemyass.com" This post is posted using hidemyass just to prove the veracity of this.

For the record, I did not make these, "preductions." Carry on.

For further info on this IP, see: http://en.wikipedia.org/wiki/User_talk:67.159.44.138

[Sen. Estes Kefauver (D-TN)]

Hosts also get bids, remember.

[Matt Weiner]

You have a few teams in the wrong division, and a few teams listed as auto-qualifiers who are not auto-qualifiers according to the listed NAQT rules (due to having less than four teams in their division at the tournament) though they are (erroneously?) listed as auto-qualifiers under the color code system on the naqt.com stats.

[DumbJaques]

Also some of those teams are div 2.

[Ondes Martenot]

Since I know basically know nothing about this "S Value", what are RPI's chances of getting a DII invitation?

[theMoMA]

Since I know basically know nothing about this "S Value", what are RPI's chances of getting a DII invitation?

Seems pretty good. You guys finished tied for first in record, had the highest PP20TUH, and fourth-highest bonus conversion at what appears to be the hardest DII sectional. I haven't really looked in depth at the results for either DI or DII yet, but from a quick perusal, it looks like any DII team with 200+ PP20TUH and a 15+ bonus conversion is at least on the bubble, and anyone with significantly better numbers should be in.

[grapesmoker]

Since I know basically know nothing about this "S Value", what are RPI's chances of getting a DII invitation?

I'm pretty sure the DII sectional at MIT was among the strongest in the nation, so I'm guessing you'll get an invite.

[Skepticism and Animal Feed]

Doesn't Matt have an elaborate formula that he uses each year to make predictions? When will he share his results.

[Matt Weiner]

Doesn't Matt have an elaborate formula that he uses each year to make predictions? When will he share his results.

My formula hasn't been so accurate in the past; while I've gotten 90% or so of the 14-18 non-host, non-automatic bids right, that's not significantly more than one could do by eyeballing it. In other words, it's pretty obvious that teams like Iowa, Stanford A, and Florida State, to take some examples of non-automatic-qualifiers from this year, are going to get invites under a formula that incorporates actual statistics at all, no matter how it does so. The 3 or 4 bubble teams are the only test we have of how well any attempt to figure out the S-value is doing. No one's attempt has ever gotten all of them right, and that's a pretty small range to test adjustments to the formula.

I can tell you who the top teams purely on some rough metric of bonus conversion and power percentage are, since we know those are components of the S-value, and I can also tell you some things that doing this has reminded me, which are:

1) There's still no reason for S-value to be secret; NAQT should publish the formula.
2) There appears to be no reliable way to account for the fact that some D1 teams played on D2 questions and/or in combined fields with D2 opponents; NAQT should prohibit this from happening again. If teams end up in an SCT like Michigan's that had two D1 teams, they could do something such as play 6 games against each other to generate qualifying stats, then enter the D2 field as exhibition teams for the afternoon to get a more varied schedule. While I know that's suboptimal, it's less so than excluding some D1 team from SCT because someone who did get a spot generated all their numbers by playing D2 questions against Freshman Team C.
3) NAQT has a bunch of teams' names highlighted in green on the official SCT results pages, despite the fact that they should not have earned automatic bids based on the rule stating that their must be four teams in {Division I, UG, Division II} playing the tournament for the winner of the division to get an autobid. The below results assume that this is an error by whoever put the stats pages up, rather than a sudden change in policy, and will not consider autobids from those divisions.

Note also that I am guessing at which division most host bids will be taken in, and that Carleton and Texas lost their host bids by qualifying teams outright for Division I.

ICT D1 BID PREDICTIONS BY BONUS/POWER PCT (this is simply the fraction of available bonus points a team answered, plus the fraction of tossups heard a team powered, divided by two; the minimum is 0 and the maximum is 1; 1 would be a team who powered every tossup it heard and 30ed every bonus.)

HOST BIDS:

1. Georgia
2. Michigan
3. MIT
4. UCLA
5. WUSTL

AUTO-BIDS:

6. Alabama A (won Miss St SCT)
7. Brown (won MIT SCT)
8. Carleton A (won UG Carleton SCT)
9. Carnegie Mellon A (won Pitt SCT)
10. Chicago A (won Michigan SCT)
11. Florida A (won Georgia SCT)
12. Harvard B (won UG MIT SCT)
13. Illinois A (won WUSTL SCT)
14. Illinois B (won UG WUSTL SCT)
15. Maryland (won W&M SCT)
16. Minnesota A (won Carleton SCT)
17. UC Irvine (won UCLA SCT)
18. Texas A (won Texas SCT)
19. Toronto B (won Laurier SCT)
20. Virginia Commonwealth (won UG W&M SCT)

D1 WILD-CARDS (12)

I've left the 15 auto-qualifiers in here so you can see their rating/ranking too.

Brown 0.4810
Minnesota A 0.4424
Illinois A 0.4393
21. Stanford A 0.3959
22. Minnesota B 0.3925
Harvard B 0.3789
23. Iowa 0.3762
24. Chicago (WUSTL site) 0.3746
Maryland 0.3599
UC Irvine 0.3576
Toronto B 0.3463
25. Harvard A 0.3411
26. Dartmouth A 0.3342
Carleton College A 0.3322
27. Stanford B 0.3282
Chicago A (Michigan) 0.3275
28. Harvard C 0.3257
29. Florida State 0.3246
Florida A 0.3161
30. Penn 0.2982
31. Princeton 0.2923
Illinois B 0.2826
Carnegie Mellon A 0.2803
32. Cornell A 0.2792
South Carolina A 0.2671
(34) Virginia Comm. 0.2664
(35) Texas A 0.2596
(36) Alabama A 0.2593

South Carolina has better stats than the last three auto-qualifiers, but is projected to be the first waitlist team. The top 10 on the waitlist are projected to be:

33. South Carolina A 0.2671
34. Dartmouth B 0.2590
35. Toronto A 0.2469
36. Florida B 0.2429
37. Ill-Chicago 0.2405
38. Missouri State 0.2372
39. Grinnell A 0.2348
40. Brandeis 0.2345
41. Lawrence 0.2302
42. Yale 0.2235

Note that both Chicago A (Michigan) who is listed as qualifying, and Ill-Chicago, who is listed as being on the waitlist, played D2 questions against D2 opponents; their actual rating by NAQT will likely penalize them for this, though Chicago A will qualify in any case due to the editing deal. Texas A played D1 questions against D2 opponents, but they auto-qualified and no other D1 teams from that SCT appear to be in contention for a bid, so it will not affect anything. Oklahoma A at 0.2061 is #46 on my list and is the next team out of that SCT. The only team that appears to be DI at the Boise State site, which played D2 questions in a common field, is Washington A, which had a rating of 0.1805 and is ranked #59.

D2 BID PREDICTIONS:

HOST BIDS:

1. Boise State
2. Pitt
3. Mississippi State
4. Wilfred Laurier
5. William & Mary

RESERVED SPOTS FOR CC TEAMS:

These spots will be determined at the CCCT on February 28.

6.
7.
8.
9.
10.
11.

AUTO-BIDS:

12. Alabama C (won Miss St SCT)
13. Carleton C (won Carleton SCT)
14. Chicago (won WUSTL SCT)
15. Gonzaga (won Boise State SCT)
16. McMaster A (won Laurier SCT)
17. Michigan (won Michigan SCT)
18. Ohio State (won Pitt SCT)
19. UCLA (won UCLA SCT)
20. Virginia (won W&M SCT)
21. Yale A (won MIT SCT)

D2 WILD-CARDS (11)

I've left the 10 auto-qualifiers in here so you can see their rating/ranking too.

Chicago (WUSTL) 0.4581
Michigan 0.4316
22. Chicago B (Michigan) 0.4183
Yale A 0.3786
23. Missouri 0.3720
24. Yale B 0.3635
25. Princeton 0.3607
26. RPI 0.3391
Virginia 0.3306
Carleton C 0.3292
27. Northwestern A 0.3272
28. CUNY-Hunter 0.3264
Ohio State 0.3246
29. Arizona State A 0.3213
30. Case Western 0.3171
31. WUSTL 0.3128
32. Maryland B 0.3127
Alabama C 0.3124
Davidson B 0.3098
NYU 0.3095
(36) McMaster A 0.3052
...
(43) UCLA 0.2883
...
(92) Gonzaga 0.1762

Davidson B and NYU have better stats than three auto-qualifiers, but are projected to be the first two waitlist teams. The top 10 on the waitlist are projected to be:

33. Davidson B 0.3098
34. NYU 0.3095
35. Miami A 0.3033
36. Cornell C 0.3032
37. Grinnell B 0.3023
xx. Georgia 0.3020
38. Columbia A 0.2919
39. Caltech B 0.2914
40. Toronto A 0.2856
41. Davidson A 0.2786
42. Mississippi State B 0.2778

Note that, at this point, it's unclear who if anybody will be receiving an auto-bid from D2 of the Georgia SCT. You can see Georgia's ranking on the waitlist; the next team in the standings was Vanderbilt B, which would be team #45 (0.2737) otherwise. If either of those teams gets a bid, then Maryland B is knocked to the first spot on the waitlist and everyone else on the waitlist moves one spot down.

Note also that all D2 teams at Michigan, Texas, and Boise State played at least some of their games against D1 opponents, and the Texas teams did so on D1 questions, so they will probably get some boost in their ratings.

So, there you have it.

[marnold]

I'm not entirely sure how this changes things, but I guess I should throw out that we (Chicago) get an automatic second bid as long as the WUSTL A-team gets a bid (which of course it will) per this. I'm not sure if they count this ghost bid as an autobid as if team nessermarnold played a full D1 field in Michigan, whether it's treated like a host bid or whether it's an additional at-large or what. I suppose the answer to that question changes the waitlist scenarios.

[Matt Weiner]

I'm not entirely sure how this changes things, but I guess I should throw out that we (Chicago) get an automatic second bid as long as the WUSTL A-team gets a bid (which of course it will) per this. I'm not sure if they count this ghost bid as an autobid as if team nessermarnold played a full D1 field in Michigan, whether its treated like a host bid or whether its an additional at-large or what. I suppose the answer to that question changes the waitlist scenarios.

Thanks for reminding me about this; I've updated the post accordingly. Since Chicago at the WUSTL site certainly qualified by the eyeball method, both Chicago teams will definitely get bids. This means that the order of bids/waitlist teams in my projection does not change, since my original post had both teams getting bids based on their stats.

[Important Bird Area]

3) NAQT has a bunch of teams' names highlighted in green on the official SCT results pages, despite the fact that they should not have earned automatic bids based on the rule stating that their must be four teams in {Division I, UG, Division II} playing the tournament for the winner of the division to get an autobid. The below results assume that this is an error by whoever put the stats pages up, rather than a sudden change in policy, and will not consider autobids from those divisions.

R. has just fixed the results for the WUSTL SCT so that they properly list Illinois B's autobid.

Harvard B should have an autobid for being the top-ranking UG team at the MIT site. (There were eight UG teams there: Harvard A, B, D, Dartmouth A, B, Brandeis, Yale, Rutgers.)

[Matt Weiner]

R. has just fixed the results for the WUSTL SCT so that they properly list Illinois B's autobid.

Harvard B should have an autobid for being the top-ranking UG team at the MIT site. (There were eight UG teams there: Harvard A, B, D, Dartmouth A, B, Brandeis, Yale, Rutgers.)

Noted and fixed. I still have questions about two SCT results:

Are these highlights correct, and if so, in what division(s) did each team qualify? http://naqt.com/stats/tournament-teams. ... nt_id=2860

Who is the actual D2 qualifier here? http://naqt.com/stats/tournament-teams. ... nt_id=2854

[Important Bird Area]

According to the Northwest SCT results thread, Gonzaga was the DII champion (8-2 in double round robin, vs. Boise State A's 6-4).

Also, Boise State will receive a host bid, which will likely be used in Division II.

I've asked for R. for clarification re: the status of Washington.

[olsb25]

Princeton was also all UG at MIT.

[lasercats]

How are DII bids generated among the tournaments that had to combine divisions? (like Texas SCT) Our TD explained that there was a system, but not how it would work.

[Important Bird Area]

How are DII bids generated among the tournaments that had to combine divisions? (like Texas SCT) Our TD explained that there was a system, but not how it would work.

If there are at least four DII teams, then the highest-ranking one gets an autobid.

http://www.naqt.com/ict/qualify.html

[rhentzel]

You have a few teams in the wrong division, and a few teams listed as auto-qualifiers who are not auto-qualifiers according to the listed NAQT rules (due to having less than four teams in their division at the tournament) though they are (erroneously?) listed as auto-qualifiers under the color code system on the naqt.com stats.

Unfortunately for the ideal of clear communication, the page on NAQT's site describing ICT qualification:

http://www.naqt.com/ict/qualify.html

was written under the assumption that DI and DII competed separately (as we used to require).

Here is the rule used when there is a combined division:

1. If there aren't at least four total teams, there are no champion bids.

2. If there are at least four total teams, then the champion receives an invitation. If, in addition, there are at least four DI, DII, or UG teams, then the champion in those (sub)divisions also receives an invitation.

To the best of my memory, NAQT has always applied this rule to every combined sectional field in the past.

That page also doesn't have the correct rules for hosts' automatic bids, which were changed this year. The correct policy can be found here:

http://www.naqt.com/sct/automatic-bid.html

And, of course, don't forget the "Seth Teitler clause":

http://www.naqt.com/sct/2009/editing.html

This seems like a good a place as any to mention one clarification that we will be using this year: If a school hosts a sectional and enters a Division I team that doesn't qualify in the initial set of bids, then chooses to take its hosting autobid in Division II, but declined invitations result in the Division I team getting a bid after all, then NAQT will not dissolve the hosting autobid as would be suggested by the text of our rules. The school will end up with bids in both Division I and Division II. We don't anticipate this happening this year, but we wanted to state a policy in advance, particularly since withdrawing the invitation would strike us as extremely unfair.

[Matt Weiner]

Hey, a question: If any of the top 6 finishers from the CCCT decline their bids to ICT, will their spots be given to the next finishers at the CC tournament, or will they become general spots for the D2 waitlist?

[rhentzel]

Hey, a question: If any of the top 6 finishers from the CCCT decline their bids to ICT, will their spots be given to the next finishers at the CC tournament, or will they become general spots for the D2 waitlist?

Following the completion of the CCCT, S-values will be calculated for the CC teams in attendance and they will be interpolated into the existing Division II waitlist in the proper place.

NAQT will continue to invite four-year teams off the purely four-year Division II waitlist until the end of February 27; starting on March 1 we will begin issuing invitations off the combined waitlist.