Likelihood of NAQT Champion.

Old college threads.
User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Likelihood of NAQT Champion.

Post by Peter13 » Sat Jan 18, 2014 3:51 am

I was randomly looking over D-values and the like, while somehow simultaneously looking at odds for some games. And this had me thinking, what would the odds for each team winning be? I know everyone picks and chooses there favourites, and Yale, Virginia and Penn are the favourites, but what is the actual percentage of any team winning? I know that this might seem odd, but I want to look at the statistical nature of this. Also, are there any dark horses who might come out and surprise us? Maryland have been so close only to lose in playoffs. Anyway, what do you think?
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
Periplus of the Erythraean Sea
Auron
Posts: 1942
Joined: Mon Feb 28, 2011 11:53 pm
Location: New York, NY

Re: Likelihood of NAQT Champion.

Post by Periplus of the Erythraean Sea » Sat Jan 18, 2014 4:52 pm

Illinois had a very good showing as a "dark horse" in 2012, so it's definitely possible.

As for this year, I think both Chicago and Maryland have a chance of making a strong showing. Chicago's biggest strength is in the arts, which gets diminished in the NAQT distribution, but Marshall does well on topics NAQT is heavy on like current events, so it should balance out. Maryland looks like it's in something of a renaissance, with Jordan Brownstein arguably being the breakout player of the year, so it'd be no surprised if they did well.

We haven't seen a full Illinois team yet, and Harvard's been able to beat Penn and Yale. I don't think they're title contenders, though, and unless Kurtis goes back to Michigan sometime this year we're probably looking at UVA/Yale/Penn/Chicago/UMD.
Will Alston
Bethesda Chevy Chase HS '12, Dartmouth '16, Columbia Business School '21
NAQT Writer and Subject Editor

User avatar
Auks Ran Ova
Forums Staff: Chief Administrator
Posts: 4042
Joined: Sun Apr 30, 2006 10:28 pm
Location: Minneapolis
Contact:

Re: Likelihood of NAQT Champion.

Post by Auks Ran Ova » Sat Jan 18, 2014 5:38 pm

Peter13 wrote:what is the actual percentage of any team winning?
Somewhere around 100%, I imagine!
Rob Carson
University of Minnesota '11, MCTC '??
Member, ACF
Member, PACE
Writer and Editor, NAQT

User avatar
Cheynem
Sin
Posts: 6557
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

Re: Likelihood of NAQT Champion.

Post by Cheynem » Sat Jan 18, 2014 5:39 pm

Well, now "vacated" won two straight ICT's, so not so fast my friend.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Sun Jan 19, 2014 3:57 am

Ukonvasara wrote:
Peter13 wrote:what is the actual percentage of any team winning?
Somewhere around 100%, I imagine!
Haha, very funny. I mean like a breakdown chart of the chart. I did this once for the world cup by looking at articles, teams form, bookmakers, and highly valueable opinions (not like the media talking heads, real statistical analysis). Since quiz bowl is even more about statistics, I thought it would be fun to make percentages (or inversely odds) for each team. The problem is that I am terrible of knowing teams outside the top bracket, so if you could give me your best guesses based on mathematical analysis or just plain opinion, it would be fun to see what everyone thinks, not just ranking wise, but actual distribution too. What is the likelihood of a lowly team winning it all (and are their chances better than Algeria winning the World Cup?).
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
ThisIsMyUsername
Yuna
Posts: 803
Joined: Wed Jul 15, 2009 11:36 am
Location: New York, NY

Re: Likelihood of NAQT Champion.

Post by ThisIsMyUsername » Sun Jan 19, 2014 10:57 am

Peter13 wrote:
Ukonvasara wrote:
Peter13 wrote:what is the actual percentage of any team winning?
Somewhere around 100%, I imagine!
Haha, very funny. I mean like a breakdown chart of the chart. I did this once for the world cup by looking at articles, teams form, bookmakers, and highly valueable opinions (not like the media talking heads, real statistical analysis). Since quiz bowl is even more about statistics, I thought it would be fun to make percentages (or inversely odds) for each team. The problem is that I am terrible of knowing teams outside the top bracket, so if you could give me your best guesses based on mathematical analysis or just plain opinion, it would be fun to see what everyone thinks, not just ranking wise, but actual distribution too. What is the likelihood of a lowly team winning it all (and are their chances better than Algeria winning the World Cup?).
This doesn't really reflect how national quizbowl tournaments work. The chance of any team besides Yale, UVA, Penn, Chicago, Maryland, and (maybe) Harvard winning ICT is basically zero. Collegiate quizbowl tournaments do not use single-elimination formats; even our playoffs are structured as round robins. Lesser teams often eke out "statistical fluke" victories over better teams, but this is not enough to propel the lesser team to the championship, because that team would have to win almost every single other match it plays. The effect of most upsets is to damage the better team's chance of becoming the champion far more than it ever bolsters the lesser team's chance of becoming the champion. Within the time I have played quizbowl, there have been just a few "surprise" champions, but these "surprise" champions still emerged from the upper tier of the top bracket (by "surprise" we meant: "not one of the top two teams we expected to be in the final together", not: "a team that no one thought was any good"). If a team not currently in the upper tier were to transform itself into a championship team between now and ICT, it would have to be through a rather intense training regimen. And by virtue of the fact that the purpose of such a regimen would be to push their performance beyond the levels they have displayed at tournaments this year, such results would necessarily be impossible to predict from current statistics.
John Lawrence
Yale University '12
King's College London '13
University of Chicago '19

“I am not absentminded. It is the presence of mind that makes me unaware of everything else.” - G.K. Chesterton

User avatar
bmcke
Wakka
Posts: 235
Joined: Sat Apr 18, 2009 1:47 pm

Re: Likelihood of NAQT Champion.

Post by bmcke » Sun Jan 19, 2014 11:10 am

I'll make some poorly-informed guesses based on past ICT winners. The frontrunner, Virginia, is maybe 30% to win, and the chance of a team outside of [Will's top-5 + Harvard + Michigan] winning is probably under 5%.

Ways a surprise team could win ICT, in order of plausibility:
1. An elite player joins the team.
2. The team gets very lucky on question selection and buzzer races.
3. Some already-good players on the team improve a lot before the tournament (presumably from a team that doesn't compete much, so the improvement could go unnoticed for a few months).
4. A very smart non-player joins the team and becomes an elite player.
5. Cheating.
Brendan McKendy
Secretary for Ontario Quizbowl Association
University of Ottawa 2011

User avatar
The King's Flight to the Scots
Auron
Posts: 1451
Joined: Mon Jan 26, 2009 11:11 pm

Re: Likelihood of NAQT Champion.

Post by The King's Flight to the Scots » Sun Jan 19, 2014 12:51 pm

bmcke wrote:I'll make some poorly-informed guesses based on past ICT winners. The frontrunner, Virginia, is maybe 30% to win, and the chance of a team outside of [Will's top-5 + Harvard + Michigan] winning is probably under 5%.

Ways a surprise team could win ICT, in order of plausibility:
1. An elite player joins the team.
2. The team gets very lucky on question selection and buzzer races.
3. Some already-good players on the team improve a lot before the tournament (presumably from a team that doesn't compete much, so the improvement could go unnoticed for a few months).
4. A very smart non-player joins the team and becomes an elite player.
5. Cheating.
I might bump #5 up that list...
Matt Bollinger
UVA '14, UVA '15
Communications Officer, ACF

User avatar
Cheynem
Sin
Posts: 6557
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

Re: Likelihood of NAQT Champion.

Post by Cheynem » Sun Jan 19, 2014 2:20 pm

There is a less than 1% chance of any team besides Penn, Yale, or UVA winning ICT. I honestly would be stunned to see any other team win it.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger

User avatar
Auks Ran Ova
Forums Staff: Chief Administrator
Posts: 4042
Joined: Sun Apr 30, 2006 10:28 pm
Location: Minneapolis
Contact:

Re: Likelihood of NAQT Champion.

Post by Auks Ran Ova » Sun Jan 19, 2014 2:55 pm

Cheynem wrote:There is a less than 1% chance of any team besides Penn, Yale, or UVA winning ICT. I honestly would be stunned to see any other team win it.
I'd be stunned to see Chicago or Maryland win it but I wouldn't count them out of the realm of possibility.
Rob Carson
University of Minnesota '11, MCTC '??
Member, ACF
Member, PACE
Writer and Editor, NAQT

User avatar
Cheynem
Sin
Posts: 6557
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

Re: Likelihood of NAQT Champion.

Post by Cheynem » Sun Jan 19, 2014 3:01 pm

UVA: 50%
Yale: 30%
Penn: 19.4%
I'd give Chicago .5% and Maryland .1%. Everyone else literally 0%.

These statistics were calculated using my advanced WIN SHARES prediction formula, so you know they're accurate.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Sun Jan 19, 2014 10:40 pm

So I do see how this isn't like a world cup or any real North American sports, which adopt a playoff system. I think its more like promotion-relegation soccer leagues. Sure there is almost no way for a bottom league team from La Liga to beat Barcelona, like there is almost no chance for my school to beat Yale, but there are always a few surprises over where each team finishes. (Sorry for all the sports analogies in advance). Those are pretty good guesses, but I feel that teams like Maryland and Illinois are much more likely for a great upset then your giving them credit for. Plus, I feel that there is not enough factoring of luck, as there are only 14 games. While I doubt a school from Florida will be the champion, you can never doubt how one lucky game can make a large difference.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
ValenciaQBowl
Auron
Posts: 2338
Joined: Thu Feb 05, 2004 2:25 pm
Location: Orlando, Florida

Re: Likelihood of NAQT Champion.

Post by ValenciaQBowl » Mon Jan 20, 2014 10:09 pm

While I doubt a school from Florida will be the champion
Keep an eye on the CCCT at the end of February. I'll offer a 99.99% chance a team from Florida will be that particular NAQT champion.
Chris Borglum
Valencia College Grand Poobah

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Mon Jan 20, 2014 11:36 pm

I am not really familiar with anything less than Div 2, and even then, its just a passing glance. I really feel like Div 1 is the major trophy to be won. For CCCT is there any team other than Chipola or Valencia that I even know that has a shot at winning it? Anyway, to refer to win shares (I had to search that up). I think it would be more in the quizbowl spirit to see the players' attributes, divide them up into probable teams for NAQT, and use a sort of Pythagorean expectation. This would be some combination of mapping out lines and using the base formula [Points for^x/(Points For^x + Points Against^x)] multiplied by some sort of constant. If someone is interested in some statistical analysis of quizbowl, I could go on, but you get the picture. Anyway, hope someone has more incite into this, whether more analytical or opinion-based.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
prodski
Lulu
Posts: 42
Joined: Sun Jan 22, 2012 10:46 am

Re: Likelihood of NAQT Champion.

Post by prodski » Tue Jan 21, 2014 9:15 pm

ValenciaQBowl wrote:
While I doubt a school from Florida will be the champion
Keep an eye on the CCCT at the end of February. I'll offer a 99.99% chance a team from Florida will be that particular NAQT champion.
Thanks Chris, since you are offering it, I'll take it. I'll gladly put up a buck to your 10,000 that a team from Florida will not win the CCCT. Heck, since I like you, I'll even put up $2 against your $20,000. I love it when illiterate math folks throw around statistics then have to back it up with cash to reflect the true odds of their statement. While the CCCT field may have more than 50% Florida schools, I'll take a shot with the rest of the states, including us, at such as nice price. I'm even locking in my bet now before sectionals this weekend - thanks again - and we'll see you in Atlanta.

I like this original post, and although I disagree with some of the numbers being thrown around later, reasonable odds could be laid on any field just by looking at the D-values. They are not as high as some folks may think though - while quizbowl is different from horse racing and most sports tournaments, a few misplaced negs against a decent team could send a top team down. Yes, they are not eliminated from winning, but a loss decreases their chances significantly. I would compare it to the NCAA basketball tournament, or the women's field, which is stronger at the top. A seed higher than 6 is not going to win come March, similarly a team outside the top 12 has virtually no chance winning a title at quizbowl. Can a 12 seed make the final 4 in hoops? Of course. Can a 15 seed similarly make the top bracket in Chicago? Sure. For those of you sure about a couple of teams, give me the field at 10,000-1 (like Chris) and we'll talk - otherwise I think you need to lower your numbers and include a few other teams in the mix, at least at small percentages.
Peter Rodski
Professor of Mathematics
CCCT Champions 2016
Jefferson Community & Technical College
Louisville, KY. 40272

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Tue Jan 21, 2014 9:54 pm

Thanks for your post. A lot of times people overestimate the chance of an outsider to win compared to the top teams. As someone with a bit of statistical background, I feel that I know this far too well. You are right that this is not like many sports. I do take this as more college sports or top soccer leagues, as top teams usually have many good players, but ever so often teams with a couple good ones and with a bit of practice can up there game in just a couple years. I remember that Toronto didn't qualify until 2007, but slowly made there way up until 2010 at number 5 (originally 6th but retroactively changed). Ever since some graduates left and they lost Jordan Palmer to Ottawa they are not the same, but they still usually a mid table team. I wish more people gave the statistical distribution of the teams, even if it is just a guess. Also, I feel that D-values can be good, but many times I feel that they underestimate some teams that are in an easy field, mostly because they don't take into consideration depth of the field, but they are a good basis. I just wish someone had a sort of percentage chart, so I could compare my results to theirs, as I don't know much about history, or if a player has left or switched schools.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

Adventure Temple Trail
Auron
Posts: 2613
Joined: Tue Jul 15, 2008 9:52 pm

Re: Likelihood of NAQT Champion.

Post by Adventure Temple Trail » Tue Jan 21, 2014 10:27 pm

I basically think that none of these comparisons to professional sports leagues/tournaments have any validity or relevance whatsoever, and that it's pretty much a waste of time to try and compare a round-robin quizbowl championship with 3-5 extremely dominant front runners to wildly different athletic events or tournament schemes.
Matt J.
ex-Georgetown Day HS, ex-Yale
member emeritus, ACF

Sailing away on my copper boat

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Tue Jan 21, 2014 11:10 pm

I agree to you with a certain respect, that quiz bowl works differently than sports teams and other competitions. But I feel that there are many things that they share. They all have statistical percentages, upsets, points gained, and different positions (i.e. categories) that one may feel they are better at. Quiz bowl has front runners, but what is the difference between a team like Virginia or Yale and a team like Chicago or Maryland. Is Penn's team more likely to have a 10% chance of winning it all, or more like 30%? This is why I started the thread, to open up discussion about how teams might match up. To be honest, I am more interested with how outsiders percentages are going to be. I have been in the middle of a distribution chart, and feel it needs to be reworked, and your opinions are really helping me see at least to quizbowlers, how strong each team might be.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
Cheynem
Sin
Posts: 6557
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

Re: Likelihood of NAQT Champion.

Post by Cheynem » Tue Jan 21, 2014 11:15 pm

I think Penn has a chance above 1% of winning and below 100%. Does that help? Seriously, what is the point of this? You can't determine an exact likelihood in professional sports, let alone something more variable like quizbowl/ICT.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger

User avatar
Unicolored Jay
Forums Staff: Administrator
Posts: 781
Joined: Mon Jun 15, 2009 3:28 pm
Location: Columbus, Ohio

Re: Likelihood of NAQT Champion.

Post by Unicolored Jay » Tue Jan 21, 2014 11:16 pm

As a player who has been on a team that's pulled the stunning upset before (we inexplicably beat Chicago A in the prelims at ACF Nationals 2011), it can and does happen, but because of the formats quizbowl nationals take as compared to the single elimination, fewer overall matches, and whatnot that sports has, I don't think the comparison works well. The upset can happen, but often it doesn't have much bearing on the final standings. A team that gets itself into the upper bracket or something from an unbelievable victory usually won't have very many wins (following our victory over Chicago A, we proceeded to lose literally every playoff game afterward to have an overall record of 3-17 or something like that, losing a rematch by a wider margin along the way). Basically, the requirement that the champion has to play at least every team that finishes at least 25% below in the standings really minimizes the luck factor to the level where it's hard to imagine the top teams (who have proven this statistically throughout the year) not winning.

To go back to the original post: I'm not sure if a statistical prediction of the chances of a team winning a quizbowl tournament could be done; there are a lot of factors that play into a match that can't be quantified. Though if someone does attempt at making such a thing, it would be interesting to see how it works out, but I find it hard to imagine it'll be anywhere accurate.
Jasper Lee
University of Tennessee
The Ohio State University '14
Solon High School '10

User avatar
Cheynem
Sin
Posts: 6557
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

Re: Likelihood of NAQT Champion.

Post by Cheynem » Tue Jan 21, 2014 11:27 pm

For the record, lest you accuse me of being a big meanie (this information courtesy of my research sources):

Only three teams have ever won ICT without being in the top playoff bracket at a previous ICT. Stanford in 1998 when who knows what was going on in 1997, Maryland in 2008, and UVA in 2012, both of which had massive lineup changes.

So I would say if you did not finish in the top bracket and did not have major lineup changes, you have 0% chance of winning ICT.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Tue Jan 21, 2014 11:40 pm

Thanks for your replies. I actually think quiz bowl is easier than many other sports as it is very statistics based. A game like baseball seems to be the easiest to set up series of equations, quiz bowl takes many aspects that are easily documented that many other sports did not do (before the introduction of fantasy sports, that is). I also like the fact that the field is very varied. As you have probably seen, I am more into soccer. As anyone can tell you, over 38 games of playing all opponents, things even out, and the best teams usually win. The same can be seen in quiz bowl. While few people can state that Virginia shouldn't have won in 2012, many might wonder how Illinois finished above Yale? The points, the powers, the relative strength of the sides all pointed towards Yale. It seemed that there was an upset that lead to this adventure lead. This is just one of those quiz bowl moments that I really enjoy. It shows that a more well rounded team, while having not as good conversion rate, still managed to win a crucial game. I digress, but it is only to show the need for statistical analysis. I want to be able to quantify how large an upset I had just witnessed.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
Periplus of the Erythraean Sea
Auron
Posts: 1942
Joined: Mon Feb 28, 2011 11:53 pm
Location: New York, NY

Re: Likelihood of NAQT Champion.

Post by Periplus of the Erythraean Sea » Wed Jan 22, 2014 12:09 am

From a purely numbers-based perspective, it's not particularly hard to predict the outcomes of particular matches on average because you rarely see massive outliers in quizbowl performance. This is probably beyond the scope of discussion here, but I'm interested in how particular teams match up against each other statistically based on their strengths in various categories? Theoretically, if you could measure the percentage of tossups a team gets by category and use something like the log5 formula, you could predict the average outcome of a matchup between two teams more accurately than just using aggregate power rates and bonus conversion.

I'm interested in seeing what using a formula like this would reveal about the returns to increasing specialization vs. general knowledge at varying levels of play. My intuition tells me that specialization gets increasingly valuable as the level of competition increases, but I want to see if a theoretical model like this bears it out.
Will Alston
Bethesda Chevy Chase HS '12, Dartmouth '16, Columbia Business School '21
NAQT Writer and Subject Editor

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Wed Jan 22, 2014 12:37 am

I think that is completely cohesive in the discussion. Is the log5 formula like ELO ratings? I wish I could hear from others about the best formula to use when trying to make a distribution chart. Also, I would like to know if there are any major teams that have undergone any major changes. Keep them comments coming.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

Adventure Temple Trail
Auron
Posts: 2613
Joined: Tue Jul 15, 2008 9:52 pm

Re: Likelihood of NAQT Champion.

Post by Adventure Temple Trail » Wed Jan 22, 2014 12:43 am

The amount of useful statistical information about quizbowl is actually deceptively low. We don't have good ways of keeping track of specialization / breakdowns by category, how early teams buzz on average (besides the crude statistic of power rate), where players buzzed on each individual question during a given match, or how much individual players contribute to a team's bonus conversion, for example. We actually have a rather impoverished statistical picture and lots of word of mouth/community dialogue about who we "hear is good at quizbowl", and until those statistics are more developed (which I doubt they ever will be, frankly), comparisons to sabermetrics/sports stats are pretty inane. Your best bet for predicting a champion is to listen to many people who are well-integrated into the community rather than attempting to create a magic number.
Matt J.
ex-Georgetown Day HS, ex-Yale
member emeritus, ACF

Sailing away on my copper boat

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Wed Jan 22, 2014 12:55 am

I agree with what you are saying. That is why I haven't started my distribution chart until I get others opinions. I don't believe that people are flawless either, as many times people overestimate favourite teams. I want to make sure I just have a rough idea at quizbowl and the percentages involved. For example, if my school qualify for ICT, what is the chance that we will be in top half of the table?
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
Periplus of the Erythraean Sea
Auron
Posts: 1942
Joined: Mon Feb 28, 2011 11:53 pm
Location: New York, NY

Re: Likelihood of NAQT Champion.

Post by Periplus of the Erythraean Sea » Wed Jan 22, 2014 12:56 am

http://en.wikipedia.org/wiki/Log5

You'd have a separate statistic for the tossup and bonus conversion rates in each category that would vary by team. For example:

"Fake UVA" - 0.8 TU/26 bonus conversion in all humanities/arts, 0.6/21 in sciences
"Fake Penn" - 0.9/27 conversion in bio/chem, 0.85/27 in world/european history, 0.65/23 in other categories
"Fake Yale" - 0.9/27 conversion in music, 0.85/27 in RMP, 0.7/25 in history, 0.65/23 in other catgories
"Fake Chicago" - 0.9/27 conversion in all arts, 0.7/25 in literature, 0.6/21 in other categories
"Fake Maryland" - 0.8/26 conversion in history, 0.65/22 in other categories
"Fake Harvard" - 0.9/27 conversion in myth & painting, 0.65/23 in European history, music, and sciences, 0.6/21 in other categories

Or something along these lines. Getting the actual numbers for something like this is basically impossible, but I think it could be an interesting exercise in quizbowl theory.
Will Alston
Bethesda Chevy Chase HS '12, Dartmouth '16, Columbia Business School '21
NAQT Writer and Subject Editor

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Wed Jan 22, 2014 1:23 am

Ok, does anyone else have any other advice in calculating this? I think I could track down some sort of stats, while not perfect, can help us learn more about the actual percents of quizbowl rate, and how powerful a side may be. Does anyone have their guesses as to how good all the teams are in a distribution chart? Anyway, I really appreciate the discussion that has come up on this thread so far, and hopefully that it will continue.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
Auks Ran Ova
Forums Staff: Chief Administrator
Posts: 4042
Joined: Sun Apr 30, 2006 10:28 pm
Location: Minneapolis
Contact:

Re: Likelihood of NAQT Champion.

Post by Auks Ran Ova » Wed Jan 22, 2014 1:53 am

Peter13 wrote:I think I could track down some sort of stats, while not perfect, can help us learn more about the actual percents of quizbowl rate, and how powerful a side may be.
Not to be too harsh, and I'm not entirely sure what exactly you mean here ("percents of quizbowl rate"?), but I would be genuinely surprised if you could track down any sort of stats that would significantly improve the quality of analysis you could perform. At the moment, quizbowl simply doesn't collect the sort of advanced/detailed statistics necessary to do something like what Will describes. Performing ever-more complicated math on the relatively meager statistics we already do collect in an effort to wring further meaning out of them is a waste of time at best. Arbitrarily renaming aspects of quizbowl using soccer terms will not make anything clearer either, I'm afraid.
Rob Carson
University of Minnesota '11, MCTC '??
Member, ACF
Member, PACE
Writer and Editor, NAQT

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Wed Jan 22, 2014 2:09 am

Wait what? I am not using soccer terms to describe quizbowl, I am just trying to use statistical analysis that has been done on other sports like soccer to fully understand statistical advantage. If I had Yale and Harvard playing NAQT, questions unknown, what is the chances of getting a victory. Is it 80-20, or a closer 60-40? It is always a fools game going by form, so I like to use the statistical points to find some meaning. I don't think it will be necessarily a waste of time, as I can better address, whatever questions come up, our chances of supposed success or failure.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
Auks Ran Ova
Forums Staff: Chief Administrator
Posts: 4042
Joined: Sun Apr 30, 2006 10:28 pm
Location: Minneapolis
Contact:

Re: Likelihood of NAQT Champion.

Post by Auks Ran Ova » Wed Jan 22, 2014 2:12 am

Peter13 wrote:how powerful a side may be.
Peter13 wrote:top half of the table
Peter13 wrote:I am more into soccer
It was a joke, mostly.
Rob Carson
University of Minnesota '11, MCTC '??
Member, ACF
Member, PACE
Writer and Editor, NAQT

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Wed Jan 22, 2014 2:31 am

Ukonvasara wrote:
Peter13 wrote:how powerful a side may be.
Peter13 wrote:top half of the table
Peter13 wrote:I am more into soccer
It was a joke, mostly.
Very funny sir, very funny.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
AKKOLADE
Sin
Posts: 15256
Joined: Thu Apr 24, 2003 8:08 am

Re: Likelihood of NAQT Champion.

Post by AKKOLADE » Wed Jan 22, 2014 9:07 am

Quiz bowl is obviously just like wrestling, so the ACF Championship will be decided in... HELL IN A CELLLLLLLLLLLL

Image
Fred Morlan
PACE President, 2018-19
International Quiz Bowl Tournaments, co-owner
University of Kentucky CoP, 2017
hsqbrank manager, NAQT writer (former subject editor), former hsqb Administrator/Chief Administrator, 2012 NASAT TD

User avatar
theMoMA
Forums Staff: Administrator
Posts: 5686
Joined: Mon Oct 23, 2006 2:00 am

Re: Likelihood of NAQT Champion.

Post by theMoMA » Wed Jan 22, 2014 1:14 pm

Log5 is probably the best statistic you could use, but as others have noted, there's definitely a garbage in, garbage out problem. (Your inputs will be rough estimates at best.) In any event, I think you'll find it exceedingly unlikely that the five most talented teams will each have 3+ losses, which is probably what would need to happen for a true darkhorse to win.
Andrew Hart
Minnesota alum

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Wed Jan 22, 2014 7:00 pm

There is no true dark horse in sports either. Of course everyone has favourites, but I still feel that people have their preferences and rarely give up and coming teams a chance. Penn has never won, given that they field teams that I believe are deserving of the chance to win. Also, who saw Minnesota win in 2011? I would call that a dark horse, since there major player Brendan Byrne left. I feel people don't take into fact the real odds about a team coming out of the blue and winning. I am not looking for a statistical chart to give me insider access of where people strengths are, I just want it to give me a rough idea of relative strengths compared with other teams in a way that is proportional, not simple ranking.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
Cheynem
Sin
Posts: 6557
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

Re: Likelihood of NAQT Champion.

Post by Cheynem » Wed Jan 22, 2014 7:04 pm

Minnesota didn't come out of the blue and win. They were a very good team prior to Brendan's arrival, had an excellent team besides Brendan in 2009 and 2010 anyway, and while I am nowhere near as good as Brendan, it's not like they added an empty chair after he left either (I had led a B team to a second bracket finish in 2010). Even if you don't quite buy Minnesota as a national title team prior to the season, entering ICT we had a strong D-Value, an excellent SCT performance, and I think were intended to be the top seed in our prelim bracket.

Edit: So we weren't by any stretch of the imagination, an "up and coming team."

Edit II: I looked it up and we were ranked in the preseason #2 (two points off from Illinois)! How in the world could this be considered a dark horse win?!
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Wed Jan 22, 2014 7:26 pm

I take dark horse to mean a team that wins that one would not think they would do such a feat the previous year. You are right, Minnesota is not a good example, but there are some examples of good teams coming up the ranks from somewhere like 28th to 14th, and in some way I think that's an achievement in its own right. Maybe you are right that the favourites flourish, but I still think there are some times when the little guy makes amazing achievements, and no one stops to notice.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
vinteuil
Auron
Posts: 1345
Joined: Sun Oct 23, 2011 12:31 pm

Re: Likelihood of NAQT Champion.

Post by vinteuil » Wed Jan 22, 2014 7:47 pm

I think that's partly because the difference between 24th and 14th isn't all that big compared to, say, 10th to 4th.
Jacob Reed
Chicago ~'25
Yale '17, '19
East Chapel Hill '13
"...distant bayings from...the musicological mafia"―Denis Stevens

User avatar
ThisIsMyUsername
Yuna
Posts: 803
Joined: Wed Jul 15, 2009 11:36 am
Location: New York, NY

Re: Likelihood of NAQT Champion.

Post by ThisIsMyUsername » Wed Jan 22, 2014 11:43 pm

I know nothing about sports, so I can't speak to the many sports analogies being made here. But I can speak to the fact that several posts in this discussion seem very out of touch with how quizbowl works, and not only those by Peter (though claims that Illinois' 2012 ICT success was due to their "well-rounded"ness, or that Minnesota was a dark horse in 2011 [when they were, in fact, one of the statistical favorites to win], or that it is somehow surprising that Penn hasn't won yet are indeed ludicrous).

Some key points:

- There is no one such thing as a "good player". A quizbowl team has many roles, but these roles are not like positions on a sports team, where you occupy only one position and you play that same position every match. How useful you are as a player is a product of what every other person playing the match knows (your teammates and opponents alike). Thus your role (and value) shifts from match to match. And the general trend of the roles you play can shift a lot over your quizbowl career: you might a decent generalist who turns into a specialist when your team acquires a better generalist; or perhaps, losing your generalist causes you, a specialist, to embrace new categories.
- Not only that, but we still have a very poor understanding of how valuable these roles are in various situations. You see this particularly in the difficulty people have of trying to figure out how to rank second- and third-seat scorers on good teams when voting in a player poll. Player X might be entirely incapable of playing solo but still be an invaluable player on any given team that already has a functional generalist. How do we assess the value of those particular points (especially, when their value is a product of what the rest of the field knows)?
- Exactly none of the subtleties of the above are captured by stats. Stats tell you only how many times a player buzzed and how many of those buzzes were correct. If our predictions and polls are sensible (and I am not entirely sure that they are), it is only because we are not relying on only stats as our source of information.
- Everything I said about players applies to teams too. The stats showing how "good" a team is tell you nearly nothing about how they would stack up against another top team, because that is not really about how "good" they are, but rather a complex sum of many strengths and weaknesses. No matter how close Eric Mukherjee and Matt Jackson's PPG might be at a given tournament, they are radically different players in terms of what they know, how they play questions, etc. No matter how similar two teams' PPB might be, the way that these points are spread across categories makes those two teams likely more or less sensitive to different packet conditions.

I have said this before, but the only way that we will one day have enough data to analyze a single tournament sensibly is if we record an entire tournament on video, every match in every room. Only then, for the first time, will we have a complete record of who buzzed where and said what on every single question. Until then, data will not solve any of the issues I've mentioned (and though I've tried reading your posts, Peter, and can't for the life of me figure out what it is you claim to be looking for, I'm fairly certain that data is not going to solve it, either). Even when we have data that sophisticated, understanding what to do with it will take some figuring out.

However, even perfect data analysis of past tournaments will be of limited use for predicting future performances:

- We are not a league of "professional-level" players. Each team consists of four individuals who each may be at any level of skill from complete novice to defending champion. I know of no team sport where rank amateurs and people who have been playing for over a decade compete not only within the same tournament, but on mixed teams; yet, quizbowl is exactly such an activity.
- It is very easy to make the jump from bad player to good player. The bar for quizbowl competence is actually not that high. It is much harder-- but still possible-- to make the jump from good player to great player; this is possible particularly when one decides not to try to cover too many different definitions of "great player", but rather to focus on serving a particular role.
- As a consequence of this, team strength is extraordinarily volatile below the top tier. Most non-top teams could rapidly advance up the rankings if even one player on their team decided to train much harder. This is why no one blinks an eye when a fourth-bracket team becomes a third-bracket team.
- Team strength near the top tends to be less volatile, because many of the players one these teams are operating at or near their "peak" (a point from which it is difficult to continue to improve dramatically).
- However, there is almost always a generally untapped potential for volatility at the top, because on none of the top teams are all four players operating at or near their "peak". This year, most people rank UVA, Yale, and Penn as the top three teams, in that order, and with all other teams being given near-0% odds. This prediction rests on the assumptions that these three teams have more or less peaked (i.e. none of them is going to improve at a dramatically faster rate than the others) and that none of the teams below them can catch up with them (i.e. none of the non-top-three teams is going to improve at a dramatically faster rate than the top three teams are improving). This kind of thinking has been right before, and it has been wrong before (see the predictions for the 2010-2011 season, which are wrong on so many counts: http://www.hsquizbowl.org/forums/viewto ... ll#p197338). But if it ends up being wrong, it will do so in a way that you necessarily couldn't have predicted before, especially from stats, because it will require one or more players (most likely the worst players on these teams, who have to potential to improve rapidly, since they are farther from peaking) breaking from their current rate of improvement.
John Lawrence
Yale University '12
King's College London '13
University of Chicago '19

“I am not absentminded. It is the presence of mind that makes me unaware of everything else.” - G.K. Chesterton

User avatar
bmcke
Wakka
Posts: 235
Joined: Sat Apr 18, 2009 1:47 pm

Re: Likelihood of NAQT Champion.

Post by bmcke » Thu Jan 23, 2014 12:11 am

John's posts here are always really good, but I'll disagree with one point: there is a pretty good way to measure a player's strength just by looking at stats, and it's how many powers they get.

Reasons you would buzz in early and power a question:
1. You're at least 90% certain of the answer (perhaps you've read a book).
2. You think your opponents are strong on this category, so you buzz in with a guess.

When a power shows up in your stats, it means that you know a lot about something, or it means that you got a tossup against strong opponents. Those are both good indicators of strength. If I were ranking the most likely teams to win ICT, I'd probably just go by which teams had the most powers at other tournaments.
Brendan McKendy
Secretary for Ontario Quizbowl Association
University of Ottawa 2011

User avatar
Ndg
Rikku
Posts: 325
Joined: Tue Oct 30, 2012 3:24 pm
Location: Wisconsin

Re: Likelihood of NAQT Champion.

Post by Ndg » Thu Jan 23, 2014 12:37 am

bmcke wrote:John's posts here are always really good, but I'll disagree with one point: there is a pretty good way to measure a player's strength just by looking at stats, and it's how many powers they get.

Reasons you would buzz in early and power a question:
1. You're at least 90% certain of the answer (perhaps you've read a book).
2. You think your opponents are strong on this category, so you buzz in with a guess.

When a power shows up in your stats, it means that you know a lot about something, or it means that you got a tossup against strong opponents. Those are both good indicators of strength. If I were ranking the most likely teams to win ICT, I'd probably just go by which teams had the most powers at other tournaments.
I'm not sure that powers are quite as useful as you're making them out to be. Regarding number 1: speaking for myself, I tend to play extremely conservatively against weak teams --- I buzz when I'm 100% sure, not 90%, since in my view the only way to lose to a substantially inferior team is to be overly aggressive and neg your way to a loss. I would venture that I'm not the only one who plays with that sort of strategy. Playing that way means that the quality of your opponents influences your number of powers as much as your knowledge does.

Regarding number 2: You yourself say that the way to score against a good team is to buzz in early with a guess. If that's true, then the number of powers is determined by how many lucky guesses you have, not your amount of real knowledge --- which would make it a pretty pointless stat.
Andrew Nadig

Mannhiem Mannheim Manheim Township, 2005-11
Carnegie Mellon University, 2011-15

User avatar
ryanrosenberg
Auron
Posts: 1308
Joined: Thu May 05, 2011 5:48 pm
Location: Chicago, Illinois

Re: Likelihood of NAQT Champion.

Post by ryanrosenberg » Thu Jan 23, 2014 1:03 am

ThisIsMyUsername wrote: However, even perfect data analysis of past tournaments will be of limited use for predicting future performances:

- We are not a league of "professional-level" players. Each team consists of four individuals who each may be at any level of skill from complete novice to defending champion. I know of no team sport where rank amateurs and people who have been playing for over a decade compete not only within the same tournament, but on mixed teams; yet, quizbowl is exactly such an activity.
- It is very easy to make the jump from bad player to good player. The bar for quizbowl competence is actually not that high. It is much harder-- but still possible-- to make the jump from good player to great player; this is possible particularly when one decides not to try to cover too many different definitions of "great player", but rather to focus on serving a particular role.
- As a consequence of this, team strength is extraordinarily volatile below the top tier. Most non-top teams could rapidly advance up the rankings if even one player on their team decided to train much harder. This is why no one blinks an eye when a fourth-bracket team becomes a third-bracket team.
- Team strength near the top tends to be less volatile, because many of the players one these teams are operating at or near their "peak" (a point from which it is difficult to continue to improve dramatically).
- However, there is almost always a generally untapped potential for volatility at the top, because on none of the top teams are all four players operating at or near their "peak". This year, most people rank UVA, Yale, and Penn as the top three teams, in that order, and with all other teams being given near-0% odds. This prediction rests on the assumptions that these three teams have more or less peaked (i.e. none of them is going to improve at a dramatically faster rate than the others) and that none of the teams below them can catch up with them (i.e. none of the non-top-three teams is going to improve at a dramatically faster rate than the top three teams are improving). This kind of thinking has been right before, and it has been wrong before (see the predictions for the 2010-2011 season, which are wrong on so many counts: http://www.hsquizbowl.org/forums/viewto ... ll#p197338). But if it ends up being wrong, it will do so in a way that you necessarily couldn't have predicted before, especially from stats, because it will require one or more players (most likely the worst players on these teams, who have to potential to improve rapidly, since they are farther from peaking) breaking from their current rate of improvement.
This is a really good post, and this is more of a personal aside than anything else, but team handball in the US has a very similar structure. Due to the sport's lack of popularity, it's rare you'll have too many people with lots of experience, but you do have some "lifers" or immigrants who have been playing handball for a majority of their lives. Due to that, it's very easy for someone to make the transition from "complete novice" to "competent handball player" if they put in the work and/or are in a good program. However, the teams at the top (NYC, LA, West Point, UNC) are consistently a tier ahead because of their success in having players at their peak handball capacity, whether this is through training or recruitment of European emigres. There is some variation among the top tier, but unless there's a dramatic shift in the handball landscape (as might be the case with Auburn now hosting a USATH Residency Program and getting some of those players for their team), it's very unlikely you'd see a non-West Point or UNC team win College Nationals in the next two or three years. So yeah, I'd say it's very possible that John's first point about the circuit composition leads to the other observations.

/end me geeking out over handball
Ryan Rosenberg
North Carolina '16 | Ardsley '12
PACE | ACF

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Thu Jan 23, 2014 2:01 am

John's post is very good, and looking at the stats and comparing what he said, I take back two of three things I stated earlier and will freely admit they are bad examples. I still can not get over Penn not winning once, since I feel as though they have had a good team for quite a while and should have won something by now. Also, I like the handball analogy. But continuing on, I know the are somethings that are a bit odd. Doing research, VCU made it 3rd or 4th in 2011. There wins over Illinois and Yale seem to shock me a bit. Am I right in this assumption? What are things that statistical analysis (other than what John mentioned earlier) can and cannot tell us about the strength of teams. Also, other than the big three, are there any teams that you would put above a 2% chance of winning? Keep the comments, questions, and discussion going.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
ryanrosenberg
Auron
Posts: 1308
Joined: Thu May 05, 2011 5:48 pm
Location: Chicago, Illinois

Re: Likelihood of NAQT Champion.

Post by ryanrosenberg » Thu Jan 23, 2014 2:21 am

Peter13 wrote:But continuing on, I know the are somethings that are a bit odd. Doing research, VCU made it 3rd or 4th in 2011. There wins over Illinois and Yale seem to shock me a bit. Am I right in this assumption?
VCU returned everyone from the team that had placed 9th at ICT the year before, plus they added Tommy Casalaspi, one of the top high school players in the country, who was taking classes at VCU and was thus eligible to play for them during his senior year of high school. Thus, while VCU is not a perennial top-5 finisher, their performance at ICT wasn't exactly unexpected.
Ryan Rosenberg
North Carolina '16 | Ardsley '12
PACE | ACF

User avatar
Peter13
Wakka
Posts: 211
Joined: Sun Nov 10, 2013 1:28 pm

Re: Likelihood of NAQT Champion.

Post by Peter13 » Thu Jan 23, 2014 3:03 am

Thanks for the reply Ryan. I really think changes in teams say a lot. While Illinois came in 2nd in 2012, they brought in some players to strength up Ike Jose, which was there only major scorer in SCT. There are a couple surprises that sometimes happen, but nothing major. Of course, I did feel VCU still overplayed compared to their player set, but I can see how they would make it in the top 5, but still, I would put less than a 20% chance of them making it 3rd. Looking at it deeper, I do see the flaws, but that makes it all the more interesting to make.
Peter Cordeiro
McMaster University, 2019 +/-1
Interim President of the ONQBA
Newest Accomplishment: SCT @ Waterloo - Scored over 40PPG on a Full Team

User avatar
Banned Tiny Toon Adventures Episode
Tidus
Posts: 721
Joined: Sun May 23, 2010 10:03 am

Re: Likelihood of NAQT Champion.

Post by Banned Tiny Toon Adventures Episode » Thu Jan 23, 2014 8:45 am

Peter13 wrote:Thanks for the reply Ryan. I really think changes in teams say a lot. While Illinois came in 2nd in 2012, they brought in some players to strength up Ike Jose, which was there only major scorer in SCT. There are a couple surprises that sometimes happen, but nothing major. Of course, I did feel VCU still overplayed compared to their player set, but I can see how they would make it in the top 5, but still, I would put less than a 20% chance of them making it 3rd. Looking at it deeper, I do see the flaws, but that makes it all the more interesting to make.
Uhh, these players that played with Ike at the 2012 ICT were ALL at the SCT but they split their team up presumably due to the field's size/strength
Andrew Wang
Illinois 2016

User avatar
marnold
Tidus
Posts: 705
Joined: Wed Jan 17, 2007 12:32 pm
Location: NY

Re: Likelihood of NAQT Champion.

Post by marnold » Thu Jan 23, 2014 9:43 am

bmcke wrote:I'll make some poorly-informed guesses based on past ICT winners. The frontrunner, Virginia, is maybe 30% to win, and the chance of a team outside of [Will's top-5 + Harvard + Michigan] winning is probably under 5%.

Ways a surprise team could win ICT, in order of plausibility:
1. An elite player joins the team.
This thread is sort of a mess, but remember that time I implied in some rankings thread that Andrew was getting a LLM at Columbia starting a spring semester and Jerry sorta believed it? Good times.
Michael Arnold
Chicago 2010
Columbia Law 2013

2009 ACF Nats Champion
2010 ICT Champion
2010 CULT Champion
Member of Mike Cheyne's Quizbowl All-Heel Team

Fundamental Theorem of Quizbowl (Revised): Almost no one is actually good at quizbowl.

User avatar
ThisIsMyUsername
Yuna
Posts: 803
Joined: Wed Jul 15, 2009 11:36 am
Location: New York, NY

Re: Likelihood of NAQT Champion.

Post by ThisIsMyUsername » Thu Jan 23, 2014 12:58 pm

bmcke wrote:John's posts here are always really good, but I'll disagree with one point: there is a pretty good way to measure a player's strength just by looking at stats, and it's how many powers they get.

Reasons you would buzz in early and power a question:
1. You're at least 90% certain of the answer (perhaps you've read a book).
2. You think your opponents are strong on this category, so you buzz in with a guess.

When a power shows up in your stats, it means that you know a lot about something, or it means that you got a tossup against strong opponents. Those are both good indicators of strength. If I were ranking the most likely teams to win ICT, I'd probably just go by which teams had the most powers at other tournaments.
I think powers are an interesting stat, but they are not as useful as you are making out.

Here are some reasons:
- The "meaning" of a power is very set-specific. Let's say I'm a Lit player: Is this a set where clues within power are likely to be titles and character names (e.g. NAQT set)? Or is it a set that focuses on plot incidents, imagery, etc.? I can say that this kind of thing dramatically affects my power numbers at a given tournament. Generalists tend to thrive on sets with shallow powers. Specialists tend to thrive on sets with deep powers.
- Powers suffer from shadow effect just as much as 10's (I am particularly aware of this as someone who played on a team with Kevin Koai for three years, where one of us would regularly beat the other to a music or poetry question within one line of where the other would have powered it, in turn beating the rest of the field.)
- Likewise, one always powers relative to one's field. So, strength of field matters. (e.g. My power numbers shot up in England because the field was weaker.) But once again, categories matter too. I will power more against an overall strong player who is weak in my categories than I will against a weaker player who is strong in my categories.
- As Andrew Nadig points out, powers are mediated by buzzer aggression. We still have no idea what the ideal level of aggression is. And this is another one of those questions that current stats cannot solve because it's too situation-specific. (e.g. If I think it's a 50/50 between "neutrophils" and "mast cells", it's a dumb idea for me to buzz and guess in most situations [especially if I have a Bio player on my team]; however, if I have no Bio player on my team, and I'm playing against Eric Mukherjee who will be able to tell these cells apart at least a couple of clues before I can, buzzing is a strategically smart move.)

I can give you some concrete examples, as I have definitely played tournaments where either I as a player and/or my team as a whole "under-powered" or "over-powered" relative to our actual strength. The two tournaments that spring to my mind from my personal experience are Penn Intergalactic 2011 and Chicago Open 2011.

Here are the Penn Intergalactic stats: http://www.hsquizbowl.org/db/tournament ... nd_finals/. Notice by what a large margin my team, Yale, which won the tournament is "under-powering" compared to the Vinokurov et al. team, which took second (we are also under-powering compared to the Magin-Weiner team). But then notice how many fewer negs we made than any other top team did. At this stage in our quizbowl careers, the Yale team usually played extremely conservatively, especially on high-difficulty questions. I distinctly remember, during the tournament, noticing that Jerry's team was greatly out-powering us, but not worrying about this, because they were greatly out-negging us by a degree that I thought would outweigh their powers. (e.g. They out-powered us 5-3 in the first game, but they out-negged us 4-1, and this allowed us to win.)

The case of Chicago Open 2011 (http://www.hsquizbowl.org/db/tournament ... standings/) is somewhat the reverse. Here, I as an individual out-powered every single member of my team and every member of the field as a whole except Magin. I was a good player already by then, but I certainly was not stronger than my teammates Matt Weiner and Matt Bollinger, let alone the rest of the field except for Magin! Our team as a whole powered more than the winning team. Why might they have been a better team, in spite of their lower powers? For one thing, they negged less. But even compare my stats to Seth's, for example. I almost doubled his power numbers, but no one in their right mind would call me the better player. My buzzes within my categories may have been characteristically earlier than Seth's, but Seth was making more of them against the field. And even if you're buzzing later relative to the questions, if you're nonetheless buzzing more often against the field, you're playing better. (e.g. Theoretically, if I never powered a music questions but still 10'd them before the rest of the field was buzzing, I'm still playing the music questions really well.)

Another way of putting one relevant part of this: if our teams both average 20 PPB, I power two tossups and you 10 three tossups, the stats weight these two figures as equivalent in our personal PPG, and you might be tempted to favor my buzzes, since they were powers. But, in pragmatic reality (assuming our teams play the bonuses at our average PPB's), I have netted 70 points for my team, and you have netted 90.

EDIT: I wrote "less" in once sentence, where I clearly meant "more"
John Lawrence
Yale University '12
King's College London '13
University of Chicago '19

“I am not absentminded. It is the presence of mind that makes me unaware of everything else.” - G.K. Chesterton

evilmonkey
Yuna
Posts: 964
Joined: Mon Sep 25, 2006 11:23 am
Location: Durham, NC

Re: Likelihood of NAQT Champion.

Post by evilmonkey » Thu Jan 23, 2014 6:48 pm

I had a longer post typed up, and then I was logged out of HSQB and lost it. So, for now, I will say that I concur with Matt, Rob, and Andrew; and they have almost everything that I would have said about this topic.

Obviously, if you have proper inputs, then a model can be built - I built such a thing last year. I abandoned it, though, after it was correctly pointed out that the likelihood-based methods of estimation that I was using were biasing some of the results based on which playoff bracket a team was in - pummelling bracket 2 teams would lead to a better rating than getting beat by top bracket teams.
gamegeek2 wrote:This is probably beyond the scope of discussion here, but I'm interested in how particular teams match up against each other statistically based on their strengths in various categories? Theoretically, if you could measure the percentage of tossups a team gets by category and use something like the log5 formula, you could predict the average outcome of a matchup between two teams more accurately than just using aggregate power rates and bonus conversion.

I'm interested in seeing what using a formula like this would reveal about the returns to increasing specialization vs. general knowledge at varying levels of play. My intuition tells me that specialization gets increasingly valuable as the level of competition increases, but I want to see if a theoretical model like this bears it out.
I'm not sure if this is what you are looking for, but it might be interesting.

Let's pretend we have 4 teams: A, B, C, D; and the 20/20 are distributed across 6 categories: Lit, Science, History each with 4/4; RMPSS and FA with 3/3 each; and Geo/CE with 2/2.

Let the teams be ranked in the categories as follows:

Code: Select all

Category	    Team A	Team B	Team C	Team D
History  (4)		1		2		3		4
Lit      (4)		4		2		1		3
Science  (4)		1		3		2		4
RMPSS   (3) 		4		2		3		1
FA      (3)	 	4		2		1		3
Geo/CE   (2)		1		3		4		2
Let's pretend that the better team will ALWAYS get a question over the worse team, for the category. Let's also say that a team will always get 40-10*rank on a bonus in a given category. I randomly ordered the bonuses 1000 times, and had each team play each other team on that ordering of questions. The games fell as follows:

A vs B: 185-700; 10 tossups each
A vs C: 112-805; 10 tossups each
A vs D: 927-13; 10 tossups each
B vs C: 3-997; 9 to 11 in tossups
B vs D: 0-1000; 15 to 5 in tossups
C vs D: 0-1000; 15 to 5 in tossups

(The numbers don't sum to 100 because I did not break ties)
Even though Team B was far better than Team A, and Team A notched occasional wins over Team C, Team C was dominated by Team B.

C: 2802 W, 115 L; 18.01 PPB, 1.93 sd
B: 1703 W, 1182 L; 17.00 PPB, 0.98 sd
A: 1224 W, 1518 L; 14.87 PPB, 3.47 sd
D: 13 W, 2927 L; 10.00 PPB, 3.67 sd

Also, the possible range of scoring margin on the basis of bonus order alone was 130 or 140 points, for each pair of teams.
Bryce Durgin
Culver Academies '07
University of Notre Dame '11
Texas A&M '15

User avatar
Periplus of the Erythraean Sea
Auron
Posts: 1942
Joined: Mon Feb 28, 2011 11:53 pm
Location: New York, NY

Re: Likelihood of NAQT Champion.

Post by Periplus of the Erythraean Sea » Thu Jan 23, 2014 8:27 pm

I'd be more interested in running an experiment like this with non-zero chances for the worse team getting a tossup, since that's a better simulation of reality. This way you use the log5 formula to determine who wins how many questions in a given category.
Will Alston
Bethesda Chevy Chase HS '12, Dartmouth '16, Columbia Business School '21
NAQT Writer and Subject Editor

Locked