Power matching sucks: a theoretical offering

Dormant threads from the high school sections are preserved here.
User avatar
Matt Weiner
Sin
Posts: 8145
Joined: Fri Apr 11, 2003 8:34 pm
Location: Richmond, VA

Power matching sucks: a theoretical offering

Post by Matt Weiner »

Good reasons to use power matching:

1) You need to do something with an incredibly large field (like, say, HSNCT-sized). You are willing to make the other tradeoffs necessary to avoid screwing anyone with power matching, which are: taking lots of teams to the playoffs, and running an elimination bracket playoff because you can't really do playoff brackets when you have an indeterminate and large number of playoff teams. Ideally, this would also include interspersing similar-record games with high-low games to make sure that there's no big advantage to tanking early or penalty for winning (I'm not sure if HSNCT does quite enough of this).

End of list.

Reasons not to use power matching:

1) You do not understand that power-matching is even more sensitive to initial seeding than bracketing and expect to overcome a lack of knowledge or effort regarding seeding in a small number of prelim games.

2) You do not understand that certain practices at national championship tournaments exist because of certain philosophical premises, relevant circumstances to exactly when within the tournament those practices are employed, or contextual necessities of the tournament as a whole, and you mindlessly ape the resulting practices without adopting or understanding the underlying reasons that those practices are in place. This applies equally to "a 192-team HSNCT with a huge, double-elimination playoff does power matching, therefore my tournament should do power matching" and to "NSC uses half-games in certain very specific situations to break ties, therefore I can run an 'ACF-format advantaged final' on a single packet if I break it into half-games." If you don't understand how, when, and why these things are done at nationals, don't just go and adopt the forms of them for your whole event like you're in a cargo cult.

3) Your power matching system does not create a clear break in record between the last team in the playoffs and the first team out, and you don't plan to do tiebreaker games, forcing you to decide who is eliminated based on PPG or any other paper tiebreaker. Yes, I know this one was not an issue at New Trier, but I'm talking about power matching in general, and this does happen at some power-matching tournaments, especially those that try to force power-matching to fit a field that is not 2^x teams.

4) You are not willing to take a large number of teams to the playoffs or use an elimination playoff to make this feasible.

5) Due to either a small number of prelim games or a failure to comprehend how the system works, you use a pure swiss pair that richly rewards tanking and punishes winning, rather than a more interspersed format.

My thesis is that in every non-HSNCT case (and possibly in the HSNCT, though I haven't thought about it at length), bracketed prelims--even those that must be followed by a single-elim playoff due to packet or time constraints--are superior to power-matched prelims in the following areas of fairness:

a) Selecting the actual top N teams in the tournament to a championship playoff of N teams
b) Making sure that it is always in a team's interest to win any given game, and, equivalently, never creating any incentive to tank
c) Giving teams a chance to play their way out of an incorrectly low seed, whether that seed was given due to the host's ignorance, the team's failure to put together its best lineup in the past, or any other reason

Bracketed prelims also offer side benefits of varying importance that are unrelated to tournament fairness, such as: making sure that intra-school matchups in the prelims are minimized, creating geographically diverse brackets, and reducing the probability of someone playing the wrong game due to misunderstanding the format or schedule.

The power-matching fad, like the half-game thing, is something that exists for no real reason beyond emulating what a prestigious tournament does, even though the other events that use it are in entirely different situations that don't match why the original tournament uses the format. I would really like people to reconsider ever using this format.
Matt Weiner
Advisor to Quizbowl at Virginia Commonwealth University / Founder of hsquizbowl.org
Kanga-Rat Murder Society
Wakka
Posts: 209
Joined: Fri Mar 07, 2008 10:52 pm
Location: NW Suburbia, IL

Re: Power matching sucks: a theoretical offering

Post by Kanga-Rat Murder Society »

You are missing the #1 reason people do power matching, which is the fact that it produces closer games that help teams more. Would Auburn A really benefit from playing three of their five matches against B teams or terrible A teams? Would Wheaton Academy B really benefit from playing St. Ignatius? The answer is no. Games against closer teams teach you when to buzz. Games against teams that are hopelessly better or worse than you do not.

I have played multiple tournaments in this format and in pools. I do prefer power matching. In my experience, it produces games that are closer and more exciting to play in. If you disagree, fine.

Also, your complaints in the NTV thread would have been solved if they just let more teams in the playoffs. In all the power matching tournaments, I have never seen a team that deserved to go 4-1 end up with a 2-3 record. Thus, if you let the top two 3-2 teams in the playoffs, you are essentially guaranteed to have the best teams reach the playoffs. Even with all of the upsets yesterday, an eight team playoff would have seen only one of the top eight teams miss the playoffs. This is certainly as effective of playoff bracket as any pool tournament, and would also have the benefits that I discussed above.
Nicholas Bergeon
Buffalo Grove High School '09
UW-Madison '12
WUSTL Law'15
User avatar
AlphaQuizBowler
Tidus
Posts: 695
Joined: Mon Dec 03, 2007 6:31 pm
Location: Alpharetta, GA

Re: Power matching sucks: a theoretical offering

Post by AlphaQuizBowler »

One idea to address the problem of teams not playing competitive games is to start a tournament with two brackets. Teams, when registering for the tournament, can choose which bracket they want to play in, so nobody's kept from playing against the top teams if they want to (alternatively, you could make the top bracket for teams that qualified for nationals, but I think that's too limiting, especially early in the season). By doing this, you end up with two manageable fields instead of a large field with widely varying skill levels. The top teams get to play each other more, and the other teams have a chance to play competitive games with teams close to their skill level. This format was used at the 2008 Rabun-Nacoochee Gap Harvard Mirror, and resulted in most the top teams getting to play each other twice.
William
Alpharetta High School '11
Harvard '15
User avatar
Huang
Rikku
Posts: 442
Joined: Wed Dec 17, 2008 9:29 pm

Re: Power matching sucks: a theoretical offering

Post by Huang »

BG MSL Champs wrote:You are missing the #1 reason people do power matching, which is the fact that it produces closer games that help teams more.
Helps them keep their fragile egos intact? There is minimal, if any, difference between playing close games and playing blowouts in relation to learning. Once someone stops listening, they stop learning. If someone keeps listening, they continue learning. It's rather simple. The problem doesn't stem from bracketed round robins or blowouts. The problem stems from a team being psychologically weak and/or lacking the necessary motivation because they're incapable of placing the results of a game into a larger context. I've learned more from blowouts than I have from close games because blowouts show teams how much they need to improve while close games only show which teams are closer in ability to us.
Games against closer teams teach you when to buzz.
I was under the impression people buzzed when they knew stuff.
Sandy
Kanga-Rat Murder Society
Wakka
Posts: 209
Joined: Fri Mar 07, 2008 10:52 pm
Location: NW Suburbia, IL

Re: Power matching sucks: a theoretical offering

Post by Kanga-Rat Murder Society »

Huang wrote:
BG MSL Champs wrote:You are missing the #1 reason people do power matching, which is the fact that it produces closer games that help teams more.
Helps them keep their fragile egos intact? There is minimal, if any, difference between playing close games and playing blowouts in relation to learning. Once someone stops listening, they stop learning. If someone keeps listening, they continue learning. It's rather simple. The problem doesn't stem from bracketed round robins or blowouts. The problem stems from a team being psychologically weak and/or lacking the necessary motivation because they're incapable of placing the results of a game into a larger context. I've learned more from blowouts than I have from close games because blowouts show teams how much they need to improve while close games only show which teams are closer in ability to us.
It is hard to pay attention in a game where you never have a chance of winning. You may call it psychologically weak. I call it being human. You may say that this is not the case for you, but you are not most people. I am one of those people that, like you, is focused fairly intently throughout the game. However, that is fairly unique. I can think of many players (including one very elite player) who have problems focusing in games that are not close.

Also, your comments seem to indicate that you think that there is no gamesmanship involved in quizbowl. This assumption is absolutely false. There are aspects of quizbowl that are psychological. When I play great teams, I will buzz if I have a hunch, because the question will probably not get to the point where I am 100% sure. Against teams that I know I am better than, I will wait until I am 100% certain.

Huang wrote:
Games against closer teams teach you when to buzz.
I was under the impression people buzzed when they knew stuff.
If this was the case, then why do we bother to travel to national tournaments? We could just play them Goldfish style, and the results would be exactly the same. We would crown the best team in the country and save everybody hundreds of dollars.

The obvious answer to my question is that teams do not always buzz at the exact instant when they know the answer.

And to anticipate your answer to the Goldfish question, we would adjust for negs by checking if teams knew the answer at the end of the question for every question.
Nicholas Bergeon
Buffalo Grove High School '09
UW-Madison '12
WUSTL Law'15
User avatar
Huang
Rikku
Posts: 442
Joined: Wed Dec 17, 2008 9:29 pm

Re: Power matching sucks: a theoretical offering

Post by Huang »

BG MSL Champs wrote: It is hard to pay attention in a game where you never have a chance of winning. You may call it psychologically weak. I call it being human. You may say that this is not the case for you, but you are not most people. I am one of those people that, like you, is focused fairly intently throughout the game. However, that is fairly unique. I can think of many players (including one very elite player) who have problems focusing in games that are not close.
Their lack of focus does not make power matching a better format. Edit: By the way, being psychologically weak does not necessarily mean you're human. Being psychologically weak means you're a weak human. It is certainly possible to not be psychologically weak and still be human.
Also, your comments seem to indicate that you think that there is no gamesmanship involved in quizbowl. This assumption is absolutely false. There are aspects of quizbowl that are psychological. When I play great teams, I will buzz if I have a hunch, because the question will probably not get to the point where I am 100% sure. Against teams that I know I am better than, I will wait until I am 100% certain.
Obviously players weigh probabilities when they're playing. No one is asserting that they don't.
If this was the case, then why do we bother to travel to national tournaments? We could just play them Goldfish style, and the results would be exactly the same. We would crown the best team in the country and save everybody hundreds of dollars.
Right because we can obviously trust everyone to be 100% honest.
The obvious answer to my question is that teams do not always buzz at the exact instant when they know the answer.
Is this really suppose to convince me that power matching is better? You don't need close games to help you know when you should buzz. There's a neat little thing called practice where you can try out all your favorite buzzing techniques.
Last edited by Huang on Sun Dec 20, 2009 3:36 pm, edited 2 times in total.
Sandy
User avatar
at your pleasure
Auron
Posts: 1723
Joined: Sun Aug 03, 2008 7:56 pm

Re: Power matching sucks: a theoretical offering

Post by at your pleasure »

There's nothing awful or even unreasonable about liking close games better; we all find it more exciting to play games that go down to the last question than games that a 610-30 game. But the goal of a quizbowl tournament should not be to give the maximum number of exciting games but to fairly rank all the teams in the field that play the whole tournament. It is not really clear that power-matching does this well, especially if no correction is made to remove the incentive to lose your first game.
Douglas Graebner, Walt Whitman HS 10, Uchicago 14
"... imagination acts upon man as really as does gravitation, and may kill him as certainly as a dose of prussic acid."-Sir James Frazer,The Golden Bough

http://avorticistking.wordpress.com/
User avatar
Captain Sinico
Auron
Posts: 2675
Joined: Sun Sep 21, 2003 1:46 pm
Location: Champaign, Illinois

Re: Power matching sucks: a theoretical offering

Post by Captain Sinico »

I don't understand at all where this thread is coming from. The initial post is rife with questionable or outright wrong premises. For example, the potential half-packet finals at the lit tournament were due to the fact that only six packets existed and, in particular, had nothing at all to do with a desire to emulate the format of anything. Furthermore, the following assertions have not been proven: first, that power matching is more sensitive to seeding than bracketing; second, that single elimination rounds are inherently fairer than paper tiebreaker use for marginal playoff teams; third, all the conclusions, even taking as written the premises. The fact is that I agree with some of the premises and conclusions, but my analysis indicates that the statements made about them are far too strong.
Perhaps unsurprisingly, the thread goes downhill from there.

MaS
Mike Sorice
Former Coach, Centennial High School of Champaign, IL (2014-2020) & Team Illinois (2016-2018)
Alumnus, Illinois ABT (2000-2002; 2003-2009) & Fenwick Scholastic Bowl (1999-2000)
Member, ACF (Emeritus), IHSSBCA, & PACE
User avatar
Matt Weiner
Sin
Posts: 8145
Joined: Fri Apr 11, 2003 8:34 pm
Location: Richmond, VA

Re: Power matching sucks: a theoretical offering

Post by Matt Weiner »

BG MSL Champs wrote:You are missing the #1 reason people do power matching, which is the fact that it produces closer games that help teams more. Would Auburn A really benefit from playing three of their five matches against B teams or terrible A teams? Would Wheaton Academy B really benefit from playing St. Ignatius? The answer is no. Games against closer teams teach you when to buzz. Games against teams that are hopelessly better or worse than you do not.
This is why bracketed playoffs exist. The expense of creating "closer games" for the top 2 teams in the tournament by using power-matching in the morning is that you completely fail at the most important task of the prelims, which is to sort out the worst teams in the championship playoffs from the best teams who are out of them. Power matching inevitably will let in the last few playoff teams based on playing far weaker schedules than the last few non-playoff teams. Brackets, on the other hand, always reflect the best possible strength of schedule balance given available data, and always determine the marginal cases through directly playing off team X against team X-1.
Matt Weiner
Advisor to Quizbowl at Virginia Commonwealth University / Founder of hsquizbowl.org
User avatar
Matt Weiner
Sin
Posts: 8145
Joined: Fri Apr 11, 2003 8:34 pm
Location: Richmond, VA

Re: Power matching sucks: a theoretical offering

Post by Matt Weiner »

Captain Sinico wrote:For example, the potential half-packet finals at the lit tournament were due to the fact that only six packets existed and, in particular, had nothing at all to do with a desire to emulate the format of anything.
I did not have any particular tournament in mind with that example, as this "half-packet advantaged finals" trend has been so prevalent over the last calendar year that it's become an abstract concern. But, think about what you're saying: someone wanted to run a five-round tournament plus an advantaged final, but only wrote six packets. That's not possible to do! Only by subscribing, for whatever reason, to the idea that "half-packet advantaged finals" make sense would one even attempt to use that prelim format to begin with.
Furthermore, the following assertions have not been proven: first, that power matching is more sensitive to seeding than bracketing
The fundamental thing that bracketing does is let you play off your marginal case. Taking the NSC as an example, every 2 seed in the prelim brackets will play the 3 seed, and if all other matches go according to the chalk, the winner of the 2-3 game will always make the playoffs, and the loser always will not. The difference between giving someone the 9th seed in a 64-team NSC and giving them the 24th seed is not effectively any difference at all; both teams play the same common opponents, plus each other, so the true #2 team will always emerge based on game results. A seeded bracket format builds in a huge fudge factor to cover the fact that information is always imperfect and always makes sure that the basic function of prelims, which is determining the playoff teams from the middle group that isn't in either "definite playoff teams" or "definite non-playoff teams," is done as fairly as possible. All you have to do in order to make this work is sort teams into rough tiers, which nearly anyone making a good-faith effort is able to do with sufficient accuracy. To make power-matching work, you have to actually be able to rank the teams almost perfectly top-to-bottom, especially with regard to the difference among teams 5, 6, 7, and 8 in a 6-team playoff system. There really isn't any way to do that in most fields, which, as they say, is why we play the games.

In power matching, the schedule played by someone who starts with card 9 and someone who starts with card 24 can vary wildly in several ways, even if the teams follow the same won-loss path for much of the morning. Which of those cards enables you to rack up four wins and clinch a playoff spot just by losing your first game and never seeing another playoff team? Which of those cards matches you up against a Top 3 team every time you win, and which of them lets you play four patsies before playing team #1 in round five, with a playoff spot already clinched for you? All power-matching is supposed to be seeded (if you don't know this, then your results are even less fair), and the seeds matter so much more for the reason that power-matching is inherently based on not having even strengths-of-schedule for all teams.
second, that single elimination rounds are inherently fairer than paper tiebreaker use for marginal playoff teams
I certainly believe this to be the case for reasons you are probably familiar with. Does it really relate to the power-matching issue?
third, all the conclusions, even taking as written the premises
My argument is that tournament formats need to accomplish certain functions in order to be fair, and that power-matching is less fair than brackets, for the reasons I outlined. I think the chain of logic is pretty clear there.
Matt Weiner
Advisor to Quizbowl at Virginia Commonwealth University / Founder of hsquizbowl.org
User avatar
Captain Sinico
Auron
Posts: 2675
Joined: Sun Sep 21, 2003 1:46 pm
Location: Champaign, Illinois

Re: Power matching sucks: a theoretical offering

Post by Captain Sinico »

I disagree and continue to. I want to stress the difference between "I believe A" and "A is true," which you've blurred here (you say: "I believe single elimination is more fair than paper tiebreakers" in your recent post, but you stated it as accepted fact earlier.) I also want to point out that you're make one (not air-tight) case for brackets over power matching in one tournament format, given one set of resources; you are far from showing that it is the best format in every case.
I want also to stress again that I agree with the spirit of a lot of what you're saying, as you well know, but I feel that you have badly overstated the case in a number of places, which harms the cause of having fair tournaments. If, for example, it happened to be the case that given a certain tournament's constraints, power matching were the fairest format, a case that you yourself acknowledge may occur, it would be the exact kind of cargo cultism you're denouncing here to use bracketed round robins instead because you prefer them.

MaS
Mike Sorice
Former Coach, Centennial High School of Champaign, IL (2014-2020) & Team Illinois (2016-2018)
Alumnus, Illinois ABT (2000-2002; 2003-2009) & Fenwick Scholastic Bowl (1999-2000)
Member, ACF (Emeritus), IHSSBCA, & PACE
User avatar
Matt Weiner
Sin
Posts: 8145
Joined: Fri Apr 11, 2003 8:34 pm
Location: Richmond, VA

Re: Power matching sucks: a theoretical offering

Post by Matt Weiner »

Captain Sinico wrote: If, for example, it happened to be the case that given a certain tournament's constraints, power matching were the fairest format, a case that you yourself acknowledge may occur, it would be the exact kind of cargo cultism you're denouncing here to use bracketed round robins instead because you prefer them.
The only tournament that I can think of where power matching isn't obviously inferior to brackets from the get-go is the HSNCT, and that only because they do several other things (most importantly, take a huge number of teams to the playoffs and spend an appropriate amount of time and effort on initial seeding) that are necessary in order to make power matching something less than a disaster. Even given that, it would probably be possible to run a better HSNCT somehow, and NAQT will be confronted with this issue soon even if they think their current format is great, because an HSNCT that grows much larger will outpace the capacity of any available building to host a one-day prelim tournament, even with teams having 33% or more of the prelim matches as byes as is the case currently. That's not really what I'm getting at in this thread, though: since local tournaments are not in either unique situation of the HSNCT (having to do something with 192 teams, and having to figure out what that thing is with no precedent as to what has or hasn't worked in the past), and because they are not willing to do what HSNCT does with the playoffs to mitigate the inherent unfairness of power matching, they should not use power matching.
Matt Weiner
Advisor to Quizbowl at Virginia Commonwealth University / Founder of hsquizbowl.org
User avatar
Captain Sinico
Auron
Posts: 2675
Joined: Sun Sep 21, 2003 1:46 pm
Location: Champaign, Illinois

Re: Power matching sucks: a theoretical offering

Post by Captain Sinico »

The point still stands that you have not excluded the possibility that another format may be as or more fair in some (arbitrarily large number of) cases.

Mas
Mike Sorice
Former Coach, Centennial High School of Champaign, IL (2014-2020) & Team Illinois (2016-2018)
Alumnus, Illinois ABT (2000-2002; 2003-2009) & Fenwick Scholastic Bowl (1999-2000)
Member, ACF (Emeritus), IHSSBCA, & PACE
User avatar
Stained Diviner
Auron
Posts: 5085
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Power matching sucks: a theoretical offering

Post by Stained Diviner »

Actually, a card system does a much better job of accounting for seeding discrepancies than a pool system. A pool system accounts for a limited number of discrepancies well, such as having 16 playoff teams and confusing your 16 seed with your 17 seed (a simple situation that any system would account for well). It accounts for a large number of discrepancies poorly, such as having both your 16 and 17 seeds being seeded incorrectly in the same direction but having exactly one of them make the playoffs anyways, or having the captain of your 9 seed team not show up, so your 24 seed has an easier path to the playoffs than your 10-23 seeds. I have explained this in other threads and can do so again if you want me to analyze a particular case. The two main reasons I use cards when I can are that they account for poor seeding better and that they provide more interesting matches. It is much easier to play your way out of a poor seed with power matching than with pools.

One of your points is that hosts are unwilling to use elimination playoffs. You make it sound like they are just being quirky in making such a decision, but there are good reasons why people don't like single-elimination playoffs that I can get into if you want me to.

Also, my system does not reward tanking. There are cases of poorly designed card systems that do reward tanking, but I don't think that matters, since we're not discussing poorly designed card systems here.

Additionally, the idea that this is done to emulate a national championship is ridiculous. For one thing, I started using the system in 2001 and found out that NAQT was doing it in 2005. For another thing, there is more than one national championship, and only one that uses a card system. I am not sure why you think people are emulating HSNCT to be national championshipish but not emulating NSC if you think that this is the reason it's being done.

I agree that power-pairing does lead to PPG being an especially poor paper tiebreaker. The remedy is to devise a system that does not need paper tiebreakers or to use PPB. Problem solved.

As to Matt's last post in the New Trier thread, eliminating Auburn for their one loss would be worse than eliminating Stevenson or Loyola for their one loss because in the cases of Stevenson and Loyola it was their second loss. Their second losses were to teams that lost to a team different than the team that beat them, as was guaranteed by the system. Eliminating a team on its first loss is not equivalent to eliminating a team on its second loss. Also, while teams may play a disparate schedule, they get eliminated by roughly the same level of opponent. In other words, a team that wins its first three matches and loses its last two plays a stronger schedule overall than a team that wins its middle three games and loses its first and last. However, the team that wins its first three plays with a cushion for its first four matches, as opposed to the team that loses its first and plays for survival each round. Furthermore, the team that wins its first three generally gets an easier opponent in its fifth and deciding match than the team that wins it second through fourth matches, so when it matters most they are rewarded for playing the tougher schedule.
David Reinstein
Head Writer and Editor for Scobol Solo, Masonics, and IESA; TD for Scobol Solo and Reinstein Varsity; IHSSBCA Board Member; IHSSBCA Chair (2004-2014); PACE President (2016-2018)
User avatar
AlphaQuizBowler
Tidus
Posts: 695
Joined: Mon Dec 03, 2007 6:31 pm
Location: Alpharetta, GA

Re: Power matching sucks: a theoretical offering

Post by AlphaQuizBowler »

Mr. Reinstein,
Could you post the 32-team card system you used? I'm trying to work out situations to compare card systems and bracketing, but I don't think I'm setting up the card system right.
William
Alpharetta High School '11
Harvard '15
User avatar
Stained Diviner
Auron
Posts: 5085
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Power matching sucks: a theoretical offering

Post by Stained Diviner »

Let me summarize the particular case of the New Trier Tournament:
1. The two weakest teams in the top bracket, going by both paper tiebreakers and the results in the afternoon, were Northmont and Wheaton Warrenville South.
2. The two strongest teams not in the top bracket, going by both paper tiebreakers and the results in the afternoon, were Loyola and Stevenson.
3. The reasons that Northmont and Wheaton Warrenville South were in the top bracket and Loyola and Stevenson were not are that in Round Five, Wheaton Warrenville South beat Loyola and Northmont beat Stevenson.

We used the same system that Dwight Wynne explained on Google Docs here. Let me know if you have access problems.
David Reinstein
Head Writer and Editor for Scobol Solo, Masonics, and IESA; TD for Scobol Solo and Reinstein Varsity; IHSSBCA Board Member; IHSSBCA Chair (2004-2014); PACE President (2016-2018)
User avatar
Matt Weiner
Sin
Posts: 8145
Joined: Fri Apr 11, 2003 8:34 pm
Location: Richmond, VA

Re: Power matching sucks: a theoretical offering

Post by Matt Weiner »

Shcool wrote:Let me summarize the particular case of the New Trier Tournament:
1. The two weakest teams in the top bracket, going by both paper tiebreakers and the results in the afternoon, were Northmont and Wheaton Warrenville South.
2. The two strongest teams not in the top bracket, going by both paper tiebreakers and the results in the afternoon, were Loyola and Stevenson.
3. The reasons that Northmont and Wheaton Warrenville South were in the top bracket and Loyola and Stevenson were not are that in Round Five, Wheaton Warrenville South beat Loyola and Northmont beat Stevenson.
Of course, if you set up a system where 3-1 is playing 3-1 in round five and only 4-1 teams make the playoffs, then the above scenario will always happen, regardless of whether you actually put the correct teams in those games, and thus is a tautology if you are arguing that you actually made the playoff split correctly. Some teams get to go to the 3-1 single-elimination game by going 3-1 in the first four rounds against no one in the top 10, and some teams get put there because they only went 1-1 in the two prior games they had to play against the top 5 of the tournament. This is why your power matching system is not fair, independent of the fact that you also used zero-elimination, which is presumably worse than single-elimination, to knock out 1/3 of the field on Friday afternoon.
Matt Weiner
Advisor to Quizbowl at Virginia Commonwealth University / Founder of hsquizbowl.org
User avatar
Important Bird Area
Forums Staff: Administrator
Posts: 6113
Joined: Thu Aug 28, 2003 3:33 pm
Location: San Francisco Bay Area
Contact:

Re: Power matching sucks: a theoretical offering

Post by Important Bird Area »

Matt Weiner wrote:Even given that, it would probably be possible to run a better HSNCT somehow, and NAQT will be confronted with this issue soon even if they think their current format is great, because an HSNCT that grows much larger will outpace the capacity of any available building to host a one-day prelim tournament, even with teams having 33% or more of the prelim matches as byes as is the case currently.
We're listening, and welcome suggestions for improving the HSNCT format. We do in fact like the current format, but I agree that some kind of change will be needed should we decide to expand the field size beyond the current 192 teams.
Jeff Hoppes
President, Northern California Quiz Bowl Alliance
former HSQB Chief Admin (2012-13)
VP for Communication and history subject editor, NAQT
Editor emeritus, ACF

"I wish to make some kind of joke about Jeff's love of birds, but I always fear he'll turn them on me Hitchcock-style." -Fred
User avatar
Stained Diviner
Auron
Posts: 5085
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Power matching sucks: a theoretical offering

Post by Stained Diviner »

Matt Weiner wrote:
Shcool wrote:Let me summarize the particular case of the New Trier Tournament:
1. The two weakest teams in the top bracket, going by both paper tiebreakers and the results in the afternoon, were Northmont and Wheaton Warrenville South.
2. The two strongest teams not in the top bracket, going by both paper tiebreakers and the results in the afternoon, were Loyola and Stevenson.
3. The reasons that Northmont and Wheaton Warrenville South were in the top bracket and Loyola and Stevenson were not are that in Round Five, Wheaton Warrenville South beat Loyola and Northmont beat Stevenson.
Of course, if you set up a system where 3-1 is playing 3-1 in round five and only 4-1 teams make the playoffs, then the above scenario will always happen, regardless of whether you actually put the correct teams in those games, and thus is a tautology if you are arguing that you actually made the playoff split correctly. Some teams get to go to the 3-1 single-elimination game by going 3-1 in the first four rounds against no one in the top 10, and some teams get put there because they only went 1-1 in the two prior games they had to play against the top 5 of the tournament. This is why your power matching system is not fair, independent of the fact that you also used zero-elimination, which is presumably worse than single-elimination, to knock out 1/3 of the field on Friday afternoon.
Wrong again. If the two best 3-1 teams played each other, then one of them would win and go on to do well in the top bracket, and the other one would lose and go on to dominate the next bracket.

Also, do you realize that you are suggesting that instead of using a double-elimination power matching in the morning that we should have used a single-elimination power matching in the afternoon? In other words, you are suggesting something obviously worse than what we did. Your other suggestion of 14 rounds was not a practical option--we could have written 14 rounds if we had wanted to, but we had no reason to want to because we didn't have enough teams that wanted to play them.
David Reinstein
Head Writer and Editor for Scobol Solo, Masonics, and IESA; TD for Scobol Solo and Reinstein Varsity; IHSSBCA Board Member; IHSSBCA Chair (2004-2014); PACE President (2016-2018)
User avatar
Mechanical Beasts
Banned Cheater
Posts: 5673
Joined: Thu Jun 08, 2006 10:50 pm

Re: Power matching sucks: a theoretical offering

Post by Mechanical Beasts »

Shcool wrote:Wrong again. If the two best 3-1 teams played each other, then one of them would win and go on to do well in the top bracket, and the other one would lose and go on to dominate the next bracket.
Which would... also suck? I don't get it.
Also, do you realize that you are suggesting that instead of using a double-elimination power matching in the morning that we should have used a single-elimination power matching in the afternoon? In other words, you are suggesting something obviously worse than what we did.
Presumably he does. Moreover, yeah, this isn't obviously worse, as record gained on uneven strength of schedule for the prelims mattered for the playoffs at this tournament.
Your other suggestion of 14 rounds was not a practical option--we could have written 14 rounds if we had wanted to, but we had no reason to want to because we didn't have enough teams that wanted to play them.
"It's not practical to say that we should have written fourteen rounds, because we could have written them."
"It's not practical to say that we should have written fourteen rounds, because we didn't have enough teams that wanted to play them--except the teams who attended the tournament, who (Matt argues) would have experienced a fairer tournament had those rounds been written, so they sure would have been desired."
Andrew Watkins
User avatar
Sen. Estes Kefauver (D-TN)
Chairman of Anti-Music Mafia Committee
Posts: 5647
Joined: Wed Jul 26, 2006 11:46 pm

Re: Power matching sucks: a theoretical offering

Post by Sen. Estes Kefauver (D-TN) »

"It's not practical to say that we should have written fourteen rounds, because we didn't have enough teams that wanted to play them--except the teams who attended the tournament, who (Matt argues) would have experienced a fairer tournament had those rounds been written, so they sure would have been desired."
This doesn't seem to be at all what was said. Having all the teams play 10 rounds was clearly pushing it for this audience, so having them all play another 2 rounds obviously would have led to many more teams leaving early than already did. Thus, probably the vast majority of their audience would have been uninterested in more games. Now, I'm always of the opinion that hosts should offer teams at least 9 games unless they have a format that simply makes that impossible, but if you are in a region where teams are used to 4 morning games or whatever, expecting teams to play more than 10 matches is probably not going to work out - just over a week ago Rolla ran a tournament where the 12 teams that weren't in the top bracket refused to play their offered consolation matches. I disagree with many other arguments Reinstein has made, but your post here seems to be an out and out lie when it purports to present his position.
Charlie Dees, North Kansas City HS '08
"I won't say more because I know some of you parse everything I say." - Jeremy Gibbs

"At one TJ tournament the neg prize was the Hampshire College ultimate frisbee team (nude) calender featuring one Evan Silberman. In retrospect that could have been a disaster." - Harry White
User avatar
Captain Sinico
Auron
Posts: 2675
Joined: Sun Sep 21, 2003 1:46 pm
Location: Champaign, Illinois

Re: Power matching sucks: a theoretical offering

Post by Captain Sinico »

It should be noted, however, that it's possible (and probably easy) to design a format that gives, say, 10 games to non-championship teams and more to championship-bracket teams.

MaS
Mike Sorice
Former Coach, Centennial High School of Champaign, IL (2014-2020) & Team Illinois (2016-2018)
Alumnus, Illinois ABT (2000-2002; 2003-2009) & Fenwick Scholastic Bowl (1999-2000)
Member, ACF (Emeritus), IHSSBCA, & PACE
User avatar
Mechanical Beasts
Banned Cheater
Posts: 5673
Joined: Thu Jun 08, 2006 10:50 pm

Re: Power matching sucks: a theoretical offering

Post by Mechanical Beasts »

You're right, Charlie; I very much misread what Reinstein intended there. To that, though, I say, okay: if a fourteen round format is necessary to
1) get the right teams in the top bracket
2) play off the top bracket fairly
then let's do that format. If teams want to leave before it's over (and I'll posit that teams that would leave early aren't generally in championship contention), then that's their business (and you'll probably know it beforehand). (And you can certainly, if you know beforehand that teams will leave, design a format that preserves the 14-round format (or whatever) for the top bracket and not for the brackets where teams are likely to leave--as Sorice is saying.)
Andrew Watkins
User avatar
cvdwightw
Auron
Posts: 3291
Joined: Tue May 13, 2003 12:46 am
Location: Southern CA
Contact:

Re: Power matching sucks: a theoretical offering

Post by cvdwightw »

Matt, it's clear to me that you have some sort of vendetta against power-matching. I will note that (in concurrence with Sorice) I agree with several of your assumptions about why power-matching is bad; however, there is absolutely no attempt to prove any of your premises, or that they logically follow from your conclusions.
Matt Weiner wrote:1) You do not understand that power-matching is even more sensitive to initial seeding than bracketing and expect to overcome a lack of knowledge or effort regarding seeding in a small number of prelim games.
For a 32 team bracket, the worst case scenario (actual seeds 1-8 in the same 8-team minibracket) gives 4 out of 6 correct teams (seeds 1, 2, 3, and 5) advancing out of the power-matching bracket. I haven't done the math on exactly how many brackets this 66.67% "real" advancement is better than, but it's better than the "worst case" bracket scenario, in which only two teams make it to eight-team playoffs. In the case that you know very little about the abilities of a significant part of your field, I find it very hard to assume that seeded brackets are better than power-matching. I'm almost 100% confident that if you were to take any random configuration of numbers and make that random configuration your 1-32 seeds, you'd end up with a higher percentage of correct playoff teams in the power-matching than in the bracketing scenario more than 50% of the time.
Matt Weiner wrote:Power matching inevitably will let in the last few playoff teams based on playing far weaker schedules than the last few non-playoff teams. Brackets, on the other hand, always reflect the best possible strength of schedule balance given available data, and always determine the marginal cases through directly playing off team X against team X-1.
Assume that for whatever reason (e.g. number of packets) only the top 6 teams can make the playoffs. If Northmont and Stevenson both end up #2 in their brackets (a reasonable assumption if Stevenson's seed is off), and you play off the game between them to decide who ends up in slot #5, how is that any more fair than what happened here? There isn't a strength of schedule balance - you're comparing across brackets, and you have no way to correct for under- or over-seedings.
Matt Weiner wrote:The fundamental thing that bracketing does is let you play off your marginal case. Taking the NSC as an example, every 2 seed in the prelim brackets will play the 3 seed, and if all other matches go according to the chalk, the winner of the 2-3 game will always make the playoffs, and the loser always will not. The difference between giving someone the 9th seed in a 64-team NSC and giving them the 24th seed is not effectively any difference at all; both teams play the same common opponents, plus each other, so the true #2 team will always emerge based on game results.
I understand what you are saying here, but you have yet to demonstrate that power matching does not play off the marginal case. Furthermore, in your rant in the New Trier thread, you are taking as a fact that Stevenson was not a "marginal case." I count 7 teams that had a bonus conversion of between 16 and 19. I don't know, maybe that makes 7 marginal cases.
Matt Weiner wrote:In power matching, the schedule played by someone who starts with card 9 and someone who starts with card 24 can vary wildly in several ways, even if the teams follow the same won-loss path for much of the morning. Which of those cards enables you to rack up four wins and clinch a playoff spot just by losing your first game and never seeing another playoff team? Which of those cards matches you up against a Top 3 team every time you win, and which of them lets you play four patsies before playing team #1 in round five, with a playoff spot already clinched for you?
This is simply false (notwithstanding the fact that Card #9 and Card #24 play each other in round one of a 32-team power-matching tournament such as is being discussed). First off let's take the assumption that the teams are correctly seeded and that the higher seed always wins, except that Team #9 may or may not decide to tank against team #24.

Team #9 (with the #9 card): Round 2 loss to Team #8 (card 9), Round 3 win over Team #17 (9), Round 4 win over Team #12 (9), Round 3 loss against Team #3 (9): total record 3-2
Team #9 (with the #24 card): Round 2 win over Team #25 (card 24), Round 3 win over Team #16 (16), Round 4 loss to Team #5 (16), Round 5 win over Team #18 (16): total record 3-2
Team #24 (with the #9 card): Round 2 loss to Team #8 (card 9), Round 3 loss to Team #17 (17), Round 4 win over Team #28 (17), Round 5 loss to Team #15 (17): total record 2-3
Team #24 (with the #24 card): Round 2 win over Team #25 (card 24), Round 3 loss to Team #16 (24), Round 4 loss to Team #21 (24), Round 5 win over Team #30 (24): total record 2-3

Yes, tanking the game gets you lower-seeded opponents, but your overall record does not change. The only incentive to tank a game is if there is a paper tie-breaker based on points per game; then you do have an incentive to lose and put up lots of points against a (hopefully) weaker team in the next round. If you don't use paper tiebreakers, there's no incentive to lose.
Matt Weiner wrote:To make power-matching work, you have to actually be able to rank the teams almost perfectly top-to-bottom, especially with regard to the difference among teams 5, 6, 7, and 8 in a 6-team playoff system. There really isn't any way to do that in most fields, which, as they say, is why we play the games.
This simply isn't true either. If seeds hold and you screw up teams 5-8:

Seed 5 plays seeds 28, 12, 4, 16, and 7
Seed 6 plays seeds 27, 11, 3, 15, and 8
Seed 7 plays seeds 26, 10, 2, 14, and 5
Seed 8 plays seeds 25, 9, 1, 13, and 6

The argument that you make that power-matching amplifies bad seeding only holds in the number of teams that are affected. OPRF was the #8 seed and played #25, #9, #17, #20, and #7 seeds. #5 seed Stevenson played #28, #21, #4, #1, and #11. Amazing News Flash: Stevenson would have been given a #2 bracket seed and Northmont a #3 seed. All of a sudden it's the same 2-3 matchup deciding who makes it in. The fact is that chalk does not always hold - DCC upset Carbondale to grab the 2 card, and then lost to St. Ignatius to get the 3 card. Upsets happen. Does power-matching have the potential to mess around with final standings more than bracketing does? Yes, because of upsets. Does power-matching have the potential to correct for incorrect seeding more than bracketing does? Yes, because of upsets.

Here's the other thing that you are absolutely discounting: to duck a higher-seeded team, you have to actually lose! That is, if you duck seed #1 by dropping the 8-9 game, then you play card #17 and your remaining matches are against card #12 and card #3 and you have to win both. If you win the 8-9 game and lose to #1, your remaining matches are against card #13 and card #6, and you have to win both. I don't see an advantage to tanking.
Dwight Wynne
socalquizbowl.org
UC Irvine 2008-2013; UCLA 2004-2007; Capistrano Valley High School 2000-2003

"It's a competition, but it's not a sport. On a scale, if football is a 10, then rowing would be a two. One would be Quiz Bowl." --Matt Birk on rowing, SI On Campus, 10/21/03

"If you were my teammate, I would have tossed your ass out the door so fast you'd be emitting Cerenkov radiation, but I'm not classy like Dwight." --Jerry
User avatar
cvdwightw
Auron
Posts: 3291
Joined: Tue May 13, 2003 12:46 am
Location: Southern CA
Contact:

Re: Power matching sucks: a theoretical offering

Post by cvdwightw »

Shcool wrote:We used the same system that Dwight Wynne explained on Google Docs here. Let me know if you have access problems.
I have a question. In Round 4, according to this system, there should be a matchup between cards 6 and 15, and between cards 10 and 11. I am assuming that teams are listed by seed order in the way that you would list them in an actual bracket (1, 32, 16, 17, 8, 25, 9, 24, etc.). According to this assumption, 6 seed Lisle beat 11 seed Northmont in Round 2. In Round 3, 6 seed Lisle lost to 3 seed St. Ignatius, keeping card 6, while Northmont beat 19 seed Fenton, keeping card 11. In Round 4, Lisle played Wheaton-Warrenville South, who according to my reverse-engineering should have had card 10; they lost, apparently receiving card 14 while WWS got card 7. Meanwhile, Northmont played Latin, who should have had card 15, and won, getting card 10, while Latin took card 11. Fenwick ended up with card 15, playing Card 17 Hoffman Estates.

What should have happened: winner of Lisle-Latin should have played Fremd; winner of WWS-Northmont should have played Stevenson; loser of Lisle-Latin should have played Hoffman Estates; loser of WWS-Northmont should have played Niles West; winner of Carbondale-Fenwick should have played Loyola; loser of Carbondale-Fenwick should have played Naperville North. Essentially, the 6-15 matchup was essentially the 6-10 matchup and the directions for after Round 4 pertained instead to the 7-14 matchup; the 10-11 matchup was essentially the 11-15 matchup; the 7-14 matchup had postmatch directions for the 6-15 matchup.

Am I doing something wrong in my analysis of this tournament, or was there a major screwup somewhere in Round 4 that meant that teams were playing the wrong teams in Round 4 and Round 5? Ultimately if I'm right, that means that at least one of Northmont and WWS should not have made the playoffs, and at least one of Lisle/Latin/Fremd should have made the playoffs.
Dwight Wynne
socalquizbowl.org
UC Irvine 2008-2013; UCLA 2004-2007; Capistrano Valley High School 2000-2003

"It's a competition, but it's not a sport. On a scale, if football is a 10, then rowing would be a two. One would be Quiz Bowl." --Matt Birk on rowing, SI On Campus, 10/21/03

"If you were my teammate, I would have tossed your ass out the door so fast you'd be emitting Cerenkov radiation, but I'm not classy like Dwight." --Jerry
User avatar
Matt Weiner
Sin
Posts: 8145
Joined: Fri Apr 11, 2003 8:34 pm
Location: Richmond, VA

Re: Power matching sucks: a theoretical offering

Post by Matt Weiner »

cvdwightw wrote:For a 32 team bracket, the worst case scenario (actual seeds 1-8 in the same 8-team minibracket) gives 4 out of 6 correct teams (seeds 1, 2, 3, and 5) advancing out of the power-matching bracket. I haven't done the math on exactly how many brackets this 66.67% "real" advancement is better than, but it's better than the "worst case" bracket scenario, in which only two teams make it to eight-team playoffs.
We're comparing realistic possibilities for actual tournaments here, not "which format can be made worse if you go out of your way to make the seedings as wrong as possible." That format probably is bracketing; I haven't thought about it because it doesn't matter.
Assume that for whatever reason (e.g. number of packets) only the top 6 teams can make the playoffs. If Northmont and Stevenson both end up #2 in their brackets (a reasonable assumption if Stevenson's seed is off), and you play off the game between them to decide who ends up in slot #5, how is that any more fair than what happened here? There isn't a strength of schedule balance - you're comparing across brackets, and you have no way to correct for under- or over-seedings.
Because Northmont never had to play any other playoff teams, but Stevenson had to play the #1 team in the field before meeting Northmont. If Northmont and Stevenson were 2 and 3 in a bracket, they both would have had to play a #1 rather than playing each other + a random, wildly different set of other opponents.
I understand what you are saying here, but you have yet to demonstrate that power matching does not play off the marginal case.
It does by its own terms, as I showed above, but we have no real way of knowing whether those teams are the actual marginal teams. If we approximate by BC stats, we see that basically teams 3 through 9 at New Trier all were pretty close, but instead of the games among those teams directly determining who made the playoffs, some of them had to play additional games against teams 1 or 2, while some did not, effectively making the playoff entry single-elimination for some teams but double-elimination for others (while remaining zero-elimination for 1/3 of the field). This is a pretty basic manifestation of unfairness.
let's take the assumption that the teams are correctly seeded and that the higher seed always wins
Um....
Yes, tanking the game gets you lower-seeded opponents, but your overall record does not change.
What changes your overall record is getting to play all of your games against teams ranked 10th or lower, or having to play two of them against playoff teams, but then comparing teams based solely on their won-loss record against those totally incommensurable schedules.
The only incentive to tank a game is if there is a paper tie-breaker based on points per game
...or if you do get the card that lets you lose one game and then play all of your remaining games against teams not in contention for the championship, which we've seen exists in at least one case at New Trier, and likely exists in multiple cases in all power-matching formats that stop at round five.

General note: I think it's pretty dumb to accuse people of having "vendettas" or "ranting" in order to characterize their opposition to something in a negative light. I'm pretty patiently explaining why I believe what I do about power matching; "you're not allowed to dislike a thing, because you obviously dislike it" is pretty fallacious reasoning.
Matt Weiner
Advisor to Quizbowl at Virginia Commonwealth University / Founder of hsquizbowl.org
User avatar
Stained Diviner
Auron
Posts: 5085
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Power matching sucks: a theoretical offering

Post by Stained Diviner »

Dwight, you got our seeding order wrong by a little bit. Saint Ignatius was actually the 2 seed. The seeds were 1, 32, 16, 17, 8, 25, 9, 24, 4, 29, 13, 20, 5, 28, 12, 21, 2, 31, 15, 18, 7, 26, 10, 23, 3, 30, 14, 19, 6, 27, 11, 22. Also, given the fact that the teams tied for fifth place did not want to play it off and two of the coaches who finished higher than them complained about the length of the tournament (which finished Round 10 around 4:15), I think your assumption that the top teams would have been fine with more rounds is incorrect.

Matt, Dwight just gave you a situation where power matching produces a better result than pools. It doesn't just happen like that in extreme situations--it happens like that in every situation I can imagine. If there is a single point of yours that has held up at this point, please point it out to me.

You pointed out in another thread that Stevenson got screwed because they played teams #1 and #5, but the only reason Northmont is considered #5 is that they beat Stevenson. Based on the most reasonable statistic, PPB, they were #9. Based on the seed they held going into that match, they were #10. Stevenson did not get screwed at all--they lost a fairly seeded match, which led to them not making the top six.

Also, if we had used Matt's system, Auburn's first seven matches would have been against Carmel, Fremd, Wheaton Academy, Niles West, and three B Teams. These are perfectly respectable programs with some decent players, but Matt is advocating a system in which Auburn would be unlikely to play a close match before 2:00 and then, if we were being practical, would be knocked out by a single loss, or, if we were being impractical, would have to play against teams that had taken their buzzers and gone home. By my count, there would have been four matches in the first seven rounds between elite teams, where elite is defined as teams that I thought had a decent shot to make the top six at our tournament. And the justification for his system is that poor Stevenson had to play Northmont to make the top six.

I challenge Matt once again to point out one of his arguments that has held up. I don't see one.
David Reinstein
Head Writer and Editor for Scobol Solo, Masonics, and IESA; TD for Scobol Solo and Reinstein Varsity; IHSSBCA Board Member; IHSSBCA Chair (2004-2014); PACE President (2016-2018)
User avatar
Sen. Estes Kefauver (D-TN)
Chairman of Anti-Music Mafia Committee
Posts: 5647
Joined: Wed Jul 26, 2006 11:46 pm

Re: Power matching sucks: a theoretical offering

Post by Sen. Estes Kefauver (D-TN) »

What is the point of this infatuation with guaranteeing more top teams playing games against each other during the prelims? I simply don't think that is a criteria that should posses any relevance at all - not all tournaments can offer anything like that to begin with - but I also think that the point of playoffs, not prelims, should be to offer teams competitive games there. If half of your games are against similar enough teams, within the constraints of a given field, that's a high enough percentage for teams to be satisfied that they are playing competitive games. To me, the point of a pool prelim is so that you can play games against a sample of teams from across the gamut of competitiveness and see which general tier you are sorted out into so you can then do playoffs and get a much more precise ranking of where you fall within your tier.
Charlie Dees, North Kansas City HS '08
"I won't say more because I know some of you parse everything I say." - Jeremy Gibbs

"At one TJ tournament the neg prize was the Hampshire College ultimate frisbee team (nude) calender featuring one Evan Silberman. In retrospect that could have been a disaster." - Harry White
User avatar
Down and out in Quintana Roo
Auron
Posts: 2907
Joined: Wed Apr 09, 2008 7:25 am
Location: Camden, DE
Contact:

Re: Power matching sucks: a theoretical offering

Post by Down and out in Quintana Roo »

Jeremy Gibbs Free Energy wrote:What is the point of this infatuation with guaranteeing more top teams playing games against each other during the prelims? I simply don't think that is a criteria that should posses any relevance at all - not all tournaments can offer anything like that to begin with - but I also think that the point of playoffs, not prelims, should be to offer teams competitive games there. If half of your games are against similar enough teams, within the constraints of a given field, that's a high enough percentage for teams to be satisfied that they are playing competitive games. To me, the point of a pool prelim is so that you can play games against a sample of teams from across the gamut of competitiveness and see which general tier you are sorted out into so you can then do playoffs and get a much more precise ranking of where you fall within your tier.
This is pretty much exactly what i was waiting to say. As long as you do accurate, thorough, and data-verifiable analysis of the teams and their "rankings," then you can place them in fair brackets, then split them off in the playoffs in appropriate groups. Then, this assures competitive games for the second half of the day for everyone who feels like getting their money's worth.
Mr. Andrew Chrzanowski
Caesar Rodney High School
Camden, Delaware
CRHS '97-'01
University of Delaware '01-'05
CRHS quizbowl coach '06-'12
http://crquizbowl.edublogs.org
User avatar
Stained Diviner
Auron
Posts: 5085
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Power matching sucks: a theoretical offering

Post by Stained Diviner »

Given the choice between playing five competitive games and playing eight competitive games, I would choose eight. I have no idea why this point is controversial.
David Reinstein
Head Writer and Editor for Scobol Solo, Masonics, and IESA; TD for Scobol Solo and Reinstein Varsity; IHSSBCA Board Member; IHSSBCA Chair (2004-2014); PACE President (2016-2018)
User avatar
Sen. Estes Kefauver (D-TN)
Chairman of Anti-Music Mafia Committee
Posts: 5647
Joined: Wed Jul 26, 2006 11:46 pm

Re: Power matching sucks: a theoretical offering

Post by Sen. Estes Kefauver (D-TN) »

If the point of running a tournament were to do nothing but offer tons of competitive games, then you would only allow the 8 best teams in your field to play one another. That obviously isn't the point of running a tournament then, so one's ability to do so should not affect the format you use, much less influence you to use one that is worse than the alternatives.
Charlie Dees, North Kansas City HS '08
"I won't say more because I know some of you parse everything I say." - Jeremy Gibbs

"At one TJ tournament the neg prize was the Hampshire College ultimate frisbee team (nude) calender featuring one Evan Silberman. In retrospect that could have been a disaster." - Harry White
User avatar
The King's Flight to the Scots
Auron
Posts: 1645
Joined: Mon Jan 26, 2009 11:11 pm

Re: Power matching sucks: a theoretical offering

Post by The King's Flight to the Scots »

Regarding Mr. Reinstein's comment about competitive matches:

Sure, everyone loves close games. But that's why we have the playoffs. Beyond being the fairest way to determine who makes the playoffs, prelims serve the purpose of letting every team see the full range of the field and find out where they stand. The middling teams get to see how great teams play, and it inspires them to improve. The bottom teams play better teams and learn things instead of trudging through 90-60 matches the entire day. Just as importantly, it lets the top teams see what the rest of the field plays like so that when they write questions, they keep those lesser teams in mind.
Matt Bollinger
UVA '14, UVA '15
User avatar
Stained Diviner
Auron
Posts: 5085
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Power matching sucks: a theoretical offering

Post by Stained Diviner »

While holding competitive matches is not the only reason to have a tournament, it's one of the major reasons, so doing it more is a significant plus.

Also, the fairest way to determine who makes the playoffs is power matching. Power matching does give some of the blowouts that some of you want, though they mostly all occur in the first two rounds rather than occurring over and over again for the first six rounds. Weak teams learn by playing weak teams, since they get to hear the entire tossup. Strong teams learn by playing strong teams, since they see where they need to buzz to do well and which bonus parts they should be getting.
David Reinstein
Head Writer and Editor for Scobol Solo, Masonics, and IESA; TD for Scobol Solo and Reinstein Varsity; IHSSBCA Board Member; IHSSBCA Chair (2004-2014); PACE President (2016-2018)
User avatar
Ben Dillon
Rikku
Posts: 323
Joined: Thu Apr 28, 2005 10:47 pm
Location: South Bend, IN
Contact:

Re: Power matching sucks: a theoretical offering

Post by Ben Dillon »

Matt Weiner wrote:The only tournament that I can think of where power matching isn't obviously inferior to brackets from the get-go is the HSNCT, and that only because they do several other things (most importantly, take a huge number of teams to the playoffs and spend an appropriate amount of time and effort on initial seeding) that are necessary in order to make power matching something less than a disaster.
Okay, I've clearly missed something here. NAQT seeds the HSNCT? They sure don't seem to advertise this fact when they explain the card system, since they say the cards don't matter at the beginning of the tournament, only at the end.

To the larger issue, from years of being a TD in chess: Power matching (Swiss system) works best with seeding (ELO rating), but it's awfully hard to seed given that there isn't one universally agreed upon rating system. The card system also uses incomplete information, because it treats the winning team the same coming out of a game while ignoring the fact that there may have been an upset. (In chess, the rounds are paired as you go.)

I do think Matt is correct about there being some "reward" for tanking, but perhaps it's not as noticeable as in chess with its draw possibility. Top players who draw their first game usually end up being paired down for a few rounds, so they end up with an easier path. (Of course, this is my pet peeve with the Swiss system in chess: if three players are tied for the lead, #1 plays #2 and, oddly, #3 gets to play down.)

Here's a few potential solutions:
(1) Reseed the teams -- that is, hand out new cards -- halfway through the tournament to adjust for any reshuffling that has occurred from the original seedings. New seedings can be calculated based on results of the previous games.
(2) Pair the teams manually for the last one or two rounds, especially to put together teams "on the bubble" for playoffs.
Ben Dillon, Saint Joseph HS

"Why, sometimes I've believed as many as
six impossible things before breakfast!"
User avatar
Sen. Estes Kefauver (D-TN)
Chairman of Anti-Music Mafia Committee
Posts: 5647
Joined: Wed Jul 26, 2006 11:46 pm

Re: Power matching sucks: a theoretical offering

Post by Sen. Estes Kefauver (D-TN) »

I don't see it proven anywhere that power matching is in fact the fairest possible format. A number of people who have had far more involvement in running good tournaments such as Dwight Wynne and Mike Sorice have posted in this thread that they do not agree that power matching is the fairest option, and I may be committing some leaps of judgment, but in my experience emulating Mike Sorice seems to work so I trust his opinion here highly. I will continue to remain unconvinced that power matching is fair until you convince me otherwise, which you have not done yet.

Nonetheless, I don't even see offering competitive games as a "major reason" for running a tournament. I directed a tournament recently where NKC averaged a margin of 263 Mrg, with their closest game coming in at a 180 point difference. Given those statistics, I would say our site failed to meet a "major reason" for running a tournament, according to you, since we failed to offer North Kansas City games that would be keeping them on their toes and possibly losing depending on how a tossup here or there goes. My choices then are that I accept your position and admit that through no fault of my own I failed to provide something integral to the experience of running a quizbowl tournament, or else I decide that's nonsense and keep doing what I've been doing. Guess which choice I'll make.
Charlie Dees, North Kansas City HS '08
"I won't say more because I know some of you parse everything I say." - Jeremy Gibbs

"At one TJ tournament the neg prize was the Hampshire College ultimate frisbee team (nude) calender featuring one Evan Silberman. In retrospect that could have been a disaster." - Harry White
User avatar
Sen. Estes Kefauver (D-TN)
Chairman of Anti-Music Mafia Committee
Posts: 5647
Joined: Wed Jul 26, 2006 11:46 pm

Re: Power matching sucks: a theoretical offering

Post by Sen. Estes Kefauver (D-TN) »

Okay, I've clearly missed something here. NAQT seeds the HSNCT? They sure don't seem to advertise this fact when they explain the card system, since they say the cards don't matter at the beginning of the tournament, only at the end.
NAQT makes a point of throwing up a cloud of smoke about it in the opening meeting, but if you listen carefully in the 2 years I played they actually made it quite clear that they seed teams going into the tournament. The point they emphasize is that the number you are given has nothing to do with seeding - they don't necessarily give out the cards in order to their seeding. Instead, what they obviously do is seed some expected top teams, then move them to random points in their card system where these teams won't be running into each other until at hopefully around the fourth game.
Charlie Dees, North Kansas City HS '08
"I won't say more because I know some of you parse everything I say." - Jeremy Gibbs

"At one TJ tournament the neg prize was the Hampshire College ultimate frisbee team (nude) calender featuring one Evan Silberman. In retrospect that could have been a disaster." - Harry White
User avatar
Whiter Hydra
Auron
Posts: 1418
Joined: Tue Dec 04, 2007 8:46 pm
Location: Fairfax, VA
Contact:

Re: Power matching sucks: a theoretical offering

Post by Whiter Hydra »

Shcool wrote:Also, the fairest way to determine who makes the playoffs is power matching.
No, the fairest way to determine who makes the playoffs is a triple RR between all the teams present, with ties broken by best-of-5 matches.
Harry White
TJHSST '09, Virginia Tech '13

Owner of Tournament Database Search and Quizbowl Schedule Generator
Will run stats for food
User avatar
Huang
Rikku
Posts: 442
Joined: Wed Dec 17, 2008 9:29 pm

Re: Power matching sucks: a theoretical offering

Post by Huang »

Shcool wrote:While holding competitive matches is not the only reason to have a tournament, it's one of the major reasons, so doing it more is a significant plus.
A significant plus would be holding a tournament that correctly determines the final standings of teams that are statistically superior to teams that rank below them and statistically inferior to teams that rank above them.
Sandy
User avatar
Stained Diviner
Auron
Posts: 5085
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Power matching sucks: a theoretical offering

Post by Stained Diviner »

To add some clarification to this debate, one of the big differences between our tournament and the alternatives that other people are suggesting is that our morning format was brutal. We went from 32 teams to 6 teams in 5 rounds. The only way to do this with pools would be to only allow pool winners to advance, a format I hope we can all agree would be bad. What we did is akin to only allowing the 8-2 teams at HSNCT to advance to the second day. The only alternatives that have been offered to what we did are to have a single elimination playoff or to force teams to play more games than they would tolerate--in other words, the only alternatives being suggested are horrible. Absolutely horrible. At least Earthquake wasn't being serious when he made his horrible suggestion.

To answer Huang's last post, let me repeat again that the reason the statistically superior teams did not make the top bracket is that they lost head-to-head against the statistically inferior teams that did make it. If this isn't proof that what we did is right, then I don't know what would constitute proof.

To answer Charlie's last post, let me state that you ran the best tournament you could under the circumstances. It would have been a better tournament experience for NKC if there were some teams there close to their skill level, but that does not mean that the tournament was a waste of time or that anybody did anything wrong. New Trier could have run a better tournament if all of our teams wanted to play until midnight, but we, like you, did the best we could under the circumstances.
David Reinstein
Head Writer and Editor for Scobol Solo, Masonics, and IESA; TD for Scobol Solo and Reinstein Varsity; IHSSBCA Board Member; IHSSBCA Chair (2004-2014); PACE President (2016-2018)
User avatar
Important Bird Area
Forums Staff: Administrator
Posts: 6113
Joined: Thu Aug 28, 2003 3:33 pm
Location: San Francisco Bay Area
Contact:

Re: Power matching sucks: a theoretical offering

Post by Important Bird Area »

Jeremy Gibbs Free Energy wrote:they actually made it quite clear that they seed teams going into the tournament. The point they emphasize is that the number you are given has nothing to do with seeding - they don't necessarily give out the cards in order to their seeding.
Cloud of smoke notwithstanding, this is an accurate description of our system.
Jeremy Gibbs Free Energy wrote:Instead, what they obviously do is seed some expected top teams, then move them to random points in their card system where these teams won't be running into each other until at hopefully around the fourth game.
This isn't quite right; we seed everyone with the intent of making the initial power-matching groups as balanced as we can (and then secondarily account for geographic diversity).
Jeff Hoppes
President, Northern California Quiz Bowl Alliance
former HSQB Chief Admin (2012-13)
VP for Communication and history subject editor, NAQT
Editor emeritus, ACF

"I wish to make some kind of joke about Jeff's love of birds, but I always fear he'll turn them on me Hitchcock-style." -Fred
User avatar
Ben Dillon
Rikku
Posts: 323
Joined: Thu Apr 28, 2005 10:47 pm
Location: South Bend, IN
Contact:

Re: Power matching sucks: a theoretical offering

Post by Ben Dillon »

Jeremy Gibbs Free Energy wrote:The point they emphasize is that the number you are given has nothing to do with seeding - they don't necessarily give out the cards in order to their seeding.
Then why number the cards? This is misleading to new teams, who draw the wrong conclusion about the numbers.

Instead, how about making the cards with fictional characters for identification. Might be fun for the hosts in the room to be able to say, "Next we'll see the match between "The Little Mermaid" and "Bill Sykes" :)
Ben Dillon, Saint Joseph HS

"Why, sometimes I've believed as many as
six impossible things before breakfast!"
User avatar
Stained Diviner
Auron
Posts: 5085
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Power matching sucks: a theoretical offering

Post by Stained Diviner »

To add one more reason why the alternative suggestions being made about our tournament are horrible--they are suggesting a reseeding after Round 7. Doing so would give us a choice of having lunch start around 1:30 or 1:45 or having a 40 minute break built into our schedule so that we could reseed teams--about 20-25 minutes to enter stats into SQBS, about 10-15 minutes to convert those stats into room placements, and about 5-10 minutes to tell everybody where to go and let them get there. Once a tournament gets beyond 16 teams or so, logistical issues like this become a real headache. Also, running pools with wildcards requires either paper tiebreakers or extra playoff matches in the middle of the tournament while most teams sit around or play meaningless matches.

Also, I like Ben's suggestion, and it is something we could do in the future.
David Reinstein
Head Writer and Editor for Scobol Solo, Masonics, and IESA; TD for Scobol Solo and Reinstein Varsity; IHSSBCA Board Member; IHSSBCA Chair (2004-2014); PACE President (2016-2018)
User avatar
Important Bird Area
Forums Staff: Administrator
Posts: 6113
Joined: Thu Aug 28, 2003 3:33 pm
Location: San Francisco Bay Area
Contact:

Re: Power matching sucks: a theoretical offering

Post by Important Bird Area »

Ben Dillon wrote:
Jeremy Gibbs Free Energy wrote:The point they emphasize is that the number you are given has nothing to do with seeding - they don't necessarily give out the cards in order to their seeding.
Then why number the cards? This is misleading to new teams, who draw the wrong conclusion about the numbers.
Because numbers are a straightforward way to keep track of 192 items. Besides, that's why we announce that the numbers don't correspond to seeding.

Would some kind of alphabetical system be better? ("Card XY will now play card QZ")
Jeff Hoppes
President, Northern California Quiz Bowl Alliance
former HSQB Chief Admin (2012-13)
VP for Communication and history subject editor, NAQT
Editor emeritus, ACF

"I wish to make some kind of joke about Jeff's love of birds, but I always fear he'll turn them on me Hitchcock-style." -Fred
User avatar
Mechanical Beasts
Banned Cheater
Posts: 5673
Joined: Thu Jun 08, 2006 10:50 pm

Re: Power matching sucks: a theoretical offering

Post by Mechanical Beasts »

Shcool wrote:While holding competitive matches is not the only reason to have a tournament, it's one of the major reasons, so doing it more is a significant plus.
The best tournament:
1) Pick the eight best teams pre-tournament
2) Have them play twelve rounds against each other; don't worry about keeping their strengths of schedule similar
3) Declare a winner

That's twelve competitive matches! You of course lose a little because you're relying on past tournament results to decide who is in the championship bracket. But it could be worse. It'll actually do a pretty good job of picking a top bracket--will you be more than two teams off or so? And since you're comfortable deciding who is the best even when strength of schedule varies, what's wrong with picking a winner out of such a bracket?
Also, the fairest way to determine who makes the playoffs is power matching.
With win-loss record alone? No. With an advanced stat that judges points against a hypothetical team by seeing your points and the strength of your opponents? Possibly.
Andrew Watkins
User avatar
Important Bird Area
Forums Staff: Administrator
Posts: 6113
Joined: Thu Aug 28, 2003 3:33 pm
Location: San Francisco Bay Area
Contact:

Re: Power matching sucks: a theoretical offering

Post by Important Bird Area »

Ben Dillon wrote:Then why number the cards? This is misleading to new teams, who draw the wrong conclusion about the numbers.

Instead, how about making the cards with fictional characters for identification. Might be fun for the hosts in the room to be able to say, "Next we'll see the match between "The Little Mermaid" and "Bill Sykes" :)
R. writes: (something that I should have thought of and missed!)

"The system has to allow a clear and un-screw-up-able determination of which team takes which card after each match. Using numbers makes that really easy.

My off-the-cuff answer is that any system we choose is going to have to have a total ordering, which will make it just like using numbers and re-introduce the misconception that the numbers directly relate to team ability.

We could use random words in alphabetical order, for example, but I think people will still think of 'aardvark' as being better than 'zebra.'"
Jeff Hoppes
President, Northern California Quiz Bowl Alliance
former HSQB Chief Admin (2012-13)
VP for Communication and history subject editor, NAQT
Editor emeritus, ACF

"I wish to make some kind of joke about Jeff's love of birds, but I always fear he'll turn them on me Hitchcock-style." -Fred
User avatar
Whiter Hydra
Auron
Posts: 1418
Joined: Tue Dec 04, 2007 8:46 pm
Location: Fairfax, VA
Contact:

Re: Power matching sucks: a theoretical offering

Post by Whiter Hydra »

Shcool wrote:We went from 32 teams to 6 teams in 5 rounds.
And you went from 48 teams to 32 teams in 0 rounds, so arbitrarily excluding B teams must be better!
Harry White
TJHSST '09, Virginia Tech '13

Owner of Tournament Database Search and Quizbowl Schedule Generator
Will run stats for food
User avatar
Stained Diviner
Auron
Posts: 5085
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Power matching sucks: a theoretical offering

Post by Stained Diviner »

Here's another suggestion, Andy. We could pick the top eight seeds before the tournament and do everything in our power to prevent them from playing competitive matches for the first seven rounds. When they win their first seven matches, we could declare them the playoff teams. It's kind of like your system, except it takes four hours longer. We could call it the pool system.

I don't know why advanced stats are needed when we already put the teams in a room and had them play each other. Advanced stats are a fallback for when you don't want to take the time or devise a system to have the marginal playoff teams play each other for the final spots, which sometimes is the case but in our case was not.
David Reinstein
Head Writer and Editor for Scobol Solo, Masonics, and IESA; TD for Scobol Solo and Reinstein Varsity; IHSSBCA Board Member; IHSSBCA Chair (2004-2014); PACE President (2016-2018)
User avatar
Mechanical Beasts
Banned Cheater
Posts: 5673
Joined: Thu Jun 08, 2006 10:50 pm

Re: Power matching sucks: a theoretical offering

Post by Mechanical Beasts »

Shcool wrote:I don't know why advanced stats are needed when we already put the teams in a room and had them play each other. Advanced stats are a fallback for when you don't want to take the time or devise a system to have the marginal playoff teams play each other for the final spots, which sometimes is the case but in our case was not.
Advanced stats that correct for variable strength of schedule would probably be useful for situations where teams have a variable strength of schedule. Right?
Andrew Watkins
User avatar
Whiter Hydra
Auron
Posts: 1418
Joined: Tue Dec 04, 2007 8:46 pm
Location: Fairfax, VA
Contact:

Re: Power matching sucks: a theoretical offering

Post by Whiter Hydra »

Shcool wrote:Here's another suggestion, Andy. We could pick the top eight seeds before the tournament and do everything in our power to prevent them from playing competitive matches for the first seven rounds. When they win their first seven matches, we could declare them the playoff teams. It's kind of like your system, except it takes four hours longer. We could call it the pool system.
Assume the PACE NSC 2010 is happening tomorrow. Give me the top 8 teams. Chances are, you won't pick the same top 8 as everyone else. This is why we play these things called games.

Plus, the point of these tournaments aren't about winning. It's also about how well you can do. If people only played to win, it would be pointless for a team like Cosby B to even enter in the same tournament as Maggie Walker A. Structured consolation brackets exist for a reason, y'know.
Harry White
TJHSST '09, Virginia Tech '13

Owner of Tournament Database Search and Quizbowl Schedule Generator
Will run stats for food
User avatar
cvdwightw
Auron
Posts: 3291
Joined: Tue May 13, 2003 12:46 am
Location: Southern CA
Contact:

Re: Power matching sucks: a theoretical offering

Post by cvdwightw »

I think this statement from Mike Sorice sums up exactly my feelings on this matter:
Captain Sinico wrote:Furthermore, the following assertions have not been proven: first, that power matching is more sensitive to seeding than bracketing; second, that single elimination rounds are inherently fairer than paper tiebreaker use for marginal playoff teams; third, all the conclusions, even taking as written the premises. The fact is that I agree with some of the premises and conclusions, but my analysis indicates that the statements made about them are far too strong.
I don't know exactly what you're arguing in that first premise. If you mess up your seedings, then I suspect that power-matching will be more robust than bracketed play. How do we test this? We can run simulations of different seeding patterns and see which one works better. Why don't we do this? Simply put, because (1) I don't have the time or computer-knowledge to run thousands of simulations of 32-team tournaments and (2) I don't know what the heck you mean by "[sensitivity] to seeding." Is this like flipping the 5 and 6 seeds? Flipping 5 through 8? Picking bracket/initial assignments out of a hat? Without knowing what the boundaries of this (i.e. the input parameters) are, it's impossible to create a simulation to verify or disprove this hypothesis. All we have is your word that given whatever definitions you've arbitrarily taken, your conclusions are correct. This is a recurring theme in your argumentation style. The reason that I've described your argument as "some kind of vendetta" is that you consistently make sweeping assumptions (e.g. "The power-matching fad...is something that exists for no real reason beyond emulating what a prestigious tournament does") that turn out to be at best impossible to prove and at worst false (this depending on whether or not you believe Reinstein's claim to have started power-matching in 2001; NAQT did not power-match HSNCT until 2002 and did so with a problematic schedule; power-matching has been standard in several non-quizbowl activities for longer than that), then ask us to accept the conclusions that arrive from premises that you have not put forth. Perhaps "vendetta" is not the best word to use, but "ranting" sure seems applicable.

You've given us a three-part thesis and the only really defended the part that "giving teams wildly different strength-of-schedules is unfair." I agree with this premise, but I do not at all agree with the conclusion that power-matching somehow results in a less fair ranking than bracketed play. In a bracketed system, if you screw up seeds 5-10 out of 32 and only take two teams out of those six, then two teams will be eliminated by finishing third in their bracket and another two will be eliminated on paper or one-shot tiebreakers. What's the way to solve this? More math. And again, since I don't know what you're defining your input parameters to be, it's impossible to set up a real simulation.

Your argument that "Stevenson was not allowed a bad match, while others were and made it in" also doesn't hold - if Stevenson finished 2nd in its pool and had to play a tiebreaker game, it would be allowed only one bad match. Meanwhile, a team that got upset could still make its way out of a 3-way circle of death or a 2-way tie for 2nd; then, that team would be allowed one bad match before having to take on another 2-seed in the tiebreaker for the playoffs.

I'm not saying that I disagree with you that bracketed prelims is better than power-matched prelims - I for one prefer bracketed prelims and playoffs - but I disagree with how you are saying it. You have yet to throw out any argument on this point that dissuades me from entertaining the possibility that power-matching might be equally good.
Dwight Wynne
socalquizbowl.org
UC Irvine 2008-2013; UCLA 2004-2007; Capistrano Valley High School 2000-2003

"It's a competition, but it's not a sport. On a scale, if football is a 10, then rowing would be a two. One would be Quiz Bowl." --Matt Birk on rowing, SI On Campus, 10/21/03

"If you were my teammate, I would have tossed your ass out the door so fast you'd be emitting Cerenkov radiation, but I'm not classy like Dwight." --Jerry
Locked