web hosting

Fun With Numbers: More Statistical Breakdowns of This Year’s Season

7
March

The final qualifying tournaments for the 2011 Tournament of Champions were held this past weekend. With the postseason now beginning, we can look back at the regular season and assess the performances of some of the country’s best teams. Who earned the most bids to the TOC? Who earned the best bids to the TOC? Who won the most preliminary rounds at octafinals-level qualifying tournaments? Who had the best winning percentage on the affirmative and negative? The answers are below the fold.

TOC Bids

Thanks to Rohan Sadogopal’s TOC Bid Spreadsheet, we know that Kush Patel and Chris Patterson of Greenhill School were this year’s leaders with nine bids to the TOC. The rest of the top ten is as follows:

  1. Greenhill — Kush Patel & Chris Patterson, 9
  2. Westminster — Ellis Allen & Daniel Taylor, 8
  3. St. Mark’s — Rishee Batra & Alex Miles, 8
  4. Lexington — Tyler Engler & Arjun Vellayappan, 7
  5. Carrollton — Anna Dimitrijevic & Fabi Urdaneta, 7
  6. Gulliver Prep — Greg Adler & Jorge Toledo, 7
  7. Greenhill — Akshay Bhushan & Lyall Stuart, 7
  8. Mountain Brook — Evan McCarty & Lee Quinn, 7
  9. Glenbrook South — Jacob Hurwitz & Marc Jacome, 7
  10. Georgetown Day — Joe Krakoff & Ben Levy, 7
  11. Edina — Mimi Sergent-Leventhal & Erin Sielaff, 7

But all bids are not created equal; those earned at octafinal-level qualifying tournaments, for example, are generally considered most impressive. What if we translate raw bid counts into weighted bid counts? A simple way to do so is to value each bid according to the number of bids awarded at the tournament. In this system, octafinals-level tournaments are worth 16 points, quarterfinals-level tournaments are worth 8 points, semifinals-level tournaments are worth 4 points, and finals-level tournaments are worth 2 points. The teams with the highest valued bids in this weighted bid system are as follows:

  1. Westminster — Ellis Allen & Daniel Taylor, 120
  2. Greenhill — Kush Patel & Chris Patterson, 116
  3. St. Mark’s — Rishee Batra & Alex Miles, 112
  4. Lexington — Tyler Engler & Arjun Vellayappan, 112
  5. Carrollton — Anna Dimitrijevic & Fabi Urdaneta, 104
  6. Gulliver Prep — Greg Adler & Jorge Toledo, 100
  7. Greenhill — Akshay Bhushan & Lyall Stuart, 96
  8. Glenbrook North — Alex Pappas & Zack Parker, 96
  9. Mountain Brook — Evan McCarty & Lee Quinn, 90
  10. Oak Park-River Forest — James Hanley & Will Hardwicke, 84

Winning Percentage

Ever wonder which teams had the best overall winning percentage? Or which teams had the best records on the affirmative or negative? While variations in travel schedules make these comparisons difficult, it is possible to analyze winning percentages at octafinals-level qualifying tournaments. Of the 65 teams that are fully qualified for the TOC, all but 16 competed in at least 20 preliminary rounds at octafinals tournaments. Among those teams, the following had the highest overall preliminary round winning percentages:

  1. St. Mark’s — Rishee Batra & Alex Miles, 94.87% (37-2)
  2. Gulliver Prep — Greg Adler & Jorge Toledo, 89.74% (35-4)
  3. Glenbrook North — Alex Pappas & Zack Parker, 89.47% (34-4)
  4. Westminster — Ellis Allen & Daniel Taylor, 88.89% (40-5)
  5. College Prep — Vinay Pai & Tatsuro Yamamura, 87.10% (27-4)
  6. Mountain Brook — Evan McCarty & Lee Quinn, 84.85% (28-5)
  7. Carrollton — Anna Dimitrijevic & Fabi Urdaneta, 84.09% (37-7)
  8. Georgetown Day — David Herman & Isaac Stanley-Becker, 82.14% (23-5)
  9. Greenhill — Akshay Bhushan & Lyall Stuart, 81.82% (27-6)
  10. Georgetown Day — Joe Krakoff & Ben Levy, 81.82% (27-6)

What about the best teams on a particular side of the resolution? The following teams had the best winning percentages on the affirmative:

  1. Oak Park-River Forest — James Hanley & Will Hardwicke, 95.24% (20-1)
  2. Chattahoochee — Maggie Davis & Mustafa Inamullah, 93.75% (15-1)
  3. College Prep — Vinay Pai & Tatsuro Yamamura, 93.33% (14-1)
  4. Pine Crest — Matt Marcus & David Rubin, 93.33% (14-1)
  5. Westminster — Ellis Allen & Daniel Taylor, 91.67% (22-2)
  6. St. Mark’s — Rishee Batra & Alex Miles, 88.89% (16-2)
  7. Glenbrook North — Alex Pappas & Zack Parker, 88.89% (16-2)
  8. Mountain Brook — Evan McCarty & Lee Quinn, 88.24% (15-2)
  9. Homewood-Flossmoor — Rufus Coates-Welsh & Laura River, 85.71% (12-2)
  10. Beacon — Damiyr Davis & Miguel Feliciano, 85.00% (17-3)
  11. Greenhill — Kush Patel & Chris Patterson, 85.00% (17-3)

The following teams had the best records on the negative:

  1. St. Mark’s — Rishee Batra & Alex Miles, 100.00% (21-0)
  2. Gulliver Prep — Greg Adler & Jorge Toledo, 100.00% (20-0)
  3. Carrollton — Anna Dimitrijevic & Fabiola Urdaneta, 95.45% (21-1)
  4. Georgetown Day — Joe Krakoff & Ben Levy, 93.33% (14-1)
  5. Damien — Nadeem Farooqi & Pablo Gannon, 93.33% (14-01)
  6. Georgetown Day — David Herman & Isaac Stanley-Becker, 92.31% (12-1)
  7. Glenbrook North — Alex Pappas & Zack Parker, 90.00% (18-2)
  8. Westminster — Ellis Allen & Daniel Taylor, 85.71% (18-3)
  9. College Prep — Vinay Pai & Tatsuro Yamamura, 81.25% (13-3)
  10. Mountain Brook — Evan McCarty & Lee Quinn, 81.25% (13-3)
  11. Greenhill — Akshay Bhushan & Lyall Stuart, 81.25% (13-3)

Overall, teams that have qualified for the TOC won 75.53% of their affirmative preliminary rounds and 71.48% of their negative preliminary rounds at octafinals-level qualifying tournaments for a combined winning percentage of 73.52%.

Posted on by Bill Batterman

60 Comments

Category: Debate News, Results

60 Responses to Fun With Numbers: More Statistical Breakdowns of This Year’s Season

  2. dtay

    Why only prelim winning percentages? It seems to me like that statistic would be the least meaningful of winning percentages, since its the most influenced by luck of the draw. Was it just a "its hard to get side data for elims" issue or was there a statistical reason for that choice?

    • Kyle S.

      how is that least meaningful? its still something and its still impressive by how much the good teams dominate so consistently

      • Ryan

        ^this guy is right. If there's a bias, it probably effects all teams equally. The only way that dtay's claim can hold water is if you presume that some teams are "luckier" than others, and that they are lucky often enough to significantly overwhelm skill disparities. Also, going negative isn't ALWAYS better, it varies on a case by case basis.

        • Vinay

          prelims and elims are both influenced by the same degree of luck – they're both paired within brackets, and because the probability of getting a side you prefer does not change significantly, computer flip is more or less the same as flipping before an elim debate

          elim data IS very important, however, because it does a better job of measuring quality of wins; good avg elim depth means you have a high win percentage against the top several percent of each tournament's respective pool. I don't think that means prelims are "least meaningful" with respect to win percentages….prelims and elims just measure different things

          • dtay

            Prelims are undeniably more erratic data than elim data. Meaningless? Certainly not. Its impressive to have good prelim winning percentages. Just more erratic than elims.

            The way prelims are paired creates far more unequal debates, and those debate are unequal by larger margins. The largest skill disparity in say a quarters round is much less than the largest skill disparity in say round 4 of a tournament. This doesn't apply to say rounds 6-7, where its mostly just the top teams again, but in the middle of prelim sets you get vast disparities in how difficult a draw teams have. Some teams will get draws that for all intents and purposes are automatic wins, and some, maybe they got a single bad speaking round, will get someone they could be debating in finals. Does this happen in elims too? Yes. But less.

            More importantly, in particular the side data for prelims is horrendously skewed. Look at the top 4 aff prelim winning percentages. All of them are teams that aren't even in the top 10 in overall prelim winning percentages. What this means is they are winning their aff rounds by… losing a bunch of neg rounds, creating easier draws on the affirmative. Ironically being good on the negative hurts your aff winning percentage, while say being bad on the affirmative would help your neg winning percentage.

            Also not all 6-1's are created equal. A team that consistently loses their 4th prelim will get much easier prelim rounds than a team that consistently loses their 6th or 7th prelim. This isn't true for elim debates, where losing early cannot ironically help your win percentage, it just knocks you out. Versus in prelims losing early, while not "helpful" allows you to achieve the same thing a team that loses late did, but the team that lost late had a much harder time of it.

            Also again, its just true that 1 judge panels are more random than three judge panels. Yeah Yeah, adopt to your audience, its always the debater's fault, but honestly, think about that for a second. If 1 judge panels were as accurate as 3 judge panels, why do we bother with 3 judge panels in elims. Judge screwups do happen, and happen more often in prelims than elims because of smaller panels. And single screwups matter alot, because if you look at those numbers above the teams at the top have single digit prelim losses, so one more loss makes a huge difference. Like multiple spots in the ranking difference.

            Most of you'll objections are off base. I'm not saying the ability to pick your side or w/e, or its lucky whether you get neg or aff in prelims is why they are erratic. Clearly that also applies to elims. And no, the fact that prelims are more random does not make it equally "biased" for everyone. Thats… not what biased/random means. And I'm not saying prelims are "biased," just a more erratic/random set of data for determining skill/win percentage than elim data. Or combined prelim-elim data, or what have you.

            And yes, Elim data measures your success against the top teams better. What exactly does prelim data measure better than elim data? Success against the not top teams? I suppose, but most of those teams above really only lose to the top teams in prelims anyway. Its just who gets more hard prelims on particular sides is much more randomly determined than in elims.

            Again, not saying prelim statistic is meaningless. Its clearly not. Its just inferior to combined prelim-elim, so I was wondering why batterman deliberately only did prelims, to which the answer was just a lack of side data for elims. I'd be interested in what teams overall win percentages are though, even without side data, which should be easy enough to do for elims as well.

          • Vinay

            I largely agree with you; I only took issue with the claim that prelim statistics are "least meaningful" of winning percentages. You say, in passing, that erratic prelim pairing "doesn't apply to say rounds 6-7, where its mostly just the top teams again." That's the point – yes, it's possible to lose rounds 1 and 2 of a tournament and have a relatively easy draw for the rest of your prelims, but no team on this list is losing more than 1 prelim debate per tournament. Going 6-0/7-0 at a tournament is pretty difficult (though, maybe not for westminster AT) given hard matchups in later prelims, and even teams that lose an early prelim will have tough draws in the down-1 bracket. For example, at St. Mark's, GBN PP lost their round 3 debate and, despite being in a lower bracket, had to debate Carrollton DU in round 6. At the same tournament, Greenhill BS lost their round 1 debate, and still ended up with a rough round 6 draw against Lexington CS. In the end, it is still uniformly difficult for teams to finish with a record better than the bare minimum necessary to clear, because of those round 6 and round 7 debates.

            I disagree that prelim statistics are only useful for measuring success against "the not top teams." My general point is that success in late prelims still demands that you be able to compete against top teams. However, one important point that nobody has mentioned is that prelim results dictate elim results, insofar as they establish seed order and thus difficulty of elim debates. Of course, this also means that elim results are inclusive of prelim results obviating the need to study prelim results at all, but I still think the statistic is a useful predictor of who will perform well in elims, if nothing else.

            Re: skewed side data. First of all, it is factually incorrect that the top 4 aff prelim winning percentages are "all…teams that aren't even in the top 10 in overall prelim winning percentages." I hate to seem like I'm tooting my own horn, but if I'm not mistaken, Tatsuro and I are in both the top 10 overall and top 4 aff winning percentages….I really hope that's not because we are, as you say, "losing a bunch of neg rounds," but maybe you're right and I'm just a totally incompetent 2n. Beyond that, though, you're drawing a number of unfair conclusions – Oak Park is not successful because they're just a horrendously terrible neg team, they have picked up several solid aff wins (Carrollton DU, GBN PP, Dallas Jesuit GM, among others). Maggie Davis is also a pretty damn good 2a, and you've essentially said that Hooch's success on the aff is less attributable to her debate skill than to Musti's unwillingness to give a shit.

            Yes dtay, obviously if Batterman had included both prelim and elim results, you would be number one and the world would be a happy place again. But an overall win/loss statistic is analytically useless because you glean no meaningful information from the data set it other than noticing "hey, westminster AT seems to win a lot of debates." Side data is important, but beyond that, by bracketing off prelim results, this becomes more than just data processing for data processing's sake; we can make make comparisons between who is consistent between prelims and elims v. who does well in prelims but still manages to lose in the doubles of every other tournament (*ahem*) v. who has a terrible prelim average but still has consistent elim success. I couldn't care less about a chart that says "Westminster AT – Win: 9000, Loss:1" because it doesn't tell many anything more than "ellis and dtay are good at debate," something I already knew. It would be an interesting statistic to supplement the existing data, but it's not objectively 'superior,' and it certainly isn't an excuse to disregard the successes of the teams represented here.

          • dtay

            Doing this paragraphs at a time.

            Yeah, mostly agree with paragraph 1. And yes, finishing with X-1, X-0 prelim records is hard. And yes, its meaningful. And yes, the down 1 bracket often, actually always, contains tough debates. You seem to make me out to saying prelim statistics are meaningless, which I'm not. They mean stuff. They actually mean alot. My point was while they are an important metric, of all the ways to look at win percentage they are the least important. Still important, just less than pure elims, and elim+prelims. Its not like I'm saying getting perfect prelims is easy, or meaningless, just that focusing soley on that statistic doesn't make much sense from a gleaning accurate results point of view.

            My point wasn't that we should exclude prelims, but that deliberately excluding elims as this article does makes no sense from a mathematical point of view. From an ease in gathering data point of view, it obviously does. Hence why I asked batterman, since i was curious if I was missing something or he just didn't have the side data. Turns out, it was just the side data.

            I didn't say prelim statistics were only useful in gathering data against the not top teams. Notice the question mark. In the succeeding sentence i then pointed out why that point of view made no sense, since as you just said, you still hit hard teams in prelims. I was hypothetically proposing one possible reason to look at only prelims, then shooting it down. Please read more carefully next time, because I basically agree with this entire paragraph, since its what I said.

            Skewed side data –
            sorry, purely a clerical error. A second look shows that yes, you are. I'm sorry about that. My point still stands. Obviously anyone on either of these lists is good, but there's no denying the pattern that 3 of the top 4 aff winning teams not being on either the neg percentage list or the overall percentage exists. And it makes sense that if you lose a neg debate, your following aff debates are easier. And vice versa, if you lose an aff debate, your neg debates are then easier.

            And i didn't make any comparative claims on what any teams success on the aff was "mainly" attributable too. Hooch/OPRF etc are obviously good on the aff, since they have to WIN their affs debates obviously. My point is their apparently being weaker on the neg makes their aff percentages easier to obtain. Not that it alone gives them a good record, but that it helps. Which is just another reason prelim aff/neg win percentages are skewed, and looking at elim aff/neg win percentages would be more meaningful, since you don't get that skew. I suspect OPRF and Hooch would still be on the list, but not as high.

            And actually, idk about the Dallas Jesuit round, but both Cton and GBN were wins OPRF picked up in the doubles, so weren't included in this statistic anyway. So yes, OPRF is good on the aff, and has good aff wins… still not a reason prelims are the statistically best selection of rounds.

            4th paragraph –

            Actually if you include prelim and elim results St Marks is still number one by quite a bit, though thanks for trying to make it personal.

            I really addressed this above. Prelim success matters. it means something. Its interesting. Side data is interesting. I asked that question though since elim wins/combined prelim-elim wins seem to mean more. Not that we shouldn't look at prelims, they have some things to say, they have some important things to say, but just that soley focusing on prelims made no sense to me since it was the statistic most prone to being erratic. Obviously the best article would include everything – overall, prelims, elims, prelims correlated to elims, all sorts of stuff. I just asked that question because i was wondering what the thought process behind only including prelims was, and as I guessed it was purely an "ease in gathering data" issue.

            Short summary of my views so if more people post they don't miss the mark.

            -Prelims matter.
            -Elims matter more.
            -Focusing soley on prelims at the exclusion of elims seems to be the least effective way to gather data, but is obviously easier to do which is why I asked batterman a question.
            -The best analysis would obviously include everything.
            -Vinay is obviously not incompetant, and instead isinsanely good. Same with maggie. Though admittedly Musty doesn't give a shit, that is true.

          • Vinay

            sorry, shouldn't have tried to make it personal
            in any case, i'm not going to 'line-by-line' your post because I think I've made my position clear and, quite frankly, I don't care enough about this
            in the end, every one of these statistical breakdowns is interesting and informative, and we can agree to disagree about their relative utility given the absence of certain data sets

          • dtay

            for amusement I'm going to literally line by line this post… lol

            line 1 – ok
            line 2 – not complete thought yet
            line 3 – ok
            line 4- i agree
            line 5 – we can i suppose, though I think we even agree about this, since you haven't said elim data would harm the analysis… Because thats what you would have to defend to disagree with what I actually said, that excluding elim data is GOOD

          • pavan upadhyayula

            "relative utility" =/= absolute utility. you say that elim data is more important that prelim data, he said that elim data is equally as important as prelim data; making a relative comparison doesn't mean saying that we should exclude elim data altogether

          • dtay

            well, to be precise I said

            elim+prelim>elim>prelim

            And I questioned why the analysis seemed to deliberately exclude elims, a stance doesn't make sense unless it is a resource/time issue in finding data, which it was. So in order to really disagree with what I advocated, you'd have to say "solely looking at prelims is the best way to determine win percentages." Which I don't think anyone is going to say.

            Though I would love for someone to actually defend prelims meaning more than elims. No one actually has, because I don't think such a defense exists. That prelims matter? Yes. That prelims matter more than elims? No.

          • dtay

            as you just said on gchat…

            prelim+elim, problem solved. And its also a deceptively larger sample size, since most prelims for teams on this list aren't ones they are likely to lose. Most top teams only lose prelims to other top teams, so you're really only looking at a sample of "good prelim rounds," which isn't that much larger than elims.

          • JForbes

            At this point, there must be a group of people making a concerted effort to neg rep dtay's posts. There's no way such low scores could have occurred naturally.

          • strong

            I wonder how rounds that are lag-paired would work into this. its a very small set of Data, but i think it would effect aff/neg loses in an interesting way.

            I also think that an analysis of prelim records at the nationals would remedy this issue since all the rounds are preset. (granted, it would make other issues arise.)

            a set of statistics on preset debates would also be interesting to see because they are completely random debates.

            probably not a reason not to see elim records. but elim records are more well known so there isnt as much of a reason to include elim results if time/energy could be spent collecting prelim data that is already time consuming to calculate.

          • Ana

            The more rounds you win, you are power-paired against other good teams, meaning the chance of you dropping the next round increases.

            Example- if you were 4-0 you would hit another good 4-0. Let's say you dropped two neg rounds and you're 2-2, then you would probably hit a team that was not as good, therefore increasing your chance to pick up your round 5 aff round.

    • dtay

      we would be number 2 if you must know, and i don't care about that much. This really started off with me just asking a clarification question and then people accusing me of things, so I defended myself.

      Let me say this again – OBVIOUSLY EVERYONE ON THAT LIST ABOVE IS GOOD AND SHOULD BE PROUD OF X WHATEVER THING LIST SAYS. Its meaningful. My sole, sole, SOLE, point was that including elims could not possible harm the analysis and could only help, so I was wondering why they were not included. So I asked, and got my answer…

    • Melanie Johnson

      I can fill in a few holes from Michigan based on rounds I judged/rounds my teams were in.

      Doubles – Westminster AT was Aff vs. Dowling DN, St. Mark's BM was Neg vs. IC West HK
      Octas – Westminster AT was Neg vs. Brother Rice CH,
      Quarters – Westminster AT was Aff vs. IC West HD

      • Melanie Johnson

        Correction – Westminster AT was Neg in Doubles. Just double-checked saved flows on my computer.

        Megan is pretty sure GBN BH was Neg against GBS GT in UMich Doubles. Said she can't be sure, though.

    • Greg

      Yo this is greg from gulliver AT, here is our elims

      Dubs greenhill- Aff
      Octos greenhill- Aff

      Dubs st marks- bye
      Octos st marks- neg
      Quarters st marks- aff
      Semis st marks- neg
      Finals st marks- neg

      Dubs glenbrooks- neg
      Octos glenbrooks- aff
      Quarters glenbrooks- aff
      Semis glenbrooks- aff
      Finals glenbrooks- neg

      Dubs blake- neg
      Octos blake- neg
      Quarters blake- aff

      Dubs emory- neg
      octos emory- aff

      Dubs harvard- Neg
      Octos harvard- aff
      Quarters harvard- aff

    • Valerie McIntosh

      In doubles at Emory OPRF HH was aff when they beat Carrollton DU not neg as the spreadsheet indicates.

  15. Bill Batterman

    Remaining missing sides — fill these in so I can add elim statistics. Facebook/email your friends if they were in these debates… shouldn't be that hard to find out. Thanks!

    Michigan – Doubles – OPRF HH d. Northside CT

    Michigan – Doubles – Lex CS d. Hoflo JO

    Glenbrooks – Doubles – Carrollton DU d. Westminster MW

    Glenbrooks – Doubles – OPRF HH d. Dallas Jesuit GM

    Harvard – Doubles – GBS VW d. Lexington BE

    Berkeley – Doubles – Juan Diego BS d. Heritage Hall HS

    Berkeley – Doubles – Hoflo CR d. Little Rock AP

    Berkeley – Doubles – St. George's LN d. Johns Creek DZ

    Berkeley – Doubles – GBN BoHi d. Johns Creek HT

    Berkeley – Doubles – St. Francis AP d. St. Vincent FS

    Berkeley – Doubles – Brophy MS d. Bingham GT

    Berkeley – Octas – Juan Diego BS d. Mercer Island GP

    Berkeley – Octas – Hoflo CR d. GBN BoHi

    Berkeley – Octas – Hoflo CR d. Juan Diego BS