Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
Math

Can Wolfram Alpha Tell Which Team Will Win the Super Bowl? 126

Posted by samzenpus
from the and-the-answer-is dept.
Nerval's Lobster writes "Which football team will win the Super Bowl this weekend? That's a multi-million-dollar question, given the amount of cash people will bet on either the Seattle Seahawks or the Denver Broncos to win. Fortunately, Wolfram Alpha (the self-billed "computational knowledge engine") can analyze the historical statistics for both teams and throw out some potentially useful numbers. Developed by Stephen Wolfram and based his Wolfram Research's Mathematica analytical platform, Wolfram Alpha is an altogether different search engine from Bing or Google, which generally return pages of blue hyperlinks in response to queries. Instead of multiple results leading to still other Webpages, Wolfram Alpha usually returns set of definitive, numerical answers. (A lengthy rundown of the engine's capabilities is found on its 'About' page.) So how does Wolfram's engine, which features sophisticated algorithms chewing through trillions of pieces of data, break down the potentials for Sunday's game? Out of the 38 times the two teams have met on the field, the Broncos have triumphed 25 times (versus 12 wins for the Seahawks), scoring 98 total touchdowns to the Seahawks' 84. It's definitely advantage Broncos, in that sense. But the teams' percentages are fairly close with regard to total yardage, penalties, penalty yards, and other metrics, although the Seahawks have managed to nab more interceptions (47, versus the Broncos' 37). But while Wolfram Alpha can crunch all the historical data it wants, and that data can suggest one team will likely triumph over another, there's always the likelihood that something random—a freak injury, or a tweak to the player lineup—can change the course of the game in ways that nobody can anticipate. Also, given how player and coaching rosters vary from year to year, the teams taking the field can change radically between meetings." EA has correctly predicted eight of the last ten Super Bowl winners using the latest Madden game.
This discussion has been archived. No new comments can be posted.

Can Wolfram Alpha Tell Which Team Will Win the Super Bowl?

Comments Filter:
  • No (Score:5, Interesting)

    by El_Muerte_TDS (592157) <elmuerte AT drunksnipers DOT com> on Sunday February 02, 2014 @11:36AM (#46134389) Homepage

    Of course not. This is just as stupid as asking if you could calculate somebody's phone number.

    • Re:No (Score:5, Funny)

      by TheloniousToady (3343045) on Sunday February 02, 2014 @12:34PM (#46134671)

      I think you're being too pessimistic. I give Wolfram Alpha a 50/50 chance of getting the winner right. That said, I give them lesser odds for predicting the coin toss.

      • by ClaraBow (212734)
        Funny!
      • Sorry to gloat, but now that the Superbowl is over I'd like to point out that I was absolutely right about Wolfram Alpha having a 50/50 chance of predicting the winner. And not only did Wolfram Alpha fail to predict the coin toss, it also failed to predict that the coin would be tossed twice. Clearly, Wolfram needs to develop a new kind of (computer) science.

      • by terryk29 (2756467)
        No, it's 55/55. This is the pro sports -cent, remember, which is 110.
    • Of course not. This is just as stupid as asking if you could calculate somebody's phone number.

      That's the great thing about close plays in sports: In one instant, you have tiny chaotic fluctuations like the air turbulence or a bounce on rough ground. The next instant, it becomes "destiny", and it gets analyzed and discussed by pundits for decades to come.

    • by gstoddart (321705)

      Ah, but I figure you could guess right with 50% accuracy, and then if you guessed right and claim you calculated it nobody will know the difference.

      If it predicts the next 10 or so, maybe. But anything else is mostly just PR.

      • This reminds me of a story I heard once (maybe it was from a movie, or an XKCD, can't track it down right now), in which a pair of guys meet a random girl:

        Guy 1: think of a card . . .
        Girl: okay
        Guy 1: your card is eight of hearts
        Girl: no it's 3 of diamonds
        \later
        Guy 2: Why did you think you knew her card?
        Guy 1: I didn't, but I figure I have about a 2% chance of guessing it, and if I do this to everyone I meet then when I do get it right the reaction will be worth all the times I got it wrong.

    • Of course not. This is just as stupid as asking if you could calculate somebody's phone number.

      Not really. Although I didn't see much in the stats given that require much of anything from Wolfram Alpha. A spreadsheet would have done as well. Though maybe it was easier to do via Alpha.

      More to the point: from OP

      "Also, given how player and coaching rosters vary from year to year, the teams taking the field can change radically between meetings."

      Comparing how the teams did when facing each other in past years isn't really going to give you a whole lot of useful information. Better to compare how they did against other teams, this year.

    • by AK Marc (707885)
      Wy can't you? The first 6 numbers are based on location, and the last 4 have a temporal component. If you know where someone lives, and when they moved here, you could calculate a range of numbers that would be likely. Much like SSN can ba "calculated" given enough information and partials of the number.
      • by danudwary (201586)

        Hasn't been that way in a long time. Until recently, my wife had an Arizona cell phone number, and we haven't lived there in a decade. I have three phone numbers, none of which has the same area code. 15 years ago I had a choice of the last four digits of my phone number from a short list (I chose 0666. The operator was like "Are you SURE?"), so that's not temporal either. Between vanity numbers, reuse of numbers, and internet VOIP numbers, etc, it's all muddy. Maybe this system worked for your parents or g

    • I'm writing this after the game. EA's Madden thought Denver was the winner. Wolfram Alpha thought a free account didn't have enough free CPU time to get the result. So in my experience, EA was wrong, and Wolfram was correct but not helpful.

    • by leaen (987954)

      Of course not. This is just as stupid as asking if you could calculate somebody's phone number.

      Actually calculating phone numbers is simple, you just need to start from contradiction. You can derive anything from that including your mom's phone number. Its called principle of explosion.

  • by bazmail (764941) on Sunday February 02, 2014 @11:40AM (#46134417)
    Try asking them that question at the start of the season and I'll be impressed if they get it right.
    • Try asking them that question at the start of the season and I'll be impressed if they get it right.

      Am I the only one who's less than impressed with Wolfram Alpha in general? Sometimes it seems like finding the syntax necessary for it to understand even a simple query is more work than figuring the problem out myself.

      • Looking at stats on how a team historically plays seems kind of useless to me to predict their next game. A football team is made of individuals who may or may not have been on the team when those stats are assembled. There's also a group dynamic. So in short; about as effective as historical predictions of stock performance without regard to what the company in question is doing and the factor that all sides are gaming the system.

        The fact that a game like Football can produce clear statistics, gives an ill

      • by EETech1 (1179269)

        The one time I slugged it out with Wolfram Alpha, I learned a lot about the question at hand just trying to figure out how to get it to calculate the right thing, and then it gave me the wrong answer!

        It needs a Watson front end.

        • by omnichad (1198475)

          and then it gave me the wrong answer!

          "I checked it very thoroughly," said the computer, "and that quite definitely is the answer. I think the problem, to be quite honest with you, is that you've never actually known what the question is."

          "But it was the Great Question! The Ultimate Question of Life, the Universe and Everything," howled Loonquawl.

          "Yes," said Deep Thought with the air of one who suffers fools gladly, "but what actually is it?" A slow stupefied silence crept over the men as they stared at the computer and then at each other.

          "Wel

  • by TheNastyInThePasty (2382648) on Sunday February 02, 2014 @11:43AM (#46134429)

    This is all information that most people deeply interested in the statistics of this match up know already. Wolfram Alpha doesn't really bring anything new to the table.

  • by prefec2 (875483)

    It can calculate probabilities on the basis of collected data. However, it cannot predict the future. Based on article, it had a success rate of 80% (failure rate of 20%). However, a sample with 10 measurements hardly qualifies as big enough to determine if the estimation model is good enough to bet on it.

    • It can calculate probabilities on the basis of collected data. However, it cannot predict the future.

      Isn't "predicting the future" exactly calculating probabilities of future events on the basis of collected data? What else do we mean when we say "the National Weather Service is predicting 1-3 inches of snow in the Baltimore region"?

  • by Anonymous Coward

    This is about as stupid as asking Bob, the trained seal, to throw a ball through a hoop with the team's logo on it. There's so many dumb random predictions out there that some of them are going to have very good records.

    But the computer is actually going to be worse than a blind guess. Why are you analyzing all the games ever played by the Seahawks and Broncos? 10 years ago, not a single one of these players or coaches were on the team, making their results irrelevant.

  • Stupid. (Score:4, Informative)

    by brunes69 (86786) <slashdot@NospAm.keirstead.org> on Sunday February 02, 2014 @11:58AM (#46134513) Homepage

    There are hundreds of millions, nay billions, of dollars at stake every year gambling on the superbowl. Bookmakers in vegas spend literally millions of dollars computing the odds to a much deeper degree than this foolishness in the summary, and even they are not even close to 100% accurate.

    If the bookmakers in Vegas can not guarantee their predictions, neither can Wolfram Alpha.

    • by rvw (755107)

      There are hundreds of millions, nay billions, of dollars at stake every year gambling on the superbowl. Bookmakers in vegas spend literally millions of dollars computing the odds to a much deeper degree than this foolishness in the summary, and even they are not even close to 100% accurate.

      If the bookmakers in Vegas can not guarantee their predictions, neither can Wolfram Alpha.

      Sadly for you, you cannot guarantee that either!

      • by Maritz (1829006)
        It seems perfectly reasonable to guarantee that no-one can guarantee their prediction. It's about the only thing you can guarantee I suppose.
    • Re:Stupid. (Score:5, Informative)

      by Anonymous Coward on Sunday February 02, 2014 @12:40PM (#46134711)

      No, bookmakers don't compute odds. They compute (and recompute) a number that will put 50% of the betting population on each side of the line. It has nothing to do with who is going to win.

      • by Boronx (228853)

        Mod parent up.

      • by psymastr (684406)

        No, bookmakers don't compute odds. They compute (and recompute) a number that will put 50% of the betting population on each side of the line. It has nothing to do with who is going to win.

        That bookmakers try to split the betting money 50-50 might be true when the betting is against the spread. But not all betting is against the spread. It is also possible, for example, to bet on the "moneyline", i.e. on who is going to win the match. It is easy to see then that moneyline betting pays different odds for each team in match. In such cases the bookmaker, if he wants to guarantee a profit, has to split the money in a different way, for example if a team has 90% chance to win the match, he needs t

      • by fatphil (181876)
        Total bollocks. Do the maths before posting nonsense.

        They want to maximise their minimum expected gain from either outcome. In a simple 2-choice scenario the expected gain is a fairly simple curved surface, and that minimum will be where the expected gain from both outcomes is the same. (And were it not for various frictions, and lack of perfect information, that gain would be precisely zero.)

        So yes, they're looking for balance, but that balance doesn't necessarily occur at the 50/50 point you claim. (Which
      • by dkf (304284)

        No, bookmakers don't compute odds. They compute (and recompute) a number that will put 50% of the betting population on each side of the line. It has nothing to do with who is going to win.

        Actually, they compute a number that balances obligations to pay out against the sums laid. They're in it for profit, not for a bit of fun, and they don't mind if it is profit from many little bets or a few big whales.

    • by Rockoon (1252108)

      Bookmakers in vegas spend literally millions of dollars computing the odds to a much deeper degree than this foolishness in the summary, and even they are not even close to 100% accurate.

      Sigh....

      If you don't know how bookmaking works, don't comment on it. We both know that you don't know, and we also both know that you are making an uneducated guess that bookies care about a games outcome. You are wrong. Bookies do not care who wins. Bookies make money regardless of the outcome.

      The goal of a football bookie is to get equal amounts of money wagered on each of the two teams. Now stop acting like a fucking expert when you are actually an ignorant fuck.

      • Re: (Score:3, Insightful)

        by brunes69 (86786)

        Of course bookmakers make money on both sides, but they also rely on odds. HEAVILY. If the odds are not set properly they stand to lose a fortune. If all it was was a 50/50 crapshoot then odds on a sportsbook would not even exist. Get a clue.

        • by Imrik (148191)

          The odds shift as more bets get placed on one side or the other.

        • Re:Stupid. (Score:4, Informative)

          by Chris Mattern (191822) on Sunday February 02, 2014 @02:49PM (#46135381)

          Yes, but bookmakers don't set their odds based on an analysis of the event. They set their odds based on the bets they receive. That's *why* they are guaranteed their cut regardless of who wins.

          • Re: (Score:2, Interesting)

            by Anonymous Coward

            Yes, but bookmakers don't set their odds based on an analysis of the event. They set their odds based on the bets they receive. That's *why* they are guaranteed their cut regardless of who wins.

            I work as a risk manager in a major company in the industry, and might have some insight here.. Lots of faulty assumptions in the comments in this thread, a lot of them upvoted for no apparent reason.
            The bookmakers start out with odds which are very much based on analyses of previous events. If you do not set odds based on an analysis of an event, you stand to lose a lot of money early on.
            These odds will then get adjusted, but not as most of you think.
            They will not really get adjusted to keep

      • True, the casinos want equal betting on both sides. Therefore, the casinos analyze the odds of who will win only when they open the betting. GP is correct that those predictions are often quite wrong. Additionally, Vegas is full of other professionals who do make predictions of who will win, and most importantly, by how much they'll win. People into sports betting read the expert predictions all the time. These professionals have a pretty poor track record.

        All of which means YOU are the ignorant fuck and

        • by Rockoon (1252108)

          True, the casinos want equal betting on both sides. Therefore, the casinos analyze the odds of who will win only when they open the betting.

          You are an idiot. You do understand that the second sentence is not a conclusion that can or even should be drawn from the first, right?

          The casinos NEVER analyze the odds of who will win. They analyze public perception. The outcome of the event is irrelevant. Public perception is relevant.

          Additionally, Vegas is full of other professionals who do make predictions of who will win, and most importantly, by how much they'll win.

          ..and they arent bookies, and they dont (ooh let me quote the ignorant fuck you are defending with your own ignorant fuckness) "spend literally millions of dollars computing the odds to a much deeper degree than this f

  • by Anonymous Coward

    This is just as idiotic as the legions of sports 'journalists' who somehow believe that a game between two teams decades ago... with different players, different coaches, different gameplans, different weather, and DIFFERENT league rules has any bearing whatsoever on a game played today. Yet another heinous abuse of statistics by the mathematically illiterate.

    And if you haven't read the article... this is not Wolfram's abuse of statistics... this is Nick Kolakowski. He even admits "Just remember that, as

  • by Anonymous Coward on Sunday February 02, 2014 @12:27PM (#46134645)

    Developed by Stephen Wolfram

    Wolfram alpha wasn't developed by Stephen Wolfram any more than Windows 95 was developed by Bill Gates. They key difference is that one of them has an enormous fortune while the other one has an enormous ego.

  • by Anonymous Coward

    It looks to me like this is a Bayesian inference problem (http://en.wikipedia.org/wiki/Bayesian_inference). The question is this: What is the probability using the Madden game one can predict correctly the winner of the SuperBowl? The prior probability on the correct probabiity p is uniform,.i.e, it's just as likely to be 80% accurate as 11% as 55% etc. -- one can't tell without data. After 10 games we have 8 successes and 2 failures. This is a job for the Beta distribution (http://en.wikipedia.org/wiki/Be

  • But while Wolfram Alpha can crunch all the historical data it wants, and that data can suggest one team will likely triumph over another

    I could've told you that without any data.

  • but only 50% of the time. ;)

  • Also, given how player and coaching rosters vary from year to year, the teams taking the field can change radically between meetings.

    This is one of the most important points here. Maybe the '78 Broncos beat the Seahawks, but that has absolutely no bearing on a game almost 4 decades later.

    • This is a point that has been made several times in the thread, and I'm not sure I fully agree with it. Although the personnel at a team may have changed several times, they don't all change at once, and the team (especially in the wider sense of backroom staff and fans) retains a distinct identity over long periods. Significant events in the past, such as strings of defeats against another team, winning against the odds in key games and especially losing a key game that you looked certain to win, remain em

  • I'm happy to see that many /.'ers are criticizing TFA & Wolfram Alpha

    Here's my $0.02...

    Their data set is fucking ridiculous...they hooked in **HISTORICAL** stats for the teams for several decades.

    There is absolutely zero relationship between how the '82 Broncos played to today's game.

    The fact that these researchers used this as their data set is laughable. We should all join in mocking this bullshit study.

    • by Firethorn (177587)

      Yeah, considering how I'd do it, I'd probably go with at most a 4 year examination - 40% for this year, 30-20-10 for the previous years. Short of building an actual game simulator like EA did, I'd probably look at median scoring, yards run/passed and such, then look at other team's averages, and whether other teams performed better/worse than their average, and by how much. For example, is team X better at blocking pass plays? Does team Y get most of it's yardage from passing, meaning it's at a disadvant

      • it's really about analyzing players not a team

        that's why EA's simulator can predict well...it's based on probabilities of events based on statistics from individual players over their career. you can plug in any players into the roster and their system does the math

        sure, a coach like Bill Walsh can make a system for a team that becomes so consistent that it *could* be analyzed over several years...but that level of consistency is very rare...the 49'ers west coast offense spanned two QB's, Montana and Young

  • (1) Story about who will win the Super Bowl between the Seattle Seahawks and the Denver Broncos
    (2) 65 comments
    (3) Zero mentions of "Seattle" or "Denver" in the comments.
    ...Priceless
  • The first game of the season the Denver Broncos beat the Ravens in a rematch of the playoff matchup that eliminated the Broncos from the post-season last year. It was the kind of record-setting (7 TDs) moral victory that set the tone for the entire Broncos' season.

    Random thing: The Superbowl Champs traditionally open the season at HOME. The only reason the Ravens were at Denver was a scheduling conflict with the Baltimore Orioles. The Ravens ended up missing the playoffs. If they had 1 more victory this

  • No mention of the Superb Owl watching over all this?

  • A seismograph?

  • The answer to the question is "No."

    The Seahawks embarrassed the living daylights out of the Broncos.

    • by Daetrin (576516)
      Betteridge's law of headlines [wikipedia.org] strikes again!

      I need to spend some time at lunch finding a screencap of Manning's face as the opening snap flies right past his head. It was priceless at the time, and only more so in retrospect as it set the tone for the entire game.
  • Much easier just to go with the ape: http://espn.go.com/nfl/playoff... [go.com]

Line Printer paper is strongest at the perforations.

Working...