Quantcast

old-computerThe volume of information and data available to professional baseball organizations these days is utterly enormous. It’s not just statistics – it’s scouting information, health information, PitchF/X data, fielding information, batted ball data, spray charts … I could go on. And when you layer the data on top of itself to find patterns and useful information, the crunching power needed can be significant.

Well, one unidentified team is taking on that data glut in a new way: by buying a Cray supercomputer. That’s actually the name of the supercomputer, by the way, not, like, cray cray. Although, from the sound of things, the computer is pretty cray cray. From The Economist:

[O]ne MLB team has invested in a Cray supercomputer according to Pete Ungaro, the company’s chief executive officer. The team, which declines to be named, exemplifies an organisation that, five years ago, most people would not have dreamed would need, or even want, a supercomputer, he says.

The team obtained one both because the machine has the capacity to analyse enormous quantities of data and because of the short time in which it can process them. Other technologies, such as cloud computing, could wade leisurely through information, helping managers make choices during the off-season (perhaps concerning which players to add to the roster, for example). Instead, a team can use a supercomputer to process data in time to affect decisions during play, explains Mr Ungaro. Cray’s Urika appliance, launched two years ago, is specifically designed to help users interpret data in unusual ways.

Business Insider reports that a Cray supercomputer starts at $500,000, so the investment here was significant.

As for the team, the one clue is that it is an organization that “five years ago, most people would not have dreamed would need, or even want, a supercomputer.” I’m not saying it is the Cubs, but that sure sounds like the Cubs, yes? And you throw in the fact that this is the kind of thing the Cubs are currently spending their discretionary funds on, and it does make you wonder.

Maybe having an in-house supercomputer for use in-game is the next market inefficiency.

Sabermetrician Tom Tango, a consultant for the Cubs, writes about the potential baseball uses for a supercomputer, and it’s pretty damn interesting to think about.

  • Jim

    Could this go with the job posting you had a few days ago?

    • http://www.bleachernation.com Brett

      I would think only in the sense that they are continuously expanding their use of analytics and technology. Gotta have people in place to do the people stuff.

      Very interesting point, though.

    • MattyNomad

      I do see the connection with the Cubs FO job and this news of a club buying some massive super computer…but I think this has Yankees [maybe Phillies] written all over it. Considering the fact that they both played in the 09 Series, I wouldnt imagine they would really look to change anything either. Neither have really been a sabermetrics kind of club, especially the Phillies. But it seems like the Yankee way….BUY yourself the best sabermetrics can offer! In turn, it could also be some of the Phillies TV money coming out to play.

      From my perspective, admitting to buying one of these is like admitting your looking to change or improve on your current baseball operations. Of which, both the Yankees and Phillies are desperately denying the need for a tear down rebuild. I’d put my money on either one of those clubs.

  • ssckelley

    It obviously is not in Wrigley Field. Where the heck would you put such a thing?

    • http://www.teamfums.org MichiganGoat

      Maybe a rooftop ;)

      • MightyBear

        LOL You kill me young man.

      • Picklenose

        But how would the computer see what is going on? The beer signs would block its view.

    • D.G.Lang

      In today’s world a supercomputer itself would not require much space. The biggest space occupying items would be the I/O devices such as the printers, tape drives (few) and hard drives requiresd to store the data.

      The processor itself would not necessarily have to be very large.

      Background: I have worked with very large mainframe systems since 1967 serving at major corporations, International Harvester, CNA Insurance, AT&T, GTE, Centel, and Sprint among others. I have owned my own consulting company and have held such titles as Programmer, Analyst, System Specialist, and Systems Engineer.

      The earliest computers such as the Vacuum tube IBM 705 were truly massive but with the introduction of the IMB 1401, and 1410 transistorized computers processor size started declining rapidly. The IBM System 360 series from the mid 60′s again continued the smaller more powerful processor trend and in the mid to late 90′s immensely powerful processors could fit in a fairly small sized cabinet.

      A hard drive array of 7 disks would require a bedroom sized room back in the 60′s but today would only require a single small cabinet.

      My own personal laptop computer contains an 8 core processor running at 4.2 ghz with over 15 terabytes of storage space and 16 gigabytes of memory takes very little space and easily outperforms most if not all of the supercomputers of previous years.

      The biggest cost factor of operating a supercomputer is not the purchase price but the support required to keep it running along with the software licensing costs and programmers salaries. $500,000.00 to buy a supercomputer is extremely and over the years many more times that amount will be spend in salaries to keep it running.

      • D.G.Lang

        Further info. Today’s PC are capable of much more than most people realize. There actually are supercomputers which are built using Intel PC microprocessors. Another interesting twist is that today’s high end graphics cards them selves contain multiple hundreds if not thousands of numeric co-processors each of which is capable of incredibly fast number crunching.

        My current ASUS Geforce 760 OC video card itself contains 1152 such processing units. Some machines are built using multiple video adaptors just to do bitcoin mining which is extremely processor intensive.

        Some of the AMD RADEON Video adaptors have over 2000 processing cores.

        Super computers being built based on PC microprocessors rely heavily on the video cards to do most of the number crunching.

  • Spoda17

    I don’t want to burst anyone’s [Cubs] bubble… but I would not classify this as a “supercomputer” … it may be a supercomputing program… but not a “computer.” This is a great thing, and I like to see the Cubs utilize software to make decisions.

    This really is not any different than a company using PeopleSoft for their HR program, or SAP, or an MRP program. These programs store a ton of data, report, analyze, cost projections, pay people… etc. But good companies invest in this type of advanced programs versus trying to keep an Access database or a spreadsheet.

    • http://www.bleachernation.com Luke

      The article pretty specifically references a team, hopefully the Cubs, buying the actual physical hardware, not just a software package.

      • http://www.friendly-confines.com hansman

        Looking up Cray, it certainly seems to be an actual computer.

      • Spoda17

        Yeah I’m with ya Luke… but I think the term supercomputer is an embellishment bay the company. Over the counter computers are pretty snazzy these days…

        • http://www.bleachernation.com Brett

          Computers at Best Buy don’t cost $500,000 to $60,000,000. :)

        • InRizzoWeTrust

          It is a legit computer. It has unworldly amounts of horsepower to rattle off complex calculations in short periods of time. No bells and whistles, just horses to analyze and calculate data

          • Edwin

            Yeah, but can it run spider solitaire?

            • mjhurdle

              SkiFree!!!

              • http://www.friendly-confines.com hansman

                Mmm…just imagine how good Kerbal Space Program would look running on this thing.

          • D-Rock

            But it still won’t be fast enough to analyze Baez’s super-swing!

        • http://www.bleachernation.com Luke

          When we’re talking Cray, it isn’t an embellishment. Cray is still making real, genuine, massive, complex, specialized, extremely expensive true honest-to-goodness supercomputers.

          Now if the team in question had bought one from Dell, I could see your concern.

          • Edwin

            Their “build your own supercomputer” has gotten better.

    • http://www.bleachernation.com Brett

      No, that’s why this is a story – the team purchased the actual hardware. A Cray supercomputer.

      • Spoda17

        Hahah… I know it’s an actual computer, and I’m sure it is real nice. But to allude you need “special” equipment to do extremely complicated analytics is not really the case.

        • http://www.bleachernation.com Brett

          It is if you need to do it almost instantaneously using insane volumes of data. That’s why these computers exist in the first place.

          • aaronb

            I can see it being helpful with things like defensive shifts or batter hot zones.

            Other than maybe that it just seems like a toy. Albeit a toy that can be used as an excuse as to why payroll dropped yet again this year.

            • http://www.friendly-confines.com hansman

              Just determined to crap all over this post. Aren’t you.

            • mjhurdle

              as someone that works in the IT field, I crack up every time I see the “I don’t understand this technology therefore it has no/little value” attitude.
              priceless.

              • DocPeterWimsey

                Scientists get the same thing: “I am too dumb and/or ignorant to comprehend what you are saying” is considered a legitimate basis for an “opinion.”

                My guess is that someone wants to do a lot of parallel processing. Baseball would be a good case for using a set of techniques called Markov Chain Monte Carlo algorithms, in which you estimate the probability of outcomes under a huge numbers of combinations of different parameters. The analyses “evolve” the hypotheses by randomly changing some of the parameters (the Markov Chain part) and the keep/reject the new idea with some probability based in how much improvement the new combination offers: if it is much better, then you almost always keep it, and if it is worse, then you usually reject it.

                Because hypothesis space usually is huge, dark and full of terror, it is great to run multiple chains: and that is where parallel processing comes in handy. Usually you have “hot” and “cold” chains, which makes it easier to find very different hypotheses that make similar predictions when intermediate ones make much different predictions. (See “on skinning cats.”). Supercomputers are great for that.

                The other thing that super-computers aid is looking at the outcomes of classes of hypotheses. Say that you have multiple sets of hypotheses that you would call “small ball” and other sets that you would call “Weaver ball.” Supercomputers can set aside much more memory still to cluster and sum across specific versions of general hypotheses.

                Also, they look really cool and can be programmed to go “bing.”

                • MightyBear

                  We paid a lot of money for the machine that goes “bing”.

                  • MightyBear
                  • DocPeterWimsey

                    heh, glad someone caught the reference.

                    Also, that’s a great example of “sorting” vs. “selection” because, yes, it’s true: someone paid a lot of money for a machine that goes bing; however, nobody paid a lot of money for a machine because it can go bing.

                    (Now, that it probably plays audiophile analogs of mp3 files might be a different issue….)

        • frank

          It’s real nice alright–that’s CIA/NSA level equipment.

  • http://www.friendly-confines.com hansman

    Bah…what does a supercomputer know?! Has it ever played the game? Can it look at a kid and say “He just LOOKS like a ballplayer, ya know?”

    The neat thing about having a supercomputer for in-game use is that you could pull in up-to-the-second weather information and tweak defensive positioning, pitcher strategy, etc. See if a guy is having issues with a certain type of pitch on that day, determine pitcher tiredness using pitch speed/break.

    The possibilities are endlessly endless!

    • http://www.teamfums.org MichiganGoat

      A computer can never be scrappy and if it has BellyFire then you have problems.

    • Boogens

      Can this so-called supercomputer measure a player’s confidence?

      (Sorry – I just had to go there)

  • http://bleachernation.com woody

    When they get a robot working with that computer I would like a cu of coffee.

  • aaronb

    Any truth to the rumors that Cubs finished 2nd in the Super-computer bidding?

    • http://www.bleachernation.com Brett

      See? Now that’s actually a little funny, and you got to do your negativity thing.

      That’s all you have to do, and everyone is happy.

      • aaronb

        We aim to please Sir!

  • http://www.friendly-confines.com hansman

    Ahhh, I figured it out. The Cubs bought a supercomputer so they can acurately model what happens when Baez hits a baseball.

    [img]http://upload.wikimedia.org/wikipedia/commons/thumb/1/1c/CMS_Higgs-event.jpg/260px-CMS_Higgs-event.jpg[/img]

  • NorthSideIrish

    How does the supercomputer fit in it’s mom’s basement?

  • candyland07

    Spending a half million dollars in discretionary funds is as dumb as having a half a million dollars in discretionary funds but it is a nice to have wad of cash to blow on discretionary objects that might net the worst performance in Cub history.

    I thought the Cubs invested in Bloomberg – they have had the worst 2 year in cub history, so a super cray cray computer cant get any worse .

    • http://www.friendly-confines.com hansman

      Wow…just, wow.

  • InRizzoWeTrust

    Cray supercomputers are made about 10 minutes from me up in Sconnie-land. Its a legit computer (not software) which is capable of high calculations of data in fractions of time as opposed to normal computers we use. Think of it as a computer on steroids( appropriate?) And are built much like those IBM supercomputers that play chess flawlessly.

    • ssckelley

      Go rummage through their trash to find a shipping receipt that has an MLB teams address on it.

      • http://www.bleachernation.com Brett

        I lol’d.

  • MaxM1908

    I do not think it’s a coincidence that the author of the Economist article specifically references the Cubs at the end. Granted, he was drawing an analogy between the Cubs’ sole bright spot in a long history of disappointment with the prospects of the Supercomputing industry, but perhaps it’s also a nod to some inside information he got from the company.

    This has Cubs written all over it.

  • Norm

    bah, the only expenditures that occur are for major league ballplayers, all other money goes straight into an owners pockets.

  • miggy80

    This goes to support my Theory that baseball is for smart people.

  • grip

    Off the topic but, at some point I’d be great to see a sidebar or glossary with definitions of some of the newer performance metrics discussed in these posts.

    • http://www.friendly-confines.com hansman

      until Brett does a series (or gets Doc to do one for him), fangraphs.com/library

    • http://www.bleachernation.com Luke

      I still have a series of that nature on the drawing board (and have incorporated some of it into some of my prospect pieces scattered here and there), but nothing coordinated has yet emerged.

  • jp3

    I heard we came in 2nd place on the bidding for the supercomputer…

    • miggy80

      and after Theo talked all that crap about “not being out bid” and how “we’re going to use the Tanaka money”. Save it for Ricketts Theo

    • Edwin

      I think Aaron beat you to that one, but I still gave a small laugh.

    • jp3

      Damnit Aaron I didn’t see your comment but great minds and whatnot

      • aaronb

        Yes they do

  • Jon

    “pshhhh, my Mac can do everything this supercomputer can do”

    /AppleSnob

  • Edwin

    Just the Cubs luck, they’ll end up buying the super computer from 2001 Space Odyssey.

    “I’d like to trade Darwin Barney for Mike Trout”

    “I can’t let you do that, Theo”

  • ssckelley

    These computers fill up rooms….not your typical desktop or laptop.

    [img]http://www.cray.com/Assets/Images/products/xc30-5.jpg[/img]

    • http://www.friendly-confines.com hansman

      If the one on the left is a standalone, I’d guess that it’d be a little bigger than a vending machine.

      For a baseball team, I’m not sure you’d need a whole row of them.

      • ssckelley

        I find it kinda funny a baseball team needs one of these at all. I doubt sites like fangraphs uses a Cray to compute it’s statistical data.

        • http://www.bleachernation.com Brett

          They don’t – but they don’t need to do it with ALL of their data in a matter of seconds. Some of the stats websites out there will run analysis that takes hours or even days (literally) to complete.

          • ssckelley

            I am trying to think of logical reasons why teams would need this data so quickly.

            • http://www.friendly-confines.com hansman

              Frees up time. You may be able to use this computer to analyze all of the video in MLB in a matter of hours and be able to accurately scout an opposing team or a trade candidate. Even beyond that, the in-game application could be very valuable.

              For 15-20 of the teams, it isn’t about getting an extra 2-3 wins advantage over the other teams, it’s about getting an extra .1 win against the other teams. For the others, they need those 2-3 wins just to get on the same level as the rest of the league.

            • Boogens

              “I am trying to think of logical reasons why teams would need this data so quickly.”

              So much porn, so little time. ;-)

        • Norm

          I think the MLBAM Fielding tracking system is going to create a whole new level of need. (x,y coordinates, acceleration speed, first step quickness, distance covered, ball trajectory, MPH off bat, hang time, etc, etc)

        • http://www.friendly-confines.com hansman

          I think this is going to go way beyond just addition and subtraction. They’ll be looking to simulate games and seasons to determine FA value, trades, advanced scouting, incorporation of the new fielding f/x data, pitch f/x. I’m guessing whatever team ponied up for this is going to be spending a bunch of money getting that technology into their minor league stadiums.

          This should allow them to create a very detailed profile on every hitter in the big leagues, compare them and dive into game situations. So it’s not just Rizzo has done X against power pitchers. It will be, given today’s weather conditions, how the pitcher is performing, today, compared to all pitchers who have performed in a similar way given similar weather conditions against all the batters that are similar to Rizzo and, by the way, the computer just ran this a 1000 times so we should expect this outcome. PH him for Koyie Hill because I ran that simulation (along with all other players on the bench) and the run expectancy is .00001 higher.

          Oh, and the opposing manager is likely to sub the current pitcher for pitcher C based on that manager’s tendancies and his bullpen’s recent usage. In that event, sub Rizzo for Campana because I did all that above and the run expectancy is .0001 higher in that event.

          Or, the computer will suggest signing player X over Player Y because, based on historical data, Player X will provide a better value.

          Or, the data on prospect X suggests Y based on a simulation of 1000 of his careers.

          All to get *maybe* 1 more win per season.

          • candyland07

            Wow- 1 more win per season under the correct situations because the variables are just so damn unpredictable- i hope this is an early April fool teaser and anybody that would be proud of this machine – should never buy first generations of anything – too many bugs it just so cray cray—.

            • Norm

              $500k for 1 win is quite a bargain.

            • http://www.friendly-confines.com hansman

              Do you realize that supercomputers have been around for decades. This is just the first time in a baseball context.

              I’m guessing not.

  • itzscott

    This is actually very interesting whether it’s the Cubs or not….

    Because of the speed of a Cray and the fact that the results can be relayed in-game, pitch by pitch…. this would be like an offensive or defensive coordinator in football calling the plays from above and relaying it to the playing field below.

    In baseball it could completely take in-game decisions, pitches to be thrown, along with batting orders, away from managers. It could clue in a batter what pitch the opposing pitcher is very likely to throw.

    • Edwin

      I don’t think you could can use technology in games to that extent.

      • itzscott

        It’s all about complex probabilities and how quickly situations can be reacted to.

        Whichever team is getting one is looking for an edge over other teams much like when the A’s started using Sabermetrics before any other team did.

        • Edwin

          What I mean is I’m not sure where MLB draws the line with what technology can be used by a manager during games.

    • ssckelley

      hmmm, maybe that is why they got rid of Sveum.

      “Um Dale, we are replacing you with a computer that can make in game decisions. The computer down the hall, our HR rep, has your severance pay.”

  • fortyonenorth

    I think Crays were all the rage about 20-25 years ago when desktop computing was in its infancy. I find it really hard to believe that crunching baseball numbers–regardless of how “advanced” the stats are–requires this kind of heavy lifting. Not saying the story is bunk, just that buying a $500,000 computer is probably unnecessary to achieve the desired outcome.

    • http://www.bleachernation.com Luke

      It is necessary. If you want to work with a huge number of variables in real time, you need some extremely serious computing muscle.

      • fortyonenorth

        I’m not saying that “serious computing muscle” is unnecessary, just that you don’t have to spend $500,000 + to get that. At some point, whether you get a return on your query in 1/10 of a second or 10 seconds doesn’t matter.

        • http://www.friendly-confines.com hansman

          I think some of it that the amount of information in baseball is about to explode over the rest of the decade and that’s just on the MLB side. This team might also be eyeing Pitch f/x and Field f/x systems in their minor league stadiums which would greatly expand the need for computing muscle.

          It’s like telling a 1980 team that they don’t need a computer. Just get a calculator and a new box of pencils!

  • candyland07

    What did the Cubs pay for the Bloomberg partnership? maybe 500,000 is cheap

    • http://www.friendly-confines.com hansman

      I may be wrong, but Bloomburg is on the software side. In theory, they’d be the ones to develop the software so the computer can do the work.

  • David

    Brett – would they even be able to use a computer during games? I’m pretty sure they aren’t even allowed to have iPads in the dugout and have to have those giant binders full of printouts.

    That said, this would make sense given what an insane amount of data systems like FieldFX are going to produce.

    • FFP

      “We had a contest to name it,” Epstein said. “I nominated Carmine because it would be a play on Carmine Hose (an ancient Red Sox nickname) and it sounded like a tough guy and a hot woman, just what we want from our software. Carmine won not only because it was the best nomination, but also because I was the judge.”There must be some rules for this, I agree. But the ways around those rules are already there. Players go into a video room just beyond the dugout to watch previous at bats from that game. You can’t review the video in the dugout, though, apparently. But, it’s closer than the can in most parks, I think. And then there’s a phone.
      In game suggestions to the field manager are the way its been done in Boston for a decade. See “Carmine”

      I have been waiting for the new computer in Chicago to get a name. (Theo picked ‘Carmine’ for Boston’s be

      • David

        Interesting.

        And I feel like an idiot for missing that because I actually just read that the other day.

      • FFP

        Sorry, I fat-fingered the paste button and the post button at the same time. (I think it’s these cashews.) Quote was from Francona’s book and should have been at end of post. The quote as posted should end after Theo saying he was the “judge” of who named her.

        Francona said he’d receive in-game suggestions, but only used them to inform his thought process. He claims he never was called on counter-manding Carmine or anyone else’s suggestions in a game. He was in fact the field manager. That was one of his strengths. Using both old school and new school stuff.

        • Picklenose

          If the Cubs really bought this computer they have to call it Robert – as in the Robert Cray Band.

  • ChrisFChi

    Maybe the Astros?

  • MightyBear

    I would love to work on that bad boy.

  • MightyBear

    They may also be looking at programming video movements from the ML and Minor league video feeds into the super computer to crank out data in real time. They did buy the minor league video equipment a few years ago.

    • Cizzle

      I noticed before one of the ST games that the Cubs’ IT guys were putting in camcorders (just normal Sony Handycams from what I could tell) in the Cubs dugout that were aimed at the 3rd/2nd base bags and the pitchers mound/1st base.

  • bushybrows74

    These computers are the norm for Commodity Trading Advisors aka managed futures funds. These firms were among the first adopters to use algorithms to make investment decisions. John Henry made more than a few dollars in this business. I think Theo worked for him at some point. Now only if someone knew how to turn that damn thing on.

  • JC Martin

    One the subject of computers and baseball…..something I have always wondered. If data and information is so important in the baseball decision making process…why have we never seen a manager with an electronic device in the dugout (either laptop or tablet)? It always looks outdated to see Lester Strode in the bullpen with his 3 ring binder

    • DocPeterWimsey

      I’ve often wondered that, myself. However, I’ve also wondered: is that even legal? (If I were a pitcher, then I’d be pulling out my iPhone to check the heat chart between each batter, and also to see where my pitches were going relative to where I wanted them to go on the prior batter. Because the game is too fast-paced, you know.)

  • GoCubsGo

    Ricketts wants some of that sweet BitCoin money.

  • Rich H

    If any major league team takes this type of step with real time analysis it gives them a huge advantage.

    Think of it like this. A kid gets called up to the bigs to face your team for the first time. You have all the scouting reports on him only if he was on your radar as a trade target or you get lucky with one of your scouts remembering what he sees when he is there is study someone else. Now with this amount of of computing power you literally can take every note ever made about the kid, condense it to a readable form with real time analysis of what his tendencies were when he was seen and have a small book on him for your big league coaches by the time he is announced as a start. Your players can literally have a book with pictures, stats and crunched numbers of how he did in every situation imaginable before he ever throws a pitch in the majors. Even if he literally just got off the plane that afternoon.

  • Darth Ivy
    • jp3

      Ok that’s funny DI

Bleacher Nation Privacy Policy and Terms of Use. Bleacher Nation is a private media site, and it is not affiliated in any way with Major League Baseball or the Chicago Cubs. Neither MLB nor the Chicago Cubs have endorsed, supported, directed, or participated in the creation of the content at this site, or in the creation of the site itself. It's just a media site that happens to cover the Chicago Cubs.

Bleacher Nation is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.com.

Google+