Inside Major League Baseball’s “Hypothesis Machine”

By Nicole Hemsoth

April 3, 2014

When it comes to sports statistics, there’s no richer source of historical data than baseball. With over 140 years of detailed information on individual players, teams, and winning trends, the addition of digital data is powering even deeper analytical capability to help fans and team owners make decisions.

Baseball data, over 95% of which has been created over the last five years, will continue to mount—leading MLB decision-makers to invest in more powerful analytics tools. While there are plenty of business intelligence and database options, teams are now looking to supercomputing—or at least, the spawn of HPC—to help them gain the competitive edge.

Last June, we talked with Cray CEO, Peter Ungaro, who dropped a hint that MLB was their big Urika graph analytical appliance user, but waited on the sidelines for more details, including who the team might be. While the team is still a secret (understandably—this could be the skeleton key to a winning season) we were able to get some details about YarcData’s role in MLB from Cray’s Tim White, who manages the government and intelligence business unit and has been the point man for the mystery team’s walk down graph analytics lane.

White, who left General Dynamics’ Advanced Information Systems division after 8 years to come to Cray, is no stranger to advanced analytics. At General Dynamics, he ran national security, law enforcement and intelligence projects on site, where he was able to understand firsthand the challenges and benefits of real-time information gathering. He said when Cray called him to push into markets with some of the toughest graph analytics problems, he immediately saw the opportunity. Government use of complex graphs to mine for relationships between disparate datasets is expected—but the same value there in understanding how multiple variables translate into unseen ties applies to baseball (and other areas, including fraud detection, drug discovery and more) just as well.

He explained that what teams, just like governments and drug development researchers, are looking for is a “hypothesis machine” that will allow them to integrate multiple, deep data wells and pose several questions against the same data. They are looking for platforms that allow users to look at facets of a given dataset, adding new cuts to see how certain conditions affect the reflection of a hypothesized reality. While he says there are multiple tools for doing this on both the hardware and software front, there is a particular marriage of both elements that works particularly well for graphs. This combination includes a large shared memory machine that lets users cut around unpartitionable problems. At the software core are SPARQL and RDF, which he agrees are specific tools but are the only ones that can tackle large-scale complex graph problems that are otherwise difficult to untangle without the blend they’ve packaged into a Urika box.

urikaTo put the complexity of the graphs MLB is working with in context, White told us that they’re operating in the 10-20 billion edge graph range. That’s no problem for Urika, however, says White, at least compared to trying to manage that kind of large-scale problem using basic joins. For background, the appliance is graph-specific and tailored to RDF, which can be an unwieldy data structure unless it’s humming on a big shared memory system. Urika uses the thread-hungry Cray native XMT processor and while it has that capability, it’s not designed for running floating point vector calculations. The value here is dipping into deep memory to hunt for similarities and associations that expose buried relationships between factors.

“Urika is unique in that it’s a global shared memory machine that lets you look at data in an unpartitioned fashion. This is very critical if you’re looking at graphs, which by nature are unpredictable. Further, certain graphs are non-partitionable—and if you do partition it, it changes the result of a query,” White explained. “There is no MapReduce job or partitioning that will do anything but fracture the graph to a point where it’s no longer reconstructable—and even for those you can reconstruct, it would take a lot of compute power.” Where this works is with memory-bound problems versus those that are compute-bound, in other words.

Urika is also fitting for the big data of MLB given the disparate data sets required to piece together a best-case-scenario for team leaders. There are lots of sources and combining that data requires a data structure that allows for federated queries. This is exactly the reason big pharma and a few others find RDF machines useful (in the case using SPARQL queries). “You could go ahead an do the equivalent of a hundred-way join from a relational database—the question is, how big of a dataset can you do that against?” asked White. “Unless you have something like Urika, which has the ability to do it memory and with massive multi-threading, you’re not able to look at enough data.” He said that when compared to what they’re doing inside Urika, for normal relational databases, this would be the equivalent of a 30-50 way join. Pulling from the large shared memory pool using SPARQL queries offers a more seamless blending of conditions to hypothesize against. And herein lies the selling point for operational budget-constrained MLB.

Although we’re not privy to pricing, the Urika appliance runs in the order of well over a million bucks. However, to put this into some context, consider the ROI. The average win in baseball brings a smaller team a couple of million, a large MLB team between $5 and $7 million. That’s one win. Let’s say Cray’s graph appliance is able to help team owners piece together what happens if you take the average player for  a particular position versus the player they’re analyzing across the course of a season. While those many factors involved are multifaceted, when it comes down to one small decision at the bottom of the ninth with loaded bases, that one switch in decision-making could mean the loss or gain of millions in a winning situation.

So this all begs the question, why buy an expensive near real-time analytical HPC platform if the time itself isn’t necessarily an issue? Couldn’t there be cheaper ways to run this on a SAS or SAP HANA box after the fact? Or for that matter, a shared memory cluster of one’s own devising primed to run the latest, greatest analytics software?

White responded by pointing to the nature of their approach to graph analytics, agreeing that real time here doesn’t have anything to do with immediate in-game decisions. Rather, their queries can be submitted, run very quickly for fast analysis, then analysts can run that same query again in a hurry with different variables. This effectively allows them to tweak the question with new conditions. This “hypothesis machine” approach is what makes the difference, says White, pointing to the burdened systems MLB and other organizations have that can’t be used to tailor one query after the next in rapid succession to find answers and hidden connection between factors they didn’t even know to ask.

The variables for each iterative query range from the outlandish to the expected (RBI, homerun histories, Golden Glove data) and even to the subjective (sentiment of fans, coach and player confidence, salaries, TV revenue, etc.). All of these are collectively referred to as “field effects” which is a certain science in itself.

White agrees that there are many excellent analytical software and hardware platforms out there to choose from, but most of those tend to focus on data that is known. Such platforms consider data from a range of databases that are all finding answers, but don’t go far enough into the masked relationships that baseball (and for that matter government intelligence, pharma and others) require to make real progress.

“We’re past the point of thinking software can solve all the problems,” said White. The hardware matters—in fact, it’s the reason why big organizations with large-scale data-driven challenges are looking seriously at HPC platforms. They’ve managed to eek all the performance out of their code, custom or otherwise, and now see the need for a marriage between the metal and the mind—in MLB, big government and beyond.

“These users want to ask a lot of questions. They want to get those results right away, then change the question. So they go through 20 or 30 different facets of the same question to find what’s interesting. It’s the iterative quality of Urika that’s interesting, it becomes their hypothesis machine—they go through all the hypothetical queries to find the difference maker,” says White.

So we’ve worked out the why on the graph analytics investment side, but with so much of this still subjective science, how on earth did Cray manage to talk an MLB team (these are historically cash-strapped organizations on the operational side) into this? To what extent are team owners concerned with data partitions and multi-threading?

Not at all, says White. However, they have been doing analytics that are much closer to Excel operations than anything bordering supercomputing. In fact, when asked about what Urika replaced at MLB shops, White hinted that it didn’t go too far beyond Excel or general business intelligence tools. This, however, he says, will be a gamechanger. Literally.

To be clear, this machine isn’t powering immediate gameplay. MLB rules prohibit the use of electronic devices during game time. When news first leaked that a mystery team was using the graph appliance there was some confusion about “real time game play data” that could shape the decision-making of coaches. This isn’t the case. Rather, the analytics are run after the fact (and before new games, seasons, or investments in player contracts). These are data for the playbook—the one analog “device” that’s allowed. And the output of the graph should lead to an approximate answer to, “when the ball hits at this level under these conditions based on history and contemporary circumstances X is the best option.”

It would be a thrill to compare playbooks before and after advanced analytics have been run but, sadly, we’d need to know the team who bought the Urika. There have been a lot of guesses, perhaps we should look to the downtrodden MLB club that suddenly pulls some amazing wins out of thin air this year for clues.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industy updates delivered to you every week!

Rethinking HPC Platforms for ‘Second Gen’ Applications

February 22, 2017

Just what constitutes HPC and how best to support it is a keen topic currently. Read more…

By John Russell

HPC Technique Propels Deep Learning at Scale

February 21, 2017

Researchers from Baidu’s Silicon Valley AI Lab (SVAIL) have adapted a well-known HPC communication technique to boost the speed and scale of their neural network training and now they are sharing their implementation with the larger deep learning community. Read more…

By Tiffany Trader

IDC: Will the Real Exascale Race Please Stand Up?

February 21, 2017

So the exascale race is on. And lots of organizations are in the pack. Government announcements from the US, China, India, Japan, and the EU indicate that they are working hard to make it happen – some sooner, some later. Read more…

By Bob Sorensen, IDC

ExxonMobil, NCSA, Cray Scale Reservoir Simulation to 700,000+ Processors

February 17, 2017

In a scaling breakthrough for oil and gas discovery, ExxonMobil geoscientists report they have harnessed the power of 717,000 processors – the equivalent of 22,000 32-processor computers – to run complex oil and gas reservoir simulation models. Read more…

By Doug Black

HPE Extreme Performance Solutions

O&G Companies Create Value with High Performance Remote Visualization

Today’s oil and gas (O&G) companies are striving to process datasets that have become not only tremendously large, but extremely complex. And the larger that data becomes, the harder it is to move and analyze it – particularly with a workforce that could be distributed between drilling sites, offshore rigs, and remote offices. Read more…

TSUBAME3.0 Points to Future HPE Pascal-NVLink-OPA Server

February 17, 2017

Since our initial coverage of the TSUBAME3.0 supercomputer yesterday, more details have come to light on this innovative project. Of particular interest is a new board design for NVLink-equipped Pascal P100 GPUs that will create another entrant to the space currently occupied by Nvidia's DGX-1 system, IBM's "Minsky" platform and the Supermicro SuperServer (1028GQ-TXR). Read more…

By Tiffany Trader

Tokyo Tech’s TSUBAME3.0 Will Be First HPE-SGI Super

February 16, 2017

In a press event Friday afternoon local time in Japan, Tokyo Institute of Technology (Tokyo Tech) announced its plans for the TSUBAME3.0 supercomputer, which will be Japan’s “fastest AI supercomputer,” Read more…

By Tiffany Trader

Drug Developers Use Google Cloud HPC in the Fight Against ALS

February 16, 2017

Within the haystack of a lethal disease such as ALS (amyotrophic lateral sclerosis / Lou Gehrig’s Disease) there exists, somewhere, the needle that will pierce this therapy-resistant affliction. Read more…

By Doug Black

Weekly Twitter Roundup (Feb. 16, 2017)

February 16, 2017

Here at HPCwire, we aim to keep the HPC community apprised of the most relevant and interesting news items that get tweeted throughout the week. Read more…

By Thomas Ayres

HPC Technique Propels Deep Learning at Scale

February 21, 2017

Researchers from Baidu’s Silicon Valley AI Lab (SVAIL) have adapted a well-known HPC communication technique to boost the speed and scale of their neural network training and now they are sharing their implementation with the larger deep learning community. Read more…

By Tiffany Trader

IDC: Will the Real Exascale Race Please Stand Up?

February 21, 2017

So the exascale race is on. And lots of organizations are in the pack. Government announcements from the US, China, India, Japan, and the EU indicate that they are working hard to make it happen – some sooner, some later. Read more…

By Bob Sorensen, IDC

TSUBAME3.0 Points to Future HPE Pascal-NVLink-OPA Server

February 17, 2017

Since our initial coverage of the TSUBAME3.0 supercomputer yesterday, more details have come to light on this innovative project. Of particular interest is a new board design for NVLink-equipped Pascal P100 GPUs that will create another entrant to the space currently occupied by Nvidia's DGX-1 system, IBM's "Minsky" platform and the Supermicro SuperServer (1028GQ-TXR). Read more…

By Tiffany Trader

Tokyo Tech’s TSUBAME3.0 Will Be First HPE-SGI Super

February 16, 2017

In a press event Friday afternoon local time in Japan, Tokyo Institute of Technology (Tokyo Tech) announced its plans for the TSUBAME3.0 supercomputer, which will be Japan’s “fastest AI supercomputer,” Read more…

By Tiffany Trader

Drug Developers Use Google Cloud HPC in the Fight Against ALS

February 16, 2017

Within the haystack of a lethal disease such as ALS (amyotrophic lateral sclerosis / Lou Gehrig’s Disease) there exists, somewhere, the needle that will pierce this therapy-resistant affliction. Read more…

By Doug Black

Azure Edges AWS in Linpack Benchmark Study

February 15, 2017

The “when will clouds be ready for HPC” question has ebbed and flowed for years. Read more…

By John Russell

Is Liquid Cooling Ready to Go Mainstream?

February 13, 2017

Lost in the frenzy of SC16 was a substantial rise in the number of vendors showing server oriented liquid cooling technologies. Three decades ago liquid cooling was pretty much the exclusive realm of the Cray-2 and IBM mainframe class products. That’s changing. We are now seeing an emergence of x86 class server products with exotic plumbing technology ranging from Direct-to-Chip to servers and storage completely immersed in a dielectric fluid. Read more…

By Steve Campbell

Cray Posts Best-Ever Quarter, Visibility Still Limited

February 10, 2017

On its Wednesday earnings call, Cray announced the largest revenue quarter in the company’s history and the second-highest revenue year. Read more…

By Tiffany Trader

For IBM/OpenPOWER: Success in 2017 = (Volume) Sales

January 11, 2017

To a large degree IBM and the OpenPOWER Foundation have done what they said they would – assembling a substantial and growing ecosystem and bringing Power-based products to market, all in about three years. Read more…

By John Russell

US, China Vie for Supercomputing Supremacy

November 14, 2016

The 48th edition of the TOP500 list is fresh off the presses and while there is no new number one system, as previously teased by China, there are a number of notable entrants from the US and around the world and significant trends to report on. Read more…

By Tiffany Trader

Lighting up Aurora: Behind the Scenes at the Creation of the DOE’s Upcoming 200 Petaflops Supercomputer

December 1, 2016

In April 2015, U.S. Department of Energy Undersecretary Franklin Orr announced that Intel would be the prime contractor for Aurora: Read more…

By Jan Rowell

D-Wave SC16 Update: What’s Bo Ewald Saying These Days

November 18, 2016

Tucked in a back section of the SC16 exhibit hall, quantum computing pioneer D-Wave has been talking up its new 2000-qubit processor announced in September. Forget for a moment the criticism sometimes aimed at D-Wave. This small Canadian company has sold several machines including, for example, ones to Lockheed and NASA, and has worked with Google on mapping machine learning problems to quantum computing. In July Los Alamos National Laboratory took possession of a 1000-quibit D-Wave 2X system that LANL ordered a year ago around the time of SC15. Read more…

By John Russell

Enlisting Deep Learning in the War on Cancer

December 7, 2016

Sometime in Q2 2017 the first ‘results’ of the Joint Design of Advanced Computing Solutions for Cancer (JDACS4C) will become publicly available according to Rick Stevens. He leads one of three JDACS4C pilot projects pressing deep learning (DL) into service in the War on Cancer. Read more…

By John Russell

IBM Wants to be “Red Hat” of Deep Learning

January 26, 2017

IBM today announced the addition of TensorFlow and Chainer deep learning frameworks to its PowerAI suite of deep learning tools, which already includes popular offerings such as Caffe, Theano, and Torch. Read more…

By John Russell

HPC Startup Advances Auto-Parallelization’s Promise

January 23, 2017

The shift from single core to multicore hardware has made finding parallelism in codes more important than ever, but that hasn’t made the task of parallel programming any easier. Read more…

By Tiffany Trader

CPU Benchmarking: Haswell Versus POWER8

June 2, 2015

With OpenPOWER activity ramping up and IBM’s prominent role in the upcoming DOE machines Summit and Sierra, it’s a good time to look at how the IBM POWER CPU stacks up against the x86 Xeon Haswell CPU from Intel. Read more…

By Tiffany Trader

Leading Solution Providers

Nvidia Sees Bright Future for AI Supercomputing

November 23, 2016

Graphics chipmaker Nvidia made a strong showing at SC16 in Salt Lake City last week. Read more…

By Tiffany Trader

BioTeam’s Berman Charts 2017 HPC Trends in Life Sciences

January 4, 2017

Twenty years ago high performance computing was nearly absent from life sciences. Today it’s used throughout life sciences and biomedical research. Genomics and the data deluge from modern lab instruments are the main drivers, but so is the longer-term desire to perform predictive simulation in support of Precision Medicine (PM). There’s even a specialized life sciences supercomputer, ‘Anton’ from D.E. Shaw Research, and the Pittsburgh Supercomputing Center is standing up its second Anton 2 and actively soliciting project proposals. There’s a lot going on. Read more…

By John Russell

Tokyo Tech’s TSUBAME3.0 Will Be First HPE-SGI Super

February 16, 2017

In a press event Friday afternoon local time in Japan, Tokyo Institute of Technology (Tokyo Tech) announced its plans for the TSUBAME3.0 supercomputer, which will be Japan’s “fastest AI supercomputer,” Read more…

By Tiffany Trader

IDG to Be Bought by Chinese Investors; IDC to Spin Out HPC Group

January 19, 2017

US-based publishing and investment firm International Data Group, Inc. (IDG) will be acquired by a pair of Chinese investors, China Oceanwide Holdings Group Co., Ltd. Read more…

By Tiffany Trader

Dell Knights Landing Machine Sets New STAC Records

November 2, 2016

The Securities Technology Analysis Center, commonly known as STAC, has released a new report characterizing the performance of the Knight Landing-based Dell PowerEdge C6320p server on the STAC-A2 benchmarking suite, widely used by the financial services industry to test and evaluate computing platforms. The Dell machine has set new records for both the baseline Greeks benchmark and the large Greeks benchmark. Read more…

By Tiffany Trader

What Knights Landing Is Not

June 18, 2016

As we get ready to launch the newest member of the Intel Xeon Phi family, code named Knights Landing, it is natural that there be some questions and potentially some confusion. Read more…

By James Reinders, Intel

KNUPATH Hermosa-based Commercial Boards Expected in Q1 2017

December 15, 2016

Last June tech start-up KnuEdge emerged from stealth mode to begin spreading the word about its new processor and fabric technology that’s been roughly a decade in the making. Read more…

By John Russell

Intel and Trump Announce $7B for Fab 42 Targeting 7nm

February 8, 2017

In what may be an attempt by President Trump to reset his turbulent relationship with the high tech industry, he and Intel CEO Brian Krzanich today announced plans to invest more than $7 billion to complete Fab 42. Read more…

By John Russell

  • arrow
  • Click Here for More Headlines
  • arrow
Share This