By Andrew Nusca
Posting in Environment
The technology known for dazzling Jeopardy! viewers could add a dose of intelligence to an industry where decisions about money mean everything.
The financial services sector is no stranger to cutting-edge technology -- after all, a microsecond may be all the difference between a successful securities trade and a money-losing one.
Much ink has been spilled about the technical arms race in which financial firms are engaged, but what is beyond the horizon? Faster, yes -- but then what?
One potential answer: IBM's Watson. The technology known for dazzling Jeopardy! viewers last year could very well add a dose of intelligence to the industry tasked with manipulating money, whether on the rapid-fire terminal displays of Wall Street or the more pedestrian versions behind your local bank's counter.
We spoke with IBM Watson Solutions director Stephen Gold.
SP: Financial firms are always gunning for the next big thing. What do they see in Watson?
SG: It's a data-rich industry, but it's an industry starved for insight when it comes to the ability to leverage that information for timely and accurate action. If you think about it from a consumer banking perspective, an institutional perspective...nobody doubts the volume of data. There is no industry that has more data available than the financial services sector. Just a few years ago they had over 619 petabytes available. How do you put that data to work?
The five new research articles coming out of Wall Street every hour every day -- how much of that is being consumed to make better actions? Are we just making something better, or are we solving a problem? With the ability to discover new insights, it's opening up a whole new world of possibilities. It's not just something better.
It's obviously very early on. Watson, as we've come to discover, really represents a shift in computing. We're used to programmatic approaches; Watson is a learning system. We're just starting to think about how to put in a learning system. The keyword may go away. If I'm a bank today, and I'm looking at my consumer, there are a couple of things that jump out to me. One: I don't have total mindshare with my audience. Consumers spread their finances out across institutions. As a consequence, institutions have a harder time getting a robust view of that customer. Mining information, it should truly be a 360-degree view -- I know that term has been over-marketed, but -- of an individual.
An individual contacts a bank. Has an interaction. They're looking to do something. The traditional response is product-centric: here's a home-equity loan we have. But what's driving the need for that home equity loan? Understanding that will give us a better perspective to what that interaction should be. With Watson, that ability to discover and mine structured and unstructured data, it's going to allow them to have a whole new capability to personalizing the level of service they can provide. The consumer gets a better combination of offerings suited to their needs.
Attrition is huge at banks. If you can build a more formidable relationship with banks, there's a really good chance you're going to retain that relationship for a long period of time.
SP: So that's the Main Street impact. How about Wall Street?
SG: When you look at the institutional side of the business, it's really more about putting content and context.
Something as innocuous as the dollar-yen relationship, for example -- if you're a client on the institutional side, you may have an interest to better understand monetary strategies. But there are something like 4.3 million trades a minute on the New York Stock Exchange. There are political events, statements, guidance, governance -- so much more that can influence monetary policy. Being able to ingest indirectly relevant data at that moment is going to provide that trader, who is tasked with a Herculean feat. They receive 1,000 e-mails a day.
None of the institutions want to [fully] automate with machines. They want to support their people, who are in fact the experts. Watson, similar to what it's doing in healthcare, becomes an assistant. Whether bonds or mortgage-backed securities or equities or currencies, Watson's going to bring back context.
On the institutional side, you'll see Watson initially used in a research capacity. On the consumer side, more in an interactive capacity.
SP: Do financial firms see Watson as a way to improve what they're doing, or get a leg up on a rival?
SG: At the senior level in an institution, they're looking at this as a mandate for how their business is going to operate moving forward. If you're talking to a trader, they're seeing this as a competitive game-changing event. But big data is not new. We've all become accustomed to a highly tailored personal interaction: I call a bank, they ask me for my social security number and zip code, I will have the expectation that they will then know who I am and won't ask me again.
It's almost becoming this imperative. Obviously this technology is still leading edge; it's still being evaluated. But the technology is a means to an end. Better decisions, better outcomes. As an institution, you never want to think you've made a decision without data. As a consumer, you don't want a loan a quarter-point higher than you could have gotten.
SP: Markets can be emotional. We've seen the irrational swings since the crash in 2008. Can contextual decision-making cut down on this?
SG: We'e been heavily dependent upon search. Watson brings it back to discovery. Putting content in context is going to make for better decisions, for the general masses. The ability for individuals to evaluate a market condition -- a trader sitting at a terminal has a different aperture to the marketplace than I as an individual do. I think you're going to see that gap close. I think that will ultimately have an effect on volatility and activity.
Watson is a system of systems, but the three critical technologies that underpin it -- natural language, hypothesis generation and evaluation, learning system -- you know, trading terminals are highly complex. If you can bring it down to a more natural way to interact with it opens up a whole new world of possibilities...
The other thing Watson does is takes my question and breaks it down into sub-components. If I said, "The relationship between the euro and yen moving forward." Its ability to help me ask better questions and bring back weighted responses so I don't have to decipher among a million of possible responses, that makes my job as a trader or investor easier.
Today, the institutional environment is highly dependent upon the individuals that participate. Trades, market swings -- all of that is predicated on a traditional, rules-based, methodologies, strategies, different ways of applying information. But an environment that gets smarter? It's going to say -- and remember, Watson has no bias, it's objective -- this is what happened. The next time you ask it this question, it's going to be smarter. As it's afforded more access to individuals and organizations, we're going to see less decisions that are based on hearsay.
Big data is the fuel in the tank, that richness of data that is available. Combine that with what research brought to market a year ago, Watson. Break it down and learn from each outcome. That's what's fueling the financial services sector interest in taking this next big step.
The question isn't whether this will be an approach this works; the question that remains to be seen is where it will be applied first. Organizations recognize there is an opportunity to make better decisions. Financial planning, credit card operations, equities, bonds -- which area remains to be seen. We know we have to do it, it's just a matter of where we put this to work first. That's the part where all the institutions are secretive -- it's a competitive edge to put it in place to work [on a given financial instrument] first.
This is an industry that's not afraid to invest. Last year, [market intelligence firm] IDC said banks spent $400 billion on IT. You're looking at an industry that's willing to push hard to advance itself.
Photo: Kathleen Tyler/IBM
May 8, 2012