Listen to Mark Palmer, TIBCO SVP of Integration and Event Processing, discuss how real-time analytics has evolved from Wall St. to Main St., revolutionizing the way we do business in all industries through streaming data and real-time decision-making.
Ellis: Welcome to the TIBCO podcast. I’m your host, Ellis Booker. Today, our guest is Mark Palmer, senior vice president and general manager of integration and event processing at TIBCO software. Hello, Mark.
Mark: Hi, Ellis.
Ellis: Mark, real-time analytics is a familiar idea on Wall Street, has been for years, but you say it makes sense for all sorts of industries. Why do you say that?
Mark: Well, because all industries really need to be more real-time and responsive to mobile devices. You have people checking in on their mobiles. There’s a huge trend right now in the internet of things. So we have sensors now everywhere and they emit streaming data. Much like if you look at Wall Street they have streaming data from the markets, from the exchanges, who’s trading what, you know, stocks are up, what stocks are down. So, all enterprises are really beginning to look like this, kind of like the way Wall Street firms started to evolve out, 10, 15 years ago, maybe even more to having streaming data all over the place and having to make real-time decisions.
Ellis: On that idea, can you tell us about an unexpected place where we’ve seen real-time analytics make a difference?
Mark: Well, a recent one that I bumped into was we have a lot of customers in the gaming casino space at TIBCO. And one of them is the provider, one of the leading providers of real-time predictive, analytics on a betting line data. Now, what’s interesting about the real-time angle is that I wasn’t aware of this until I started meeting with our customers, that 80% of the world’s betting activity today starts . . . actually happens after the game starts as opposed to, you know, you put a bet down and then you just watch the game go. With the prevalence of mobile devices and the internet people, you know, bookies and betting firms keep it interesting by, you know, if the games goes off, if it’s a soccer game or something and it goes up 4-0 and they will actually issue new odds to, you know, say if pretend that if the game was 0-0, who’s going to win from here on out?
If a player gets hurt of if conditions change in the game they are constantly sort of resetting and setting new lines, new odds as the game’s happening to create a more interesting experience for people betting and also just obviously increasing the volume of bettings going on. So that’s . . . it’s an area where I don’t think most people think about it but obviously real-time data and insights are critical to actually participate in that kind of new style of betting.
Ellis: So, real-time analytics in the gaming industry, let’s stay on that. Can you tell us a little bit more about that customer? Did something happen that can tell them to move in this direction?
Mark: Yeah, well, you know, what happened was that they, so one of the firms in this space called TXODDS, the customer of TIBCO’s and what they had an idea for was, they were already publishing out the data, the real-time odds information to online media publications. So, if you go to a website and you see what the odds are. That information, they outsource from firms like TXODDS and they’ll get that stream of betting line data from them and they serve the purpose of being an aggregator of the data. Just like in the financial markets, Reuters is sort of the . . . they provide all the market data from all the exchanges in the world. So, they have the idea though that based on their knowledge of the way the betting lines work that they could come up with a set of predictive analytics to understand, based on activity in the game and based of movements of line by particular bookies and institutions, that they could actually predict in real-time which way the line was going to move for these in-game betting scenarios.
So, they actually came up with that analytic and they sell it as an advanced service of predictive analytic feed that obviously can be really valuable for betters that are interested in making these in-game decisions on which way the line is going to go, with some additional insights.
Ellis: That’s fascinating. And I suppose, just like the Wall Street scenario we talked about earlier, this is going to prompt their competitors to move in the same direction. Have we seen that already?
Mark: Well, yeah there’s a lot of users of this data feed and of course just like on Wall Street, you know, people have their own proprietary special sauce on how they trade on Wall Street and just like that people have the same . . . their own kind of mechanisms and interests in figuring out which line, which way the line is going to move. So, yeah. So, anywhere from, you know, sort of the general public which might subscribe to this feed via a media outlet to, you know, in the world of betting there are professional betting institutions as well. They’re just the people that speculate a lot. So, there’s a lot of different uses for that kind of predictive stream of analytics that come from TXODDS and I think also you can get a lot of this information now if you’re in a casino as well. That’s really kind of a transformation towards a different style of getting information on how to bet.
Ellis: Right, now if I’m a gambler the first if I hear about this story, the thing I’m going to ask about is can they predict the future of any one gambler based on their activity? In other words, am I as the individual gambler now at some sort of a disadvantage because my activities are understood and the people [inaudible 00:06:15]?
Mark: Well, you know I think that most firms that, and this holds the same for Wall Street, I mean most of this information is used to sort of improve the experience of the individuals that are . . . for example walking into a casino taking the kind of stepping away from just the data parts, you know, Caesar’s for example is a customer of ours and they use a lot of this real-time infrastructure to know when you walk into the casino, what kind of games you tend to like to play, what are your preferences for where to stay in the resort? And if they have access inventory, perishable inventory that they’re not using for other purposes, they’ll give you a special offer or, you know, comp your room or something that aligns with your interest. So, the whole goal is really to improve the experience of the player rather than try to beat them anyway. We just give them more information so they can have more fun and be more engaged with the whole experience.
And I think that’s the same for this predictive line. Because at the day these are all still odds. Nobody knows which way a game is going to go. Nobody knows where a line is going to go but, you know, by getting more information it becomes a more immersive experience, a multi-gaming type experience in the whole betting process. So, I think real-time is the way that that really can be uniquely achieved.
Ellis: Moving off the casino example, more broadly when real-time analytics help predict future behavioral events, what does that mean for business strategy? Won’t that challenge many businesses that are existing today?
Mark: Well, you know, predictive is a very large field and you know there’s a couple of different . . . I view it as there’s really almost two classes of predictive analytics. One, which is really based on big data Hadoop, deep data warehousing and analytics tools and TIBCO has a great one in the Spotfire environment where you can explore history and try to predict what’s going to happen next. And, you know, companies have been doing that for a long time. They’ve used more rudimentary tools for forecasting purposes and research purposes but that really is evolving into quite a science now in the enterprise. The other class of predictive analytics is what we’re really talking about here in this betting scenario which is real-time predictive analytics. And that really is being used by enterprises to predict operational changes. Very small changes during the day based on the conditions that are happening right now. So, the betting example is a good one.
The game score changes, lines begin to change. If you can correlate those two events then you can statistically predict which way a line’s going to go, which way a game might move. So, you know, I think fundamentally this is about making businesses smarter, making them more efficient, to making it so that they can provide a better interactive and immersive experience for customers who are increasingly interacting with companies and casinos and firms of any type. Retail operations via their mobile phones so they can do it real-time, all the time, anywhere and that predictive aspect is important to providing that experience in a positive way.
Ellis: I see. So, for an organization that is already engaged in, let’s call it, traditional forecasting but wants to move in this direction toward real-time analytics, what’s required from an infrastructure standpoint, how long is this all going to take, how much is it going to cost?
Mark: Well, you know, again it depends. It always depends on the firm, how prepared they are for real-time. A lot of companies already have access to streaming data everywhere. You know, in Wall Street for example you can’t trade unless you have a real-time feed of market data. Now, that said, in, you know, I met recently with the CIO of a major theme park in the U.S. like of the one of the bigger ones. The CIO there was saying, “You know, we know we need to make these real-time predictive decisions but we still to this day don’t have sort of sensors in the theme parks. We don’t really know what customers are looking for or trying to do or what they’re interested in buying.” So, they need the data, right? So, gathering that real-time data is the first step in that but then once you have that then you need a platform, an infrastructure and of course this is TIBCO’s specialty which is this fast data platform which allows you to absorb all that data and then make sense of it and then create or offer these predictive analytics.
That’s sort of our sweet spot but once you have that infrastructure in place, I mean it goes very fast. I was meeting with the chief architect of a trading operation just last night in New York and he was telling me that they bring out new applications, new ideas sometimes in just a day or two. Significant new ideas that they test in the market, they make the system smarter and then they evolve them. So the flexibility and the ability to move very quickly once you have that framework and that architecture in place is really what TIBCO is all about. It’s like making that foundation of . . . you need a fast one to build upon.
Ellis: So, the CIO who is intrigued with this concept, I’m sure a question that comes up is what happens to our traditional investments in the traditional data warehouse? Do those things still have a purpose?
Mark: Absolutely, every system that we do that is a real-time system leverages an existing warehouse or an existing Hadoop repository or existing historical information. You know, again the two aspects are predictive sort of forecasting based on what’s already happened and then sensing and responding to real-time events as they happen, those are two different types. You know, the decision on how to respond to real-time events always has to be informed by the history and the mathematics of what’s likely to happen. And the only way that you can come up with that predictive signal is to analyze the past. And let’s go back to that, the predictive line, betting line example. In order to come up with those analytics that they execute in real-time, they need to dig through every piece of history. The information they have about a particular match, who played who, what players were hurt, what players weren’t? What part of, you know, what are their statistics?
What, you know, were they playing home or away? Were they, you know, what was the situation in the season? Was one team behind, were they way up? You know, all that data needs to be considered before you can sort of predictively determine in real-time, which way things are going to go. So, that investment is, the short answer to the question of the investment is still extremely important, relevant and invaluable in terms of moving more real-time.
Ellis: And, finally Mark, do you have any advice for organizations that might move in this direction and where should they start?
Mark: Yeah, I just did a whole session on that in [inaudible 00:13:51] at Las Vegas just last week about sort of the steps to adopting real-time. And I think one of the biggest mistakes is that CIOs try to move too fast, too soon, too big. The reality is that thinking about operations in an algorithmic way, in a mathematical way is very, very different for most organizations. In fact, more than that it’s also threatening to the existing people that work there. You know, they kind of have the perception . . . and this happened in the Wall Street trading. The traders thought that algorithms were going to replace them. Well the reality is that algorithms augments what they do. They make what they do more informed and more intelligent.
It does require them to understand mathematics more. That is one of the most important, I think, jobs in the 21st century. So, for CIOs to start it’s really picking through, you know, what is the biggest impact that you can move towards real-time taking kind of a baby step. And then also picking the people that are going to be able to adapt to that new model, right? People that have a mathematics background and the analytical background and that are actually innovators in terms of thinking of a way to disrupt an existing way of doing things. You know, you have to think in terms of how can we do this better in a machine aided, machine guided way and come up with some of these, you know, ideas like predictive signals for which way the line and a sporting event might go? I mean things like that are quite often, I think, surprising to a lot of people that they exist and which requires a little bit of vision. So, start small, get some innovators and think in terms of math is, I think, the best advice that I’d like to give to a CIO starting out.
Ellis: And that will have to be the last answer. We’ve run out of time. I’d like to thank Mark Palmer of TIBCO for joining us today with some fascinating information and a case study. You’ve been listening to the TIBCO podcast. I’m Ellis Booker, your moderator, thanks for listening.