In a series of monthly video debates, Computer Weekly and the Financial Times invite a panel of experts to discuss business and IT issues.
In this first video, Duncan MacInnes, founder of Xenfin Capital, Steve Grob, group strategy director of IT solutions provider Fidessa join Computer Weekly editor in chief Bryan Glick and FT technology correspondent Maija Palmer to discuss the key IT issues facing financial services firms.
The debate covers the influence of cloud computing, high-volume trading and latency, and the costs of technology to the finance industry.
Read a transcript of this video:
Maija Palmer: Welcome to the Financial Times' Business and Technology Video Debates. The financial services sector is estimated to account for more than 9% of the total global spending on technology. The industry is known for its love of counting money in, but it is likely to see its IT outgoing rise to nearly $400 billion a year by 2013. In particular, it faces vast costs related to collecting, managing, and storing data. Will the companies simply have to pay up, smile gladly, or can they manage their costs and resources so that they do not literally bust the bank? With me, to discuss the issues are Duncan MacInnes, Founder of Xenfin Capital, Steve Grob, Group Strategy Director of IT solutions provider Fidessa, and Bryan Glick, Editor-in-Chief of Computer Weekly. Gentlemen, welcome. Let us first address this issue of whether financial services companies can streamline their IT systems so that they can manage these costs,especially to do with the vast amount of data that they handle. Bryan Glick, what can they do? Can they turn to Cloud computing, for example? Bryan Glick: I do not see many banks or many financial services companies looking at Cloud for their real competitive edge systems, but for those back-office systems, where they are looking to cut costs, Cloud is something that is becoming a viable option for them. Maija Palmer: Steve Grob, is this something that you are seeing and your clients ask for? Are they moving to Cloud computing, in any way?
Steve Grob: I think, to come back to your point about the pressures on these firms, if you look at the wall of regulation that they now face, it is a huge problem. I think, as well as looking at new technologies, there is combining that with practicality. For example, if you were to look at market abuse, that particular directive means some firms have to keep up to 10 years worth of data. That is ideal for Cloud computing, but also why not work together, because that is not a competitive application area. It would be much better for banks to have one collective utility where they share all of the resources, and perhaps deliver it through a Cloud computing application.
Maija Palmer: Duncan MacInnes, what solutions are you looking at, and is it easy to reconcile them with regulation?
Duncan MacInnes: I'm not really looking at it, personally, for my business. Obviously, we are an algo trading firm and we are very much looking at the performance of systems. I have a real issue with looking at the regulation side of the business and putting that information into a Cloud infrastructure. The fairly big providers of Cloud infrastructure are all US-based, so we get the real issues of putting information that is very much UK or European based into the States, you get issues with the SEC and what have you. Looking, actually, at our trading engine side, there is just no way that I would put that into a Cloud computer environment at the minute.
Maija Palmer: The other area in which technology is revolutionizing the financial services industry is this area of high frequency trading, where the issue of latency, the speed of trading, is very crucial. This is where doing a very small value trades but large numbers of them at high speed. Is it becoming something of an arms race? Can companies afford not to take part, Duncan?
Duncan MacInnes: I definitely agree with that. We started out, very much, as an ultra-high frequency firm back in the early days, '04 and '05. in the States, near Chicago, next door to the CME, low latency links down to New York for FX venues. We got to the point where we just spending so much on IT, and we were a relatively small company, and it was just getting ridiculous, the amount that we were paying every month just to keep up with the other guys. We actually backed down out of that arena, and have now very much concentrated our efforts in other areas in the market where we find it much easier to make money.
Still, very much, we do look at latency. We are a market maker, in effect, so it is very important to us. Certainly, we are not looking at it to the nth degree of shaving three or four milliseconds off of a link. We always like to have the best, but there is a level where that becomes, 'We are spending too much on this, against what we need to spend just to be in the game.'
Maija Palmer: Steve Grob, what costs are companies facing when they get into this game? Are there alternatives for them?
Steve Grob: I think there are. I think Duncan is probably right, that there is a distinct law of diminishing marginal returns with all of this stuff. You can almost see a number of different tiers within that high frequency game. The first, if you think of the premier league, you got the very top tier for whom they will simply spend whatever the money is to get the absolute very best because that is the basis of their trading strategy. Then, I think, you got a sort of mid-tier of players for whom latency . . . they are aware of latency; it is an important issue. And then, perhaps, at the bottom, you got perhaps the fashionistasm who are not quite sure why they are even involved in that particular game, and probably would do much better to look at their complete trading workflow, because it will only go as fast as the slowest piece.
Maija Palmer: Bryan Glick, are there risks involved in moving to systems which are so highly automated?
Bryan Glick: Yes, there is a risk involved. A few years ago at the London Stock Exchange, for example, who spent hundreds of millions of pounds developing an in-house trading system, and by the time they went live with it, it was a few microseconds slower in trades than a lot of their rivals. Within a couple of years they bought a third party software company that provided a faster trading platform in order to replace that. They are caught in a speed war now, with all the other trading platforms to be the fastest.
Duncan MacInnes: I think that is actually quite an interesting point in that, now, it is almost turned a lot towards the exchanges that are really latency-sensitive, in terms of the users, the buy-side are really demanding that the sales-side are actually decreasing their latency and upping their spend on their core kit.
Maija Palmer: It is hard to see where it is all going to end?
Duncan MacInnes: Yes, exactly. You just get into that circle, and it just keeps going round.
Steve: I think you are right. As the focus gets on to exchanges and other platform operators, you then say, 'Perhaps that is where the risk management needs to go.' If you are building a faster and faster racing track for people to drive around, then you are probably duty-bound to put the tires in the corners in case people run off. I think that is certainly something that we are starting to see now.
Bryan Glick: People are still building the race track though, at the moment. At the moment, there is a new cable being laid under the Atlantic.
Steve Grob: C2.
Bryan Glick: Yes, that is right. Which is purely designed to provide a few extra microseconds shaved off the trading time between London and New York. It is a hugely expensive exercise, but the people who are laying the cable have pretty much got that cost more than covered by these real high, sorry, low-latency specialists, who really need those milliseconds.
Maija Palmer: Gentlemen, I am afraid we will have to stop there.
Thank you very much. If you are interested in this topic, then you can read more online at FT.com/ConnectedBusiness.