Extreme Connectivity & The Rise of The Cyber Cities
Thu, 21 Oct 2010 00:01:00 GMT
An Interview with Bob Giffords
In this week’s interview for the High Frequency Trading Review, Mike O’Hara speaks with expert banking and financial technology consultant Bob Giffords.
Bob is an independent analyst and consultant for the European banking sector with over 25 years experience in financial technology, mostly as an independent consultant. Bob writes for several leading financial journals, including Automated Trader, The Trade, Swift Dialogue, and Financial World. He has been a judge on two banking technology innovation awards and has chaired major banking conferences. Previously, he was Director, EMEA Research and Consulting, for Financial Insights, an IDC Company where he built up a pan-European consulting practice. Before that he was Chief Technology Officer with EuroMTS Ltd., the leading European electronic bond-trading platform.
High Frequency Trading Review: The question I ask everybody to start with is what’s your definition of high frequency trading?
Bob Giffords: I have a fairly open definition of high frequency trading because I find that the participants themselves have a fairly open definition, which I don’t want to constrain.
For example, some of the investment bankers, the brokers, will consider what they do, in terms of smart order routing and order slicing and those kinds of algorithms, not to be high frequency trading.
However, as it is speedy and happens quite quickly, then as far as I’m concerned and many who responded to a recent survey I analysed, it’s high frequency trading. Some dynamic hedging is high frequency trading if it’s fast enough and dynamic enough during the day. So I’m very open about it. I don’t limit it simply to flat-trading market-makers with holding times of milliseconds.
There’s a wide variety of what you could call high frequency traders. The key thing obviously is that they are doing a lot of trading during the day. Most of them, in order to optimize that trading, will have a high orders-to-trades ratio, but that’s not a requirement. In studies that I’ve seen, fairly low orders-to-trades are also claimed by some high frequency traders.
There are no real rules but clearly, the speed of trading, the volume of trading intraday and increasingly low latency are all involved in it.
HFTR: So in that case, with such an open definition, would you say that the terms “algorithmic trading” and “high frequency trading” are synonymous?
BG: Again, for some people, they are and for some people, they are very different. From my own point of view as long as it is high volume intraday and exploits some low latency techniques, it’s high frequency trading, but there’s no hard and fast rule.
For example, you might want to include some of the passive trading that keeps ETF’s in line with other instruments like futures and underlying stocks. Those may or may not have a lot of trades. It depends really on the volatility of different parts of the market at different times; so they may or may not be particularly high frequency and indeed some passive funds are able to keep in line with the indices in suitable market conditions with very few trades per week let alone intraday.
We tend to get confused if we focus too much on a particular definition. Using terms like high frequency trading and algorithmic trading may help to separate some of those things out if we do define them clearly for the particular problem we’re wishing to analyse. It really depends on what you’re trying to consider, what definitions would be appropriate for the time.
HFTR: How do you think the growth of high frequency trading (and algorithmic trading) has impacted the markets?
BG: Well, first of all, I think we need to be very clear about how the markets have impacted high frequency trading, because the main thrust has been in that direction rather than the other way around.
Without fragmentation, without the trade-through rules in the United States, we probably wouldn’t have seen so rapid a take-up for high frequency trading. So it really is a response, a quite natural market response, to a series of regulatory changes in the structure of the market.
Having said that, clearly high frequency trading has impacted the markets themselves and on the whole, it has been beneficial. It has reduced spreads. It has arguably reduced intraday volatility (except in some circumstances which we can come back to).
Certainly, there is less volatility and there are tighter spreads around macroeconomic releases. When these hit the market now, it’s all over in a second or two, unless there’s a big surprise. Also, trading costs have come down.
On the other hand, because of the increased volume of transactions and because of the latency implications of high frequency trading, for some firms data costs have gone up hugely. Although those for whom it has gone up are usually the ones who are trying to reap the benefits of high frequency trading. So, maybe that’s not a bad thing.
HFTR: Presumably you’re talking about the firms who take raw data feeds from the market with the intention of getting the fastest possible prices to feed their algorithms?
BG: Precisely. So, they are paying more, yes. But with increasing competition on the data front, there are cheaper solutions. They may be throttled, but for those who are not trying to optimize at the millisecond level, that’s perfectly okay.
We’re going through a transition, certainly on the data side. At the end of the transition, we’ll find that prices will be much more aligned to trading requirements than they are perhaps at the moment and therefore some of the higher costs of data will have been worked out of the system.
Clearly, if you need very fast data, you have to pay for it and that costs money. If there’s no alpha to be had for that money, then people won’t do it. Some people at the moment (because of the transition) are paying probably a little bit more than they have been in the past and therefore need to manage their market data costs better than they have in the past. But, overall, I think people can get into the market cheaper than they ever have in the past.
I go along with those who say that high frequency trading has democratized the markets and opened them up. That doesn’t mean there’s nothing to do though. There are definitely things to do, but I think high frequency trading has been, on the whole, very good for the markets and a very natural development.
HFTR: Coming back to something you just said, that high frequency trading has generally reduced volatility except in certain circumstances, can you expand on that? Which circumstances? The “Flash Crash” for example?
BG: Well, clearly in the sense of market-making when you’re buying and selling very fast, high frequency trading will tend to amplify any natural market movements. So, in the Flash Crash of 6th May, it moved the market extremely fast in a downward direction, because that is the direction the market was heading.
The problem there, was not so much high frequency trading as a lack of fundamental buyers and sellers because of the huge uncertainty that was taking place. There were some relatively lower frequency trades taking place and some big options and futures trades, which weren’t particularly fast, they were just going on and they were done in the normal course of trading.
Those things unsettled the market. In that situation and because of poor microstructure, the markets overreacted and high frequency trading tended to accelerate and amplify that, so we saw the prices plummet to a penny a share and so forth.
That, to my mind, was clearly a microstructure failure and a regulatory failure. My analysis is simply that the regulators were not on top of their game. They allowed a poor microstructure, which accelerated the problems and turned what should have been a fairly ordinary situation into a totally uncontrolled one.
If you compare what happened on CME for example, the futures market dropped, but it went to a bottom of a few percentage points, stopped, and recovered perfectly normally. Most of the futures trading in fact was done after the turn in the market. The problems were all in the cash markets, where there were some serious microstructure failures. That had nothing to do with high frequency trading. Much of the SEC/CFTC analysis focuses actually on the futures market where there really wasn’t that much to talk about, it was fairly normal. Okay, it was dramatic, but then the conditions were hugely pessimistic.
What is completely underplayed is how dramatic the conditions on that day actually were. You had rumours flying around the market (and this has been confirmed by the ECB itself) that two European banks, within the next day or two, were going to go under. Everybody thought it was another CreditAnstalt from 1931 and that completely spooked the market. Then the ECB had a briefing at 8:30 in the morning U.S. time and made no comment! I mean, the markets were totally spooked and the ECB didn’t help, so rumours were rife. Nobody knew what was going to happen. People thought that potentially we were going into another bust. We’d had all this quantitative easing so there were huge amounts of liquidity. There could have been a flight to cash. It was an extraordinary set of circumstances. A bank in Greece had been firebombed the day before, there was rioting in the streets and all this was being shown live on television.
All of this has been totally underplayed. That was the problem. The fact that high frequency trading accelerated it, well yes, that’s what you would expect, but what we need is much better market microstructure, rather more like the CME, where that kind of problem doesn’t happen. Of course with the futures market, liquidity is much more concentrated, so you don’t have the same problems. What we do have is a very uncontrolled situation in the cash market and that does need to be dealt with.
In fact, curiously, while most of the SEC/CFTC analysis was focused on 16 or 17 high frequency traders in the futures market, it turns out that 50% of the broken trades were executed by, a large broker/intermediary on the sell-side and a market-maker on the buy-side. So half of the broken trades were from a large ordinary broker (i.e. algorithmic, but not high frequency) basically turning market orders into limit orders and following the market all the way down to the bottom.
Perhaps in that case it was the failure of an algorithm to think a little bit more sensibly, but it certainly shouldn’t be described as a failure of high frequency trading. I mean, there are a lot of algorithms out there which are not terribly sensible in terms of their ability to evaluate a totally unstable market.
Clearly, for those who are writing algos that should be a lesson that they need to learn. On the other hand, even with that, there will always be people who get it wrong. There will always be rogue algos. The markets really need to be much better at preventing extreme behaviours and they should have just pulled the plug.
In Europe, we’ve got volatility interrupts which would have basically changed the whole thing. Most of the people I have spoken to don’t think it would have happened in Europe. I think it was a very specific microstructure failure in the United States rather than a problem of high frequency trading as such.
As for the absence of the fundamental traders that really establish fair value in the markets, that is part of a long term trend, and has more perhaps to do with socio-economic changes, globalisation, and other factors. The shift of the big money managers into index based trading could be another factor and there are the effects of currency volatility to consider as well. However, these issues are a million miles away from high frequency trading, but probably crucially affected market sentiment and liquidity on the day.
HFTR: No doubt we’ll continue to see a lot of analysis regarding what happened on that day and the controversy will continue. High frequency trading does seem to be surrounded by controversy and is a very emotive topic with some people, almost like three dirty words in some quarters. Why do you think that is? Is it ignorance? Is it fear? Envy? Why do you think high frequency trading has such a bad press and do you think any of the concerns are particularly valid?
BG: It’s probably all of the above and a lot of people elaborating and adding drama to things for the sake of it. Clearly, we’re in the middle of a huge global economic convulsion. That means that people will be looking for scapegoats. The politicians and the regulators have played heavily on the “naughty banker” theme since the beginning of the crisis in 2007. They don’t really want to admit that they are the ones who got it wrong and they haven’t been able to deal with the crisis so they’re out to find scapegoats.
Because high frequency traders trade very fast, because for some purposes they have some advantages over those who are not able to trade faster, people see that as an unfair advantage. As soon as you start to lose money, you look for somebody to say the level playing field has been tilted. So it’s a natural human reaction and I think it’s not surprising, we always like to blame our problems on other people.
Having read quite a lot about this and spoken to a huge number of people, I hear some of the suspicions but what I don’t hear is evidence. There are suspicions about “quote stuffing” for example. Quote stuffing means you pour so many quotes and orders into a particular market that the data feeds get clogged, which slows down the distribution of data, which means if you’re going very very fast, you can take advantage of the slow data because you get some bits that are slow, other bits are fast and you can arbitrage the differences. The regulators found nothing in the flash crash to suggest that happened.
Maybe it does exist. Maybe there is some quote stuffing taking place, but despite their best efforts, the regulators were not able to find any evidence. If you look at “front-running” as another example, I’ve seen reports that try to prove that it’s happening but are totally unable to prove it in any serious way.
Yes, there is always going to be market abuse and there were market abuses in open outcry trading. I’m not saying there is no market abuse now, but I really don’t think it is any worse than it has ever have been and in some respects, there are probably grounds for thinking there is less of it going on because of the increased transparency.
There are going to be periodic advantages where one clever trader might be able to game the market for a while, but since everybody is watching like a hawk, I really don’t think those advantages will last very long and progressively, they are being ironed out. I really cannot find (and I have tried) serious evidence to suggest that high frequency trading per se is causing any real problems at all that a sensible market microstructure couldn’t deal with.
On the contrary, in fact, without it, we wouldn’t have the tight spreads, we’d have higher fees, we’d have all the problems we’ve had in the past and if you look at the total amount of costs of intermediation that existed prior to the introduction of all these innovations, they were higher. Investors were losing more money before, as far as I can tell. Now I’ve seen there are some studies saying that things haven’t improved (it depends really how you do your numbers) but I’ve seen none that suggest things have actually got worse.
I would tend to think that things have got better. For example, you now find these one and two-man shops who are trading seriously and earning money in the new environment where they would never been able to do so before, and they’re trading globally.
HFTR: So is it just a case of educating the public then? Do firms who are involved in high frequency trading need to do a better job of being transparent about what they’re doing and countering some of the arguments and accusations that are continually levelled against them?
BG: I don’t think necessarily the high frequency traders are going to do much self-defence. A few of them are coming out and speaking, but not out of defensiveness. It’s basically to sell their services and show how clever they are!
I think this is just a phase we’re going through. We’ll get over it and hopefully the regulators won’t do anything too stupid and will correct some of the mistakes they’ve made in the past.
We have to accept a certain amount of “playing to the galleries” from the political stage. Everybody’s big enough and ugly enough to take that, so it will pass, it will blow over (I hope!) and we won’t see huge problems because, unfortunately, if people make things worse, then we’re going to have different problems and we have enough as it is.
HFTR: Indeed. Now Bob, I’d like to turn to the subject of m2m (machine-to-machine) eCommerce, which you’ve written and spoken extensively about. What do you think are some of the key challenges we face in this world of machines trading with each other?
BG: Well, machines move much faster than we do and they can work on a much larger scale than we can, so they can keep track of many, many more things than we can. The human brain is very good for some things; it’s very good for very complex and innovative pattern matching, for example. But machines are extremely good at very large scale, very fast “standard matching” and looking at things from lots of different angles simultaneously.
So there are a couple of issues. Clearly, one of them is that life is going to get faster (as we have seen), which creates a challenge for control. It’s going to become hugely more complex and the scale of things is shooting up unbelievably. And no one is properly addressing it. I am trying to make people recognize what’s going on, not to shout “Bogeyman! Stop it! It’s got to reverse!” No. We can’t reverse progress. We’ve never been able to reverse technological progress.
What we can do is learn how to deal with it. What we’re seeing is what I call extreme connectivity, and that extreme connectivity is hugely magnified by machine to machine activities. That is a global phenomenon. All the markets are increasingly interconnected. The FX markets are interconnected. The futures and commodity markets are increasingly connected too, and through ETFs they link with the stock markets, for example. There has been a huge increase in correlations because everybody’s watching everything else and moving in step, particularly because of the negative economic sentiment of our current crisis times. And there are very fast shifts taking place.
On top of all of that, you’ve got this huge amount of money sloshing around, mainly from quantitative easing and huge government deficit spending. You have enormous amounts of liquidity. You have an extremely fast and highly interconnected set of markets on a global level with money moving cross border, cross market, cross everything. You have a huge amount of speculation in terms of public policy, which is increasingly driving FX volatility. If you just look at what’s happened in the FX markets, the major currencies have been bobbing up and down 20%, 30% over the last few years.
So, given all of that, it is not surprising that we are going to have problems. The world is changing and we really do need to get to grips with it, but that’s not to say it’s a big bogey. I’m not saying that. It’s just that the world has changed and we have to deal with it. Let’s get on with it.
HFTR: One of the ways it seems to be changing, which I’ve also heard you speak about as part of this machine to machine e-commerce thread, is the massive growth we’re seeing in proximity hosted data centres and co-location facilities and so on, which I’ve heard you referred to as “cyber cities”. Can you explain this concept of cyber cities?
BG: Yes of course. But first of all, let’s be very clear about what we’re talking about. At the moment, proximity trading hubs are an extremely small part of the total trading environment. It is just a fraction of what is happening. The vast majority of traders trade from their own offices or use brokers, where the bulk of what they trade is also through their own offices, their own data centres. They may have some proximity trading presence, but it is only to serve the 1% or 2% of firms in the market that is, in any sense, high frequency. It is not really to support their own main brokerage activities or the vast bulk of flow that they manage.
Of course, that 1% or 2% of firms controls, in terms of transaction volume in the market, anything from 40% to 70%, depending on how you define things. So there’s a huge difference between the number of transactions, the volume of trading in terms of the total value of the turnover, and the number of firms. It’s chalk and cheese, and that’s partly why the high frequency traders have come in for such suspicion, because there are so few of them and they’re having such a large impact on the market. So that clearly is going to colour people’s views.
The cyber cities side of it refers to these large, lights-out warehouses holding thousands of computers that are talking to each other and transacting, but without any people in them, that phenomenon is only just beginning. My feeling, my strong forecast would be that this is going to build and build and so just as we saw huge growth in human cities in the 19th century, I believe the task of the 21st century will be to build the cyber cities of the world. This will not just be for financial markets; it would be for all industries.
The key thing in all of this is that everybody is connected via extremely high speed links that are very cheap to switch from one to another. So if I want to talk to Mr. X, Mr. X may be a market, Mr. X may be a data vendor, Mr. X may be a risk analytics guy, Mr. X might offer algorithmic trading capabilities, automated news or whatever. The key thing is that if I want to talk to him in the morning, I can be up, connected and interacting or transacting with him at high speed on the same day. That just completely changes the game. It enables a very creative supply chain to evolve and become much more dynamic.
Instead of saying, “Well, you and I are going to do a deal and here is some money upfront and we spend 3 months plumbing in a link between our data centres, and we know we can’t pull it out very quickly because I’m tied into you because of the investment that I’m making and you’re making” — that changes to, “Hey, I want to talk to you for a couple of hours. I switch you on. If I don’t get value, I switch you off. If you start to raise prices, maybe I can find somebody else I can switch in just as fast.” It completely changes the whole economic game, dramatically increases competition, and that applies to every aspect of commerce not just to financial markets.
What the financial markets are doing, because they are extremely high value, is they are subsidizing the technology growth to make it all work and to turn this stuff, hardware and software, into a commodity. Once that happens, prices will drop dramatically and as we’ve seen in every other computer evolution that’s ever happened, it will start to move out to much lower value transactions in all areas of e-commerce. So, from that point of view, I see cyber cities as a major, major development of the 21st century.
HFTR: Do you see any parallels evolving along similar lines to the development of human cities? I’m thinking in terms of crime, good and bad neighbourhoods, etc.
BG: Essentially yes, we will see everything in cyber cities that we see in ordinary human cities. We know, for example, there are cyber criminals. There are people who will try and outwit the system or use sharp practice to build a niche for themselves. If you look at the way criminal activity has evolved in ordinary cyber space and ordinary e-commerce, you have the so-called BotNets, which involve thousands (hundreds of thousands in some cases) of corrupted PCs that have been infected with various viruses in order to take them over and use them for co-ordinated cyber attacks, either through denial of service type attacks or through ordinary computer fraud, but on a massive scale of low value transactions that do not arouse suspicion.
So yes, we can expect to see all of that in cyber cities, but because they are so connected and so global, it will be increasingly difficult for national police forces to deal with them. That is one of the risks. But it doesn’t mean we stop there. It just means we have to get better in how we manage these things.
We’ll see bad neighbourhoods too. For example, at the moment, it’s very important to be in the same building as certain other key players because if you are in the same building, you can gain advantages. If you get shunted off to one of the “poorer neighbourhoods” as it were, you can find that even being relatively close in proximity may still not give you the advantage that you want. When there are thousands of orders in the queue a few microseconds can make all the difference.
And so what we see is that when people move from separated, but close buildings, transaction times or latencies come down from 1 to 2 milliseconds to a hundred microseconds. That’s a world away. We’re talking 10 or 20 times the speed, so being in a nearby proximity centre may not give you enough to really be at the top of the queue. So there will be upper class and lower class neighbourhoods in these cyber cities, but then prices will just have to adjust for those differences and I’m sure eventually we’ll have the normal urban regeneration-type schemes and so forth. But that’s still very much far along down the line!
HFTR: In conclusion Bob, do you have any other predictions of how the high frequency trading space will develop?
BG: Well, the “Big If” is of course the regulatory side. If that doesn’t cause problems, then I would expect to see a steady expansion of high frequency trading. I’m talking to more and more hedge funds in particular and even some long-only funds who are looking at higher frequency strategies and overlays which give them a certain amount of advantage, particularly for some market conditions. It does offer significant, uncorrelated opportunities for alpha but still with fairly limited capital capacity.
I would expect that to continue to grow. I think the problem that we have is not so much high frequency trading as the general fall of longer term trading in the market place and that’s what I’m currently focusing on.
High frequency trading on its own is not bad unless it’s only on its own, in which case there’s nothing to tie it down. What we need in the market place are much more longer-term “bets”, which are not going to just flutter around on the latest market sentiment. We’ve got to bring what I call this ‘long finance’ perspective back into the market and that’s where our focus should be, not really worrying too much about high frequency trading except to recognize that it’s changed the markets and, therefore, the market microstructure has to get its act together and really stop any silly excesses.
Once you do that then, on its own, the playing field is levelling out anyway. There is only so fast you can go before sooner or later, the fractions of a second are so small that the opportunities for alpha start to disappear and we have to move on to the next thing. There are some signs that that’s already happening in fact. Some people are finding it harder to make money with some of these strategies.
I’ll leave you with one final thought. Alvin Toffler wrote a book called “Future Shock,” back in the 1970’s and the one thing in that book that really struck me was the concept that knowledge is the fuel that accelerates the pace of change. If you look at high frequency trading, that is all about very, very specialized knowledge in low latency and high frequency, high volume automation. That has accelerated the pace of change.
It is not so much high frequency trading today as all the things it is going to bring. That’s why I talk about the global market fabric of extreme connectivity: speed, volumes, scale, diversity and a growing set of increasingly valuable intangibles. These are the big issues that the 21st century will have to deal with. How do we control things? How do we prevent things going over the edge? How do we basically manage our lives in this wholly new world that’s just going to accelerate and accelerate and accelerate?
Every year, things have speeded up dramatically. And the number of transactions in the world is just enormous. We’re moving from around a billion people in the developed world 20 years ago to what within the next 30 or 40 years will be somewhere between 5 and 10 billion people — the revolution of our times.
So in one lifetime, the number of active economic participants will have gone up by potentially a factor of 10 and each of those participants will be generating N times more transactions, which we will have to be able to cope with. That is a huge change.
High frequency trading is part of that change, but it’s only part of it and the bigger change is really how do we control the machine to machine world?
HFTR: Thank you Bob.