Is high frequency trading good for financial markets?

(piano music) – Flash crashes, exchange shutdowns, incorrect orders and algorithms gone wild. Technology has completely
transformed financial markets, and not always in a good way. For many, the culprit is
high-frequency trading, the practice in which black
boxes automatically send orders to buy and sell
at lightning-fast speeds and firms and exchanges compete by buying ever-more expensive technology to trade faster and faster. High-frequency trading now
dominates the equity markets and is growing in the derivative sector, a trend that concerns many observers. Regulators in Washington fret
about a race to the bottom in which speed is more
important than risk management. But is more regulation really the answer? And if so, should it be directed
at high-frequency trading or at the overall structure of markets? Welcome to The Big Question, the monthly video series from Capital Ideas at Chicago Booth. I’m Hal Weitzman, and with me to discuss the issue is another expert panel. John Heaton is the Joseph L.
Gidwitz Professor of Finance and Deputy Dean for
Faculty at Chicago Booth, an expert on asset pricing,
portfolio allocation, and time series econometrics, his teaching focuses on
the practical problems facing investors and institutions. Toby Moskowitz is the Fama Family Professor of Finance at Chicago Booth, an authority on asset
pricing, portfolio choice, risk sharing and market efficiency. He also knows a fair bit about the business and analytics of sport, which is showcased in his
best-selling book, Scorecasting: The Hidden Influences
Behind How Sports Are Played and Games Are Won. Steve Brodsky is the
Chief Executive Officer of Spot Trading, a
proprietary trading firm that specializes in equities,
equity options and futures. He was formerly president
of the parent company of the Chicago Stock Exchange, the exchange’s chief financial officer, and a managing director
at a private equity firm. Eric Budish is an Associate
Professor of Economics at Chicago Booth. A former Goldman Sachs analyst, he researches auction and matching markets and the design of market institutions. Panel, welcome to The Big Question. I want to start by deciding what we’re actually talking about. Toby Moskowitz, let me start with you. What do we mean when we
say high-frequency trading? – Well, that’s a good question. I think what I think it means is, like you said in the opening, algorithmic trading at
incredibly high speeds. Now with technology, that’s
become in milliseconds. Previously, it would
have been on the order of minutes or seconds. But I think the idea is that they’re essentially creating markets, something we’ll probably talk about later. But I think that’s what people mean by high-frequency trading. It’s just the speed of
getting in and getting out at lightning-fast speeds. – Okay, and when we say
lightning-fast speeds, Steve Brodksy, you’re
actually in the industry. How fast is lightning-fast? – Very, very fast. We are talking milliseconds
to microseconds. I do believe at some level, though, just in the nature of trying to define it, it’s become so pervasive
in the marketplace that at some level, the market has moved
a bit past the debate. I mean, the tools that we’re, I’m sure, going to talk about today, if you’re a large market maker, a large proprietary trading firm, these technologies persist
throughout your organization. It’s part of being in the industry. So to talk about, to try to
define it as any one strategy, I think that’s where
people have difficulty. It’s more of a technology,
it’s a technique. It’s really the way the markets function. – There’s not a trading strategy, per se. It’s just the technology
that uses the speed– – That underpins the marketplace, certainly from a
liquidity-providing perspective. – Okay, and that’s
important when we discuss who’s to blame for particular
problems in the market. Eric Budish, how important
is high-frequency trading nowadays to financial markets? – So high-frequency trading
is commonly attributed as constituting more than half of volume across a wide variety of markets. So, data’s a little hard
to come by at GetGo, which is a Chicago-based
high-frequency trading firm, recently had to file public documents with the Securities and
Exchange Commission. So you can get a detailed
glimpse of one firm. And they alone constitute
a five-to-ten percent of volume across a wide
variety of asset classes, stocks, currencies, funds. And just to Steve’s point, I would argue that
high-frequency trading firms… First of all, they’re trading
at incredibly fast speeds, at lightning-fast speeds. So put into context, blinking
your eye is 400 milliseconds, so 4/10ths of a second. So when we’re talking about
milliseconds or microseconds, which are millionths of a second, we’re talking about unfathomably
small amounts of time. And they’re also sophisticated
information technology firms. And when we think of developments in financial markets over
the last few decades, there’s a lot of evidence to suggest that the developments in information technology have benefited fundamental
investors in lots of ways, that IT has been very
healthy and very good for the function and performance
of financial markets. But some aspects of the speed race, and I’m sure we’ll discuss this in our discussion, the speed race component of
high-frequency trading, right? I take your millisecond… Take your millisecond latency and try to be just ever-so-slightly faster than you, I’ll argue will have some
negative consequences. – Let’s talk about how it’s
changed the markets, then. John Heaton, what effect has this ever-fasting trading
had on the markets? – Well, I think that’s
open to debate, frankly. We’ve certainly seen an increase
in liquidity in the markets in the last many decades. – So that’s just more orders coming in. – One argument is that
the high-frequency trading is part of it, but it could be just the increase in technology. So if we just define high-frequency trading
as use of technology and almost by definition
it’s had a massive impact in the markets, in terms
of increasing liquidity. So I think it’s open to debate whether it’s had a major impact on liquidity in the markets, per se, independent of the technology itself. The other thing in terms
of thinking about size, if you thought about fees or profits in the trading environment, standard mutual funds would dwarf this industry many times over. So in terms of importance in developments, I think it’s open to debate. – So we shouldn’t exaggerate it. One of the claims or concerns
that’s been raised is that exchanges are forever sort of pandering to
high-frequency traders. Building expensive facilities where they can locate
their black box computers right next to the exchanges, match an engine and get
the fastest possible trading times, and that kind of thing. The allegation being that there’s
sort of a two-tier market. The high-frequency traders
have privileged access, either to information or
to certain kinds of trades, and the sort of mom-and-pop
investors are left out. Steve, how is that criticism
viewed within the industry? – I think it’s important to acknowledge where we’ve come from. We can’t look at the
technology developments isolated from the regulatory developments. All of this came out of an environment where you’ve had a specialist system, a floor-based exchange model. You had very little transparency into what was going on in the marketplace. You had high fees. You had slower turnaround
times for the retail investor. All of these things were in place when these changes occurred. So I think it’s important
to have that in mind when you start talking
about think like colocation. Firms colocate in exchange facilities to be near the servers to generate the types of latency
that we’ve talked about. – And they pay huge amounts
of money for the privilege. – They pay, and what’s
different about it today than certainly years ago, I could have gotten the
same privilege ten years ago if I paid $2 million for a
New York Stock Exchange seat. And I would have been limited in terms of my access at that point. What occurs today is if, yes, you are willing to invest in technology, the exchange, by rule, has
to treat everyone equally, and this is important also
when we talk about colocation. For opponents of colocation,
if you remove it, all that’s gonna happen is that the most valuable piece of real estate is gonna be the person who owns the building next door to the exchange. So I think the good thing about the way it’s structured
today is it’s regulated. People have… The exchanges are responsible for making sure that
it’s fair across members. And fairness isn’t necessarily that everyone gets the same… Everyone has the same opportunity to– guarantee of making the
same amount of money. Fairness is about access
and making the investments. And it is no doubt expensive and part of the business model, but it’s fundamentally different than it had been many years ago. So to look at it in isolation I think is a little bit flawed and I
think important to acknowledge why we’ve come to where we are today. – Toby Moskowitz, are
the markets better today than they were before
we had this technology? – You know, again, it’s hard to say. You can make arguments in either case. I mean, one argument saying
that markets are better because of this is
liquidity may have improved. I don’t know if that’s caused
by this kind of trading or a consequence of technology and improved liquidity. And there’s some evidence to suggest that maybe that’s the case. But it’s also the case that
maybe it causes other problems. And I think, for
instance, some of the work that Eric’s been doing is
just scratching the surface on some of these issues. And I think before we have
any policy debates about this, there’s an awful lot of
evidence that we need to gather. We don’t know much about it yet. – Well that was a nice segue
into Eric Budish’s research because, Eric, you’ve
recently put out a paper about high-frequency trading, about the markets that enable
this kind of technology to flourish and be successful. So, give us your kind of judgment on whether the markets are better now than they were in the past, and if there are any downsides. – Great question. So first thing’s first. To Steve’s point, no nostalgia for financial markets in the 1990s or, more generally, pre the IT revolution, pre the information technology revolution. That said, my research suggests that current financial market design is flawed in an important way. And that flaw is that trading
occurs in continuous time. What I mean by continuous time is that investors, algorithms
can trade at literally any instant of the trading day, from the time that markets
open at 9:30 in the morning to the time that markets close at 4 o’clock in the afternoon. And continuous-time trading is
problematic for two reasons. The first reason is that
continuous-time trading builds in a speed race. When trading is in continuous time, any time there is a news event, so if the Federal Reserve announces it’s not gonna taper bond purchases, or some stock price goes up and that has implications
for other stock prices, any time there’s a news event, market participants are gonna race to react to that news event. And if I react a
millisecond or a microsecond or a picosecond faster than you do, I win the right to trade on
the basis of that information. So number one, continuous-time trading builds in a speed race. And number two, my research shows that the speed race has a
lot of negative consequences for fundamental investors. So we can go through this
in some depth if you like, but the speed race, number one, we show that it harms
fundamental investors in reducing liquidity. Number two, we argue that the speed race is destabilizing for financial markets, makes the markets more
vulnerable to events like the Flash Crash or
the Facebook IPO debacle. And number three, and put this off to a corner if you like, the speed race is bad PR. It makes financial markets look bad. It might undermine investor confidence. That might also be important
from a regulatory perspective. Although, I would put that
after liquidity and stability in terms of importance. – Okay, let’s get into
those in a little bit. Just let me press you on those
two points just very briefly. So, the markets are
more volatile because of the kind of trading environment
that we have, is that right? – Algorithms acting at
millisecond or microsecond speeds is potentially destabilizing
for financial markets. So, when I think of volatility, I think our stock prices
on a day-to-day basis moving around a lot. When I think of stability, I have in mind extreme events like the Flash Crash or when NASDAQ’s computers couldn’t… – Went down. – Went out for three hours this summer, or current keep up with
the surge of orders during the Facebook IPO. – So the stability of
market is affected by it. Toby Moskowitz? – I think that the issue’s not… I mean, Eric’s research I
think does show some of that, but I think, generally, the
evidence is not quite clear. We had a student here
last year, Brian Weller, who was working on this
in different markets in commodities futures. And there, what he found
is actually flash crashes typically occur in the absence
of high-frequency traders, than when they’re removed from the market. Now I don’t know whether you want to blame them for that as well. It’s hard to blame them for
being there and causing things as well as when you remove
them, they cause things. I think they are unpopular politically. – But actually, isn’t that the point? Because in the old system, bad as it was, there was always a
market maker who’s forced to make a market, even when
things were disastrous. – There’s also a second side to this, which is maybe, even if we do blame them for some of these rare events, and I’m not sure that’s even clear, the question is are there other
benefits on the other side? So, for instance, you
may be able to tolerate the occasional additional flash crash, if most of the time, 99% of the time, you get better liquidity,
maybe less volatility, maybe all kinds of other things that we haven’t quite documented
across all these markets. So I think the issue is open to debate. – John Heaton? – Yeah, I was just gonna
say very similar things to what Toby was just saying. I mean, I just wanna
talk about Eric’s work, Eric is one of my colleagues but independent of that,
it’s a wonderful paper. And it does point out these potential problems with
continuous trading. I would be a little careful,
more careful than Eric, about saying that the magnitude of the effects that he’s finding, I think that’s open to debate. Certainly, I think we
need empirical evidence. What’s interesting is that we might see exchanges experimenting. So, for example, there’s
private foreign exchange markets that are doing not
exactly what Eric wants. They are doing some batch
processing of orders. The other interesting part
in Eric’ paper, by the way, is beyond just limiting trading is coming up with the
appropriate mechanisms designed– – Right, so let’s come to that. – Which is a really interesting part of what he’s coming up with. What they’re doing is just randomly assigning ordering
within batches of time. It’ll be interesting to see
what’ll happen in those markets as we add these different features. So I think an interesting… We’re gonna talk about
regulation at some point. An interesting thing would be to allow the markets to experiment and to see with these
different mechanisms. I think Eric has an
interesting one to try. But I’d be careful about magnitudes. I agree with Toby, we have
to be a little careful before we draw too many conclusions from this one piece of theory, however good it is. – Eric, I’ll let you come back on that. A couple of points I’d like to make. Number one is, is I wouldn’t… I think of continuous
trading as the culprit, not high-frequency
traders as the culprits. So a lot of the public policy debate seems oriented around a debate of whether high-frequency
trading is good or evil. I think that’s the wrong debate. The debate we should be having is whether continuous-time trading is
the right market design. Number two, on magnitudes, our work does get at the magnitudes for one piece of the puzzle, and I think with millisecond-level data directly from exchanges. Where what we are able
to document empirically is the amount of money at stake in the arms race for particular kinds of latency arbitrage trades. – Hang on, so tell us what you’re saying about latency arbitrage trades,
just explain what that is. – So say there are… Say there are two securities that are very highly correlated to each other. So in the paper, we use two securities that track the S&P 500 Index, one of which trades in Chicago and one of which trades in New York. Now, to the human eye, these securities’ prices are gonna move almost in perfect lockstep, which they should, because the securities are very highly correlated to each other. They attract the same index. So, to the human eye,
the securities’ prices go roughly like that. But when you zoom in to the
high-frequency-level data, so to millisecond-level data that we purchase from exchanges, the price path is a lot choppier. Really, prices kinda go like this, where one security’s price moves and then the other security’s price– – And traders can make money
from that anomaly there. – And that’s the key is that this choppiness creates
arbitrage opportunities, where if the prices are
supposed to go like this, but really what happens
is first this one moves and then this one follows. There’s a momentary arbitrage opportunity, buying the cheap one and
selling the expensive one. And there’s a lot of money to be made from those kinds of
arbitrage opportunities. So in our paper, we show that for this one opportunity alone… So, S&P 500 arbitrage between a futures contract that trades in Chicago and an ETF, and exchange-traded fund, that trades in New York, the annual sums at stake are
on the order of $75 million. And that’s just one pair of securities that’s highly correlated. Of course, if you teach a finance class, the correlation structure of the market is the very foundation
of asset pricing theory. Correlations are
everywhere, and this is just one example of a pair of
securities that’s correlated. So I grant that that’s one number, but it’s suggestive of large
sums of dollars at stake. – Steve Brodsky? – Couple things on that. First of all, I enjoyed the paper. I’m the only one who’s
not a professor here, but I enjoyed it nonetheless. And I do appreciate the fact that, exactly as you said, it
was an analysis based on not the right or wrong of one type of liquidity versus another, but, simply, is the
market structure correct or optimal? Just a couple of points. Do you think the fact
that arbitrage exists is, that’s just a reality of
whenever you have two products, two correlated products trading together, and that’s the role that
liquidity providers play. The reason an S&P 500 Index ETF is a product that’s marketable is because there are firms out there who are compressing that
arbitrage opportunity. So there is a utility to that function and it will always exist in various forms. It happens to exist today
in a low-latent capacity. Eric can talk about
the continuous markets. That’s a function of just competitive markets in and of itself. We have 13 or 14 exchanges. We have multiple dark pools. We have an environment where a good idea will be backed by investors,
order flow providers. There are ways to express, certainly, all these types of models which are being experimented
with in various forms. The complication is that when you have these competitive markets, and you’re trying to design
a system where you say this market in now locked
up for a period of time, be it a second, whatever increment, to the extent it’s
slower than the increment that the other markets are
trading the same product, it complicates things for firms like ours. I am, to the extent that we were able to all lay down our weapons and no longer engage in the arms race, I am the first one at the peace table. The reality, though,
is that we’re competing against other firms who won’t. That gets to your prisoner’s dilemma. And there will always be markets that cater to those type of firms. So we’re constantly locked
in this competitive battle. I think that’s a good thing. I think the competition has
created tremendous choice for investors, including,
ultimately, the retail investor. So I’m a proponent of that. My struggle is just how, say, an auction or a break
of the continuous market would function in that ecosystem. – I understand. Well, Eric, seeing as we’ve sort of obliquely referred to it several times, let’s talk about what you
think the solution is, the “peace at last”
that Steve referred to. Tell us a bit about your alternative to the current continuous market. So we’re proposing that
financial exchanges, instead of using a
continuous-time limit order book, which is the predominant
market design today, use what we call frequent batch auctions. And a batch auction is
a very standard auction that actually has its roots
in work by Milton Friedman 50 years ago in the design
of treasury auctions. And we’re proposing that batch auctions be conducted at very frequent
but discreet time intervals, such as once per second or
once per hundred milliseconds. So the market structure we’re proposing is that for each stock or for each, each futures contract, each
foreign exchange contract, etc., once per second, there is an auction to determine the price for that stock. And in the auction, investors
submit bids and asks, desire to buy, desire to sell. Market-making firms also
submit bids and asks the way they currently do in
the limit order book system, and the market clears
where supply meets demand. – Gathers up all those
orders and every second it– – Exactly, it gathers up or batches a full second worth of orders, then computes based on
that second of activity where does supply meet demand? And then anybody that’s
willing to pay more than the market-clearing price, or sell for less than the
market-clearing price, gets to trade at the
market-clearing price. It’s a uniform price. – So what problems does that resolve? – So, batching solves two problems. One benefit of batching is that by moving from a continuous-time design to a discreet-time design. So from continuous to once per second, you eliminate the incentive to spend substantial amounts of money
on tiny speed advantages. So, to Steve’s point, you’re solving the arms race. This would save Steve a lot of money in the sense that he wouldn’t have to spend substantial resources
to be a millisecond faster or a microsecond faster
than he was yesterday. – It’d probably cost him more money. – Well, we can debate that. I think that’s an interesting debate. – So let’s just put– Toby, what is that concern there? – Well, no, I was just saying that there’s a reason Steve wants
to spend that money, right, which is he’s presumably
making more than it costs. So it’s gonna cost him more
than it’s gonna save him, and other people like him. – Maybe there are two, I
think, critical points. I mean, that feature of trading today has become more commoditized. There are more firms able to spend money on this type of technology. There’s just a limit
to how far this can go and you described in your
research, you hit a plateau. And then, really, it’s
just a zero sum game between market makers. And I think that’s evidenced by, you see firms like GetGo who made significant amounts of
money several years ago. Now you can publicly
see, really, they’re the one of the representative
firms losing money. Where when things are
moved from my perspective, are how do you apply the technology to other things you do well, be it analytics, be it valuation, be it access to customer order flow, if it’s providing liquidity to customers. – So is your point there would still be a technology arms race, just in different areas? In my mind, if you look
back time immemorial, there’s been this tension between market maker rights and
obligations and speed. If you wanna slow things down, you give market makers certain levels of benefits. A specialist, you get to see every order that comes to the exchange. A market maker, you get to
participate in a certain number– You favor one participant
versus the other, that will slow it down. That will bring stability
to the marketplace. But at a cost, be it transparency, be it explicit cost. – Eric, just to be clear, who would actually institute this rule? Would it be regulators? Should it be the exchanges themselves? Who do you envisage
would put this in place? – So the cleanest way to implement this would be via a regulator, such as the Securities
and Exchange Commission. And I’m not a pro-regulation
or anti-regulation guy. The current market structure
is a product of regulation. And I would argue that
frequent batch auctions is a smarter regulation than
the continuous-time framework that’s currently the
product of regulation. – But this is still a
regulatory intervention in what is a freer market,
right, John Heaton? – The reason you need a
regulatory intervention is to allow coordination across markets. Because if one market does this, you could see a migration
to another market. I mean, another interesting
issue of coordination is across securities. I mean, something… We all talk about liquidity but really when we’re talking about the high-frequency trading, it’s mostly in large cap stocks, what we really think
as very liquid stocks. I’m not exactly sure how this mechanism helps or hurts, say, for someone who wants
to trade in small cap stocks. And how to coordinate the
auctions across the stocks would be an interesting thing. I’m not sure if Eric’s thought about it. I think he’s thinking about it. It’d be an interesting situation. I mean, I am the last person
to advocate regulation, but to implement what Eric has in mind, I think you would need, you know, a national regulatory
body to implement it. – Toby Moskowitz, would that
be a step forward in your mind if we regulated to batch
auctions in this way? – Again, I’m gonna take the cop-out, which is it depends. I think we’re still gathering evidence. I think Eric in his paper
has some very good arguments. I think, you know, that
may be a better system. I kind of like what
John was saying earlier, which is it’d be nice to let the exchanges experiment a little bit. I think Eric’s proposal
is extremely interesting, and I think it does solve
some potential problems. We don’t know, necessarily, what the unintended consequences might be. This would be a way to find out. I think if we let the exchanges
experiment a little bit, that would be terrific. – That’d be better than
having it imposed on them? – I think, you know, any… Well, at least for a while, right? Let’s try some different things. I think that would be interesting. – So, just to Toby’s comment, two things: one is that there are current regulatory
impediments to exchanges experimenting with
frequent batch auctions. Frequent batch auctions would
potentially be in conflict with a regulation imposed
about 10 years ago by the SEC called National Market System, or NMS, which imposes continuous trading, or implicitly assumes continuous trading. So number one, there are impediments to exchanges experimenting with this idea. And number two is I’d like to see individual companies, or listed companies, be able to elect for
frequent batch auctions over the current market structure. A lot of… I’ve heard industry participants say that they hear CEOs of
listed companies say, “My stock trades in a casino. Why does my stock need to
trade at a millisecond?” – They also complain about shorting. – They complain about all sorts of stuff. But I like the idea of
giving listed companies the– Yeah, “Why does my stock go down?” So I like the idea of
allowing listed companies to elect a different market structure, but the current regulatory
environment would prohibit that. There’s a regulatory assumption in favor of continuous trading. It’s as if the regulation assumes that retail and fundamental investors, what they really want
is immediate execution. – I wanna come to Steve. Is this the peace that
you’ve been praying for? – I do think that what I’m struggling a bit
to reconcile in my own mind, and again, I preface it by saying that I would like it to work, is that the second you
have two market models trading the same security, or
heavily correlated security, you inevitably inject race
and speed into the equation. So if the goal is to
eliminate that as an input, therefore, all of us can invest
more in providing liquidity, and therefore, either narrowing
spreads or increasing depth, I’m not sure it gets there unless there’s a very large regulatory mandate. And I’m talking about a
central limit order book for every correlated product. – And presumably, that would have to be international as well, wouldn’t it? – Certainly, you could
start small and just do, say, the products that you studied in your proposal where you have an SEC-regulated set of equities, a CFTC-regulated set of futures who are heavily correlated,
actively arbitraged, and really subject to the
arms race that you described. Unless they’re trading
on the same platform, and that would involve getting into the corporate structures of both the Chicago Mercantile Exchange and the New York Stock Exchange. That’s a complexity,
that’s a really big hurdle. My concern would be unless you’re willing to go all the way to that, which I think is a difficult process, I’m exposed to being
held up in an auction. – Alright, Eric, let me give
you the final word, briefly. – Thank you. So, to Steve’s point, if I can bring this
debate with my research to where a panel of experts
says this is a good idea. We agree that continuous-time
markets have problems and that frequent batch auctions
is a good idea in theory, but there is regulatory complexities with implementing that idea, then I’ll consider that
a tremendous victory as a research paper. – Okay, wonderful. I hope today we’ve at
least started to do that. But unfortunately, on
that discussion of speed, our time is up. My thanks very much to our panel: John Heaton, Toby Moskowitz,
Steve Brodsky and Eric Budish. For more research,
analysis and commentary, including much more analysis
of Eric Budish’s work, visit us online at And join us again next time
for another Big Question. Goodbye! (piano music)

5 thoughts on “Is high frequency trading good for financial markets?

  1. Interesting video, thanks for posting. Here are some additional thoughts on the same topic:

    Feel free to comment. Thanks!

  2. I wouldn’t be this happy if it wasn’t for ( ) who has literally changed my life by keeping me away from all the financial issues that used to haunt me a lot. It not only solved my financial issues but also the time commitment thing by working on its own without me being the part of it.

  3. People who have made this =>> ( ) have done years of research in this field before making this app and it is quite evident from the results that it produces from the very first day of use. I can say all that because I am using it myself and I have never seen anything like this before. So hop on and try this app right now because you’ll never get a better one.

  4. FTD is a highly recommended trading app (Details Here : ) for all people out there who are trying so hard in binary trading but fail to get positive output from it. The app stands out from the rest of the pile because of its user-friendly nature and the services it provides to the user.

Leave a Reply

Your email address will not be published. Required fields are marked *