As long as there has been a transfer of money, there has always been a market for faster access to information and faster use of that information. Its only in the last ten years that the speed is so fast it has to be done electronically.
There's a autobiography of a stock trader written in the early 1900s called Reminiscence of a Stock Operator where, in 1907/1908, the author is complaining that the time to execute a trade is too long. Specifically, he would notice a price movement on the tape, tell his runner to go (for example) buy 100 at $60, but by the time the runner was actually able to execute, the price was at $80, hence he's getting a worse price. The solution back then was faster runners, then faster communication (such as hand signals), then computers, etc.
Wall Street is not just HFT. Different firms have different ways of trying to make money. Some are speculators, some are value investors, some are arbitrageurs, some are market makers. Some companies trade to hedge risk in, for example, commodities that they expect to buy in the future. Others trade to hedge currency risk against countries they do business with.
To claim that Wall Street used to bet on companies that build things but now only bet on companies that make money is terrible mass-market journalism at best.
I guess I'll pose my question about HFT directly: what value do you add to the economy by being in the HFT business?
Surely the markets must at some point be "liquid enough". i.e. that the companies that are supposed to actually benefit from liquid markets no longer do see any additional benefit from increased speed. If trades could only be made once per day, then traders would be very, very hesitant to invest in a new company IPO, because if there was new information or a problem, they couldn't sell it for 24 hours, at which point the price could have plummeted. Companies would have a bona fide problem raising new capital.
But does that problem really still exist if trades could only be executed once per second? per quarter-second? Billions are now being made that justify shaving a few milliseconds off of a signal -- how do you justify that in terms of benefit to the companies?
I think its on the people who don't want HFT to be regulated out of existence to convince the rest of us that it actually has value for someone other than themselves and isn't basically legalized gambling.
If you are a market participant, you benefit from tighter spreads. If you "regulate HFT out of existence," you will instead get to buy higher and sell lower.
I was assuming that the desired course of action was to enter or exit a position, rather than to enter or exit a position with moderately high probability.
Generally, the HFT player provides "impatience service" - they give you immediate execution, and charge you a fee for it.
Unfortunately, since they invest in speed, they impose this service fee on everyone: When you join the bid or the offer, you are always behind the HFT player, and as a result you will only execute if there is a huge execution, or, (more probably) if the price moves against you.
This aspect of HFT is completely fair, btw - the HFT players are taking a risk, and they should be rewarded for it.
The part that is unfair is described very well by nanex in this thread - go read it.
I know a lot of people at major investment firms and not one of them trades HFT. I'd assume HFT is an extremely small percentage of the overall market in terms of average position size.
If they are trading any stocks at all, they will be using essentially the same systems as HF traders. Any big broker-dealer ('big' in terms of quantity of orders processed) will be a high frequency trader by definition, because they aggregate large quantities of customer order flow, process it rapidly, and send the orders out in rapid succession.
it might be 1% of the average position size (and thus margin), and still account for 30% of the trades.
In "long term trading", you buy 100,000 shares and weight, locking down margin big enough for 100,000 shares. You sell it after a month.
In "hft", you buy one share, and sell it. You repeat that 100,000 times in one day. If you did this smartly, you come up with some profit, and you only need enough capital to have a margin for 1 share.
Furthermore, at the end of the moneht, the hft guy had 20 times as much volume (assuming 20 trading days in a month), which is why the exchange loves them so much (they live off volume!), and why the exchanges will never let hft be regulated out of existence.
Vast amounts of money are being spent on HFT with no benefit to the market, but only to the firms that are currently the fastest at the time. This is a pure greed move and demonstrates so much of what is wrong with much of the modern banking and investment industry.
I would be incredibly easy to come up with a workable compromise, like a throttle or minimum time an order must be maintained, but vested interests will keep that from happening until something truly disastrous happens.
It's a Red Queen Race: everyone's running as fast as they can to stay in the same place - and burning cash by the ton to do so. Red Queen races tend to be good for advancing the state of the art, but they're wasteful. And if you're trying to do something other than advance the state of the art, they're a trap.
I don't mind HFT - I use their services every time I trade and I love them.
Never understood why the quants always get blamed for so many negative situations - most of it is essentially predetermined by the directions of "financial directors" in conjunction with their lawyers (monopoly on markets/one way regulation).
That shows that volume increased. Which means higher competition between market makers resulting in smaller bid-ask spreads and higher liquidity, meaning the market can correct itself much faster.
I'm going to go ahead and qualify this next statement by saying that I know little to nothing about high frequency trading and little about finance in general, but when Wired covers technology subjects I do have in depth knowledge about it's riddled with sloppy reporting and sensationalism. This has happened enough times that I usually skip their coverage entirely and when I do read their stories I do so with a very skeptical eye.
This piece may be different, but I would have know way to know since it's outside my field. It just makes me sad that I can no longer give them the benefit of the doubt. I used to love Wired. However, as I've become more and more knowledgeable about web technology and marketing over the course of my career, the more I see them as one more sensationalistic rag doing bad science and technology reporting for the mainstream.
The problem isn't trading fast, it's trading unfairly. Not following regulations. Heaping the costs of speed on others, while reaping the benefits. We have documented hundreds of examples of clear rule violations. http://www.nanex.net/aqck/aqckIndex.html
Did you know that in 1999, during the internet bull market, the system processed just 1000 quotes/second, and today it's 1.5 million/second? Guess who pays for that? Guess who benefits? http://www.nanex.net/aqck2/3528.html
Did you know that your orders to ETrade and other retail brokers are tagged as "dumb" and sold to the highest bidder like Knight, UBS, Citadel, who then internally match your order (it never goes to the exchange, except during the flash crash). They give you the price from the slower system, but buy/sell from the faster one. The incentive to skew the two is irresistible. http://www.nanex.net/aqck2/3519.html
When you get a price-improvement of a penny on a 100 share $60,000 apple trade, they steal 99 cents - and front run some other investor, who's order is left hanging (sometimes forever). It happens in Apple more than 2,000 times a day. http://www.nanex.net/aqck2/3520.html
When you offer to buy something at a price, then yank that price before the other party can see it (speed of light), is that fair? It happens millions of times a day. http://www.nanex.net/Research/bloodbot/bloodbot.html
Great to have nanex here. I am very interested in hearing why you (and surprising number of other vendors) support windows, but not unix. I don't mean that in an accusatory sense, I'm genuinely curious if there are that many people who run automated trading models on a windows machine.
Regarding the topic on hand. When e-trade sells their order flow, aren't they obligated to provide 'best execution?' You are speaking of rebate for liquidity, correct? If NITE and BATS are both quoting the same price, then e-trade has the choice of sending to either, but if NITE has better price, they e-trade has no choice but to route the order to NITE, correct?
Rebate for liquidity is entirely different. So is retail price improvement (aka sub-penny pricing).
Retail order flow is sold to a wholesaler like Nite who matches it internally. They are allowed (best execution) to give you the price at the NBBO during a 1 second window. For less active stocks, the NBBO price can change 1,000/second or more. For all stocks, the NBBO price will be slower than the direct feed price. This latency and differential makes it a risk free trade (so long as you don't accidentally release your test code into production: see http://www.nanex.net/aqck2/3525.html)
There is no way to prove what prices existed on all markets at the time your trade was executed, meaning it's on the honor system. So this depends on your trust in Wall Street to give you the best deal.
One of the dirty little secrets is that a lot of all the easy money was made a while ago. The media's just kinda catching up now. The increased cost of playing this game means fewer and fewer firms are able to chase fewer and fewer dollars.
They're squeezing themselves out of the market. It's like any other thing in the market: as people see profits they get into a market and accept a lower profit target until the market stabilizes on the marginal cost to produce the good. In this case the good is liquidity for the market.
Just give it time to play out. The "problem", if there is one, is kind of taking care of itself. If you want to direct your energy at a real problem, focus on the old-boys-network of getting stuff done in banking: commoditize the crap out of the financial sector's business by breaking down the moats and walls that allow high finance to run scams like the dotcom bubble, the housing bubble, and manipulating the LIBOR.
Wall Street firms are not "addicted" to faster trading: they NEED faster trading to keep up with the competition. That's not news. For as long as financial markets have existed, market players have been in a never-ending technological arms race, constantly seeking to get information and execute trades faster, hopefully gaining a temporary advantage until the competition catches up.
Leaving aside the hyped-up nature of this Wired article, there is a legitimate debate, ongoing, as to whether automated high-frequency trading is beneficial or detrimental to society. On one side of the debate, there are economists, regulators, and market participants who sincerely believe markets normally are, or otherwise tend toward, equilibrium, so allowing more trading motivated by profit, regardless of whether it's automated or not, makes markets more efficient and stable, which in their view is beneficial to society.
On the other side of the debate, there are many critics (e.g., the late Benoit Mandlebrot, Nassim Nicholas Taleb, economist Steve Keen, Keynes biographer Lord Skidlesky, etc.) who believe markets are never in equilibrium, and can unexpectedly 'spin out of control' just like other highly complex, dynamic systems, so it's a terrible idea to have market stability relying on the second-to-second behavior of automated trading systems that can make decisions and act on them much, much faster than humans could ever react to a market debacle.
In my view, the second group is right, but if anyone here disagrees, I'd love to hear their thoughts.
Crashes happened in 1929, when everything was done on paper, and even long before that, when transactions took even longer to complete. Everything is faster these days, but it is not qualitatively different.
"Benefit to society" is not, and never was, an optimization criterion to how markets are being run. Forget the rose tinted economics text books. The way markets have been run in the last 200 years or so (at least) is to benefit the big players. There are some benefits to society, such as a standardized way of raising money by companies that lets the investors remain liquid, as well as some standard forms of insurance (all derivatives are, one way or another, a form of insurance and COMPLETELY EQUIVALENTLY, a gamble, depending on how you view life).
The problem is that it's a wild west now. There are a lot of laws on the books that are not being observed, or are being observed only against some players, but not others.
There's a regulation that says any order you put into the market must have behind it the intention to execute. Now, please tell me how 10,000 order/cancel/order/cancel events per second (often with the same price but different quantities) can satisfy that regulation? I had been in algo trading for 7 years, and I'm not aware of a single plausible explanation, let alone one that would work for 50 firms, 250 days a year, 4 hours a day.
Modern trading is a wild west casino. Regulators have been long ago captured, and the few that weren't appear to be either powerless, incompetent, corrupt or worse. (see: madoff w.r.t markopolos, the total lack of action against mfg/corzine - there are hundreds of examples; read zerohedge or denninger for new examples daily)
beagle3: you make several great points, all of which, AFAICT, are in agreement with the second view: financial markets are inherently unstable -- and yes, regulatory capture, rampant fraud, and 'wild west casino' behavior are all factors contributing to instability -- speed just magnifies their impact. Many of the people in the second camp have made the same points as you in their calls for increased regulatory supervision and enforcement.
Note that the years preceding the Crash of 1929, commonly referred to as the "Roaring 20's," were also years of 'wild west' behavior. Consider that the CEO of Chase National Bank, one Albert Wiggin, was caught secretly selling short his bank's stock in 1929, at the same time he was committing his bank's money to buying the bank's shares, and he did this through a Canadian shell company to avoid US taxes.[1] Or consider that the Chairman of the NY Stock Exchange, one Richard Whitney, was sent to jail at Sing-Sing for embezzlement in 1938.[2]
The forced separation of commercial and investment banks by Glass-Steagall in 1933, the creation of the SEC in 1934, and the enactment of many investor- and market-protection laws and regulations via the Investment Company act of 1940, were all a response to the Crash of 1929 and the resulting Great Depression. The underlying assumption behind all these efforts was that Wall Street could not be left to its own devices.
The country subsequently had decades of relative financial market stability: there were no crashes like 1929 until the regulatory apparatus was dismantled in the 1980's and 1990's, leading to the crisis of 2008.
> all of which, AFAICT, are in agreement with the second view:
As you presented it, the speed is an important factor in the second view - I disagree. Proper regulations (assuming they are observed and diligently enforced) should be time-scale independent.
Glass Steagall kept the system-as-a-whole safe -- e.g., it would have greatly reduced the effect of the "subprime" crisis (though sadly, blame lays just as much with the Clinton administration, which means GlassSteagall might not have helped). But is mostly unrelated to HFT - e.g. Knight Capital, which lost $440M million in the course of 45 minutes last week, would not have been subject to Glass Steagall.
The fact that everything seemed stable does not indicate everything was indeed stable - e.g. Madoff was doing his thing for the last 17 years, and things blew up only because many people wanted to get their money back in the same time (as a result of their losses in other places).
There is an interesting result from complex system theory, which is that when such a system breaks down, it always looks like there is a specific culprit / event, but the truth is that any one of hundreds of qualitatively different events would have started the ball rolling.
All of the big banks (and most governments, including the US, Japan, UK) are bankrupt - or at least would be, if they were run like a business, no one dares admit it (although greece and ireland were already forced to, spain italy and portugal will need to admit that sooner than later).
We will hopefully know the truth about this crisis in the future. It is unlikely that we will have a reasonable idea before 20 more years have passed (and especially the statute of criminal violations....), though.
beagle3: yes, in an ideal world, regulations should be time-scale independent, and they should be observed and diligently enforced. The problem is that human judgment -- which is still necessary for enforcing regulations in the face of market participants who're constantly trying to game them -- is not time-scale independent.
As I presented it in my first comment above, the second side believes "it's a terrible idea to have market stability relying on the second-to-second behavior of automated trading systems that can make decisions and act on them much, much faster than humans could ever react to a market debacle." (Emphasis added.)
Here, let me save the HFT apologists from their post:
liquidity market-makers benefits retail investors handwave handwave march of technology fat guy in a suit waving his fingers around liquidity liquidity liquidity paying 1/8 of a cent for a trade obviously harms the retail investor more than the entire market crashing when algorithms fight with each other handwave handwave market making you don't understand the importance of liquidity market maker liquidity liquidity no fat guys in suits.
There's nothing intrinsic about markets that says they have to move towards faster and faster trading. For example, markets could decide that they now want to clear every second, on the second. That's larger than the RTT around the globe and so global markets could remain in sync.
The spread would increase marginally but the upside is that PhDs could focus on making their algorithms smarter, not just faster.
As someone who knows just about nothing about HFT(theoretically this article's target market) I don't see anything wrong with it. Care to elaborate about why its "bad on so many levels"?
tl;dr: We missed out on Facebook because of regulation. regulation lowers market efficiency, markets have always evolved, HFT improves pricing and order fulfillment
Like seemingly everything wired writes these days this article is nonsense, written from the perspective of an agenda. Wired regularly comes out with these neo-luddite articles. The reality is, software bugs cause money. Knight was punished for their "sins".
Meanwhile I profit because HFT makes the market more efficient for me. I know this with certainty because I trade derivatives (options spreads) and I've done so in areas where there is no HFT and where there is a lot of HFT. My orders go better at closer prices when there's HFT going on. I've experienced the additional market liquidity first hand, and the underlying is shockingly stable for the nature of the business, so I don't think anyone could say that HFT is destabilizing the stock price.
At this point people might be tempted to shout "anecdote" but the thing is, economics and reason also support this view. Thinly traded markets are simply less price efficient because there are less people analyzing price and thus willing to buy or sell at a little better price. Computers are efficient at this (I know almost all the counter parties to my trades are computers, it is really easy to see because they react instantaneously to my actions.)
Finally let me point something out. Facebooks IPO has been pretty poor, while Google's was great. Google was able to go public when it was still in its high growth phase. This meant that the public was able to invest in google and make very high returns. Facebook, however, grew up in a different regulatory environment, especially after 2008 (Which was itself a result of regulation in the first place, though for political reasons we're fed a constant stream of lies to the contrary) ... in this regulatory environment going public is much more difficult.
Thus, instead of you and me profiting from buying facebook stock in 2006, where we would be up 100-fold by now, only venture capitalists and large companies were allowed to participate in that. This is really a shame. In the 1990s a company like facebook would have gone public much earlier. People seem to think that because there was a bubble (again government created) and some companies went bust that somehow magically "regulation" will "protect" people. It doesn't, it just hurts people. We all missed out on Facebook, and now facebook was forced into a situation where its stock is going to suck wind for awhile-- hurting the entire startup sector.
But it always starts with articles like this-- nonsense to scare people about HFT, this creates the political support for legislation (which always favors certain interests, not the least of which are the politicians who wouldn't do it otherwise) at the expense of society.
>Meanwhile I profit because HFT makes the market more efficient for me.
High frequency trading does not increase liquidity in a way that directly benefits anyone but high frequency traders. Being able to liquidate assets in a fraction of a second less does not benefit the normal investor. Not even getting into the volatility it adds.
This is utter crap.
As long as there has been a transfer of money, there has always been a market for faster access to information and faster use of that information. Its only in the last ten years that the speed is so fast it has to be done electronically.
There's a autobiography of a stock trader written in the early 1900s called Reminiscence of a Stock Operator where, in 1907/1908, the author is complaining that the time to execute a trade is too long. Specifically, he would notice a price movement on the tape, tell his runner to go (for example) buy 100 at $60, but by the time the runner was actually able to execute, the price was at $80, hence he's getting a worse price. The solution back then was faster runners, then faster communication (such as hand signals), then computers, etc.
Wall Street is not just HFT. Different firms have different ways of trying to make money. Some are speculators, some are value investors, some are arbitrageurs, some are market makers. Some companies trade to hedge risk in, for example, commodities that they expect to buy in the future. Others trade to hedge currency risk against countries they do business with.
To claim that Wall Street used to bet on companies that build things but now only bet on companies that make money is terrible mass-market journalism at best.