Contrary to what some might think, I'm not opposed to investment bank and bankers, I'm just opposed to the current crop of banks and bankers. But even more than usurious middle men, the guys who really make my head explode are the flat-out speculators--they serve, even on their best days, no useful purpose whatsoever. And thanks to the recent stock market computer glitch that caused the Dow to temporarily dive ten percent, speculators known as high-frequency traders are finally getting noticed (italics mine):
Depending on whose estimates you believe, high-frequency traders account for 40 to 70 percent of all trading on every stock market in the country. Some of the biggest players trade more than a billion shares a day.
These are short-term bets. Very short. The founder of Tradebot, in Kansas City, Mo., told students in 2008 that his firm typically held stocks for 11 seconds. Tradebot, one of the biggest high-frequency traders around, had not had a losing day in four years, he said.
But some in Washington wonder if ordinary investors will pay a price for this sort of lightning-quick trading. Unlike old-fashioned specialists on the New York Stock Exchange, who are obligated to stay in the market whether it is rising or falling, high-frequency traders can walk away at any time.
And this is probably one of the most unproductive uses of mathematicians and computer scientists--it probably runs neck and neck with online porn:
The Tradeworx computers get price quotes from the exchanges, decide how to trade, complete a risk analysis and generate a buy or sell order -- in 20 microseconds.
The computers trade in and out of individual stocks, indexes and exchange-traded funds, or E.T.F.'s, all day long. Mr. Narang, for the most part, has no idea which stocks Tradeworx is buying or selling.
Showing a computer chart to a visitor, Mr. Narang zeroes in on one stock that had recently been a winner for the firm. Which stock? Mr. Narang clicks on the chart to bring up the ticker symbol: NETL. What's that? Mr. Narang clicks a few more times and answers slowly: "NetLogic Microsystems." He shrugs. "Never heard of it," he says.
This is not entrepreneurial investing. This isn't even an attempt to find an investor dumber than you and fleece him. Instead, very bright people are spending far too much time trying to game the system--it's no different than the first wave of day traders who attempted to quickly capitalize on minute differences between markets before everyone else could catch up. There is absolutely no useful economic activity here. And here's how they
earnextract their money:
Most of these firms typically hold onto stocks for a few seconds, minutes or hours and usually end the day with little or no position in the market. Their profits come in slivers of a penny, but they can reap those incremental rewards over and over, all day long.
What all high-frequency traders love is volatility -- lots of it. "It was like shooting fish in the barrel in 2008. Any dummy who tried to do a high-frequency strategy back then could make money," said Manoj Narang, the founder of Tradeworx.
A 0.1% transaction tax would put a stake through the heart of these parasites, while not overly burdening long-term investors. It would also mean that many of these very technically specialized people--whose educations were subsidized by their fellow citizens*--would actually do something productive: hell, porn would be a step sideways, if not up.
*A while back, I calculated that the cost of training various science PhDs, in terms of cost to the federal government, was comparable to the cost of training a special forces operative (if not greater). As far as I'm concerned, someone who uses that training only for pecuniary gain and also imposes a societal cost is no better than those Blackwater mercenary scum.
As a computer scientist, I don't really want to stop high-frequency traders.
Sure, they extract money from minute changes in stock price. However, technology developed to achieve this is pretty interesting. Investing banks for a long time were drivers of large multicore realtime systems (pretty much no other industry needs them) and highly parallel algorithms.
I don't know if the benefits from these developments outweight harm from market crashes, though.
large multicore realtime systems (pretty much no other industry needs them)
Perhaps you have heard of this organization called the National Weather Service. They issue a variety of products, among which are severe weather bulletins (tornado warnings, severe thunderstorm warnings, flash flood warnings, etc.) that have to be issued in real time based on highly localized weather conditions. I'd say that the NWS, as well as its counterparts in many other countries, needs this kind of thing. Maybe the banks were funding research into this kind of thing, for the reasons you describe. I agree with Mad Mike that the HFT application has negative social value: these firms are capturing profits that otherwise would have gone to retail traders. I can't prove that they are actually changing closing prices on stocks, but there is at best zero value to society for someone to hold a stock position for all of eleven seconds.
The proposal would not necessarily eliminate HFT altogether. There are many occasions where markets move by 0.1% or more (that's only about 10 Dow points) in a matter of minutes. It would certainly reduce the amount of HFT going on, and that would be a good thing. Many retail investors would not notice a big difference, because we already have to pay trading commissions which are close to or more than that amount.
The founder of Tradebot boasted about "not having a losing day in four years" doesn't mention it is not from trading, buying low and selling high. He actually loses money in that regard. His matching engine is paid generously by exchanges for computer driven liquidity. The problem as we all saw on May 6 was these HFTs get to choose when to play pulling liquidity when most needed. They're getting paid well for not doing their job. Wall Street Reform needs to address these market whores.
High-speed trading also allows for cheating in the form of a modern version of beating the wire.
If giant fund ABC puts in an order to buy at market price 20,000,000 shares of XYZ stock, and I can delay that order by milliseconds, I can slip in ahead of theirs my minor fund's order to buy 200,000 shares of XYZ. After ABC's purchase triggers a market response, that is, pushing up the price of XYZ, when the price stops rising I can sell my 200,000 shares for a nice profit.
How can this be stopped? I suspect it cannot. The stock market was designed by gamblers, not by engineers.
I suspect that these HFT folks---the really big players, such as Goldman Sachs---also have the ability to use HFT to drive the market up or down as they wish for their own benefit. The May 6 "glitch", where the Dow fell 700 points in 10 seconds, certainly worked out well for the big boys. They stepped in when the Dow was 1000 points down, bought up huge amounts of stock at bargain prices, and, in minutes, the market rebounded 600 points. "Greater" wall street made off with billions taken from "lesser" wall street. If a crime is committed so rapidly you can't see it, is it still a crime? Gee! How did my pockets become empty?
On a good day, speculators are useful. Example: item "should" sell at 100, seller offers at 90 but highest bid is 75. If I step in and buy at 90, seller gets price 15 closer to "fair" price. [And people who are basing their actions on the market price have a more accurate "signal" which with to work.] How do we know item "should" sell at 100? Because price eventually gets there and I sell for 10 profit. On a "bad" day, speculators are a problem: I think item should be worth 120, but is really only worth 100. I pay 110. I have cause the price to move away from "correct" price. Even if I find a bigger fool who pays 120, I have distorted the market. But the fool's loss will be more than my profit, so on average "bad" speculators will lose, good speculators win. The problem is bubbles, where even good speculators can add to the problem and cause systematic harm. Thus I see no problem with a transaction tax to build up a bailout fund for the next crash.
High frequency trading is a different problem, but I suspect a bailout fund tax would be justified there too. Front-running is just a crime, but very difficult to prove.
What I can't figure out is why some of the transactions were canceled when things went off the rails this month. Everything I have read about the issue is that it was largely a bunch of high speed trading systems that reacted badly to a trading slow down.
I can certainly understand halting trading and letting everybody readjust to avoid a total meltdown, but if your clever computer program decided to offload all of your shares of Accenture for $0.01 each, tough. Live by the sword and die by the sword. All that ends up being is a massive transfer of wealth away from parasites who shave coins for a living.
The stock market was designed by gamblers, not by engineers.
This. One might as well complain about the dilapidated state of the Three-Card Monte industry.
Yikes, where to start...
From comment #2:
> I can't prove that they are actually changing closing prices on
> stocks, but there is at best zero value to society for someone to
> hold a stock position for all of eleven seconds.
High functioning capital markets are of value to society. Liquidity
added on the buy side and the sell side adds value to capital
markets regardless of how long the position is held.
> The proposal would not necessarily eliminate HFT altogether. There
> are many occasions where markets move by 0.1% or more (that's only
> about 10 Dow points) in a matter of minutes. It would certainly
> reduce the amount of HFT going on, and that would be a good thing.
It would reduce the amount of liquidity and would increase spreads.
It would make markets more susceptible to manipulation by traders
willing to throw size around.
BTW, nobody has shown actual evidence that HFT is a bad thing. Hand
wavy "those stupid HFTs caused this!" is not evidence.
> He actually loses money in that regard. His matching engine is paid
> generously by exchanges for computer driven liquidity. The problem
> as we all saw on May 6 was these HFTs get to choose when to play
> pulling liquidity when most needed. They're getting paid well for
> not doing their job.
Liquidity is valuable to markets, they are thus willing to pay for
it. Nobody has an obligation to add liquidity, traders add liquidity
when they believe that a price at which they can bid or offer will be
profitable (taking liquidity add rebate or take fee into account).
Markets work best and are the least susceptible to manipulation when
there is a large number of individual participants.
When massive sell size takes out all bids noone is going to stand
in front of that freight train (not even specialists). Get rid of
market orders and install circuit breakers (or better yet, futures
style limits), eliminate erronious trades and the corresponding breaks
and presto, we avoid the irrationality of the selloff and are left
with the rational (still significant) sell off.
> I suspect that these HFT folks---the really big players, such as
> Goldman Sachs---also have the ability to use HFT to drive the market
> up or down as they wish for their own benefit.
Most HF trading is market making and is not intended to move the
market. You can't really move a market without applying size at an
opportune moment. There very well could be people doing that, but it
is not HFT.
> The May 6 "glitch", where the Dow fell 700 points in 10 seconds,
> certainly worked out well for the big boys. They stepped in when the
> Dow was 1000 points down, bought up huge amounts of stock at bargain
> prices, and, in minutes, the market rebounded 600 points.
Both Tradebot and Tradeworx (probably others but we don't know) pulled
out of the market during all of that because trading wasn't rational
and markets weren't operating properly. Also, significant number of
trades were broken. Anyone in the industry knew that trades would be
broken but nobody knew at what prices thus firms who traded through
the whole episode did not have a good idea what their holdings were
over night (and wondered about potential margin calls if one leg of
lots of their trades was broken.)
There are huge numbers of floor traders, market makers, specialists
and day traders who have become largely irrelevant due to HFT.
They're pissed and they're using every opportunity to paint HFT as the
evil that causes every percieved problem in the market.
The bottom line is that there has never been a better functioning
market than ours for the investor and this is the result of HFT. It's
not perfect but it's good.
Contrary to what some might think, I'm not opposed to investment bank and bankers ...
er, what? No, really, WHAT!?!? Have you been in a cave, smoking crack, for the last two years?
ok, I admit, investment banking isn't all bad. But if investment bankers cannot understand why many people feel they are, they are in deep trouble.
BTW, nobody has shown actual evidence that HFT is a bad thing. Hand wavy "those stupid HFTs caused this!" is not evidence.
I suppose it's possible that humans just decided based on no new information that Accenture and Sam Adams had both lost 99.99% of their value, but it smells like a broken feedback loop to me.
I understand the benefits of the liquidity that traders like this add, but we can't lose sight of the fact that these are basically automated feedback control systems where the control variables are not especially well understood. This sort of thing has me torn because my background puts me on both sides of the fence: I studied economics and computer engineering with an emphasis on signal processing.
One side of me says that there's a something to be gained from having fast moving counterparties available for just about any trade. The other side of me knows that feeding anything that looks like a day's stock price information into any sort of digital control system will absolutely make it crap the bed on a regular basis. Once you have enough of them doing it to actually noticeably affect the price, it's bound to happen.
Anybody who has written sensible looking code to control a target tracking camera and had a small, unexpected movement in the target cause the camera to jump off the table knows how this works. That's how feedback systems act if they're not dealing with well characterized systems. It's an important life lesson that everybody who dabbles in these sorts of things in the physical world learns early on. My experience is that most of my friends in finance... well... they just haven't had that experience.
I will say this: The fact that these systems hold on to stocks for seconds instead of minutes or hours is a good thing. If they operated slower than that, these sorts of issues would be far worse. And of course, if your computer offloaded all of your stock for pennies, suck it up. Let's not create yet another moral hazard issue.
"Perhaps you have heard of this organization called the National Weather Service. They issue a variety of products, among which are severe weather bulletins (tornado warnings, severe thunderstorm warnings, flash flood warnings, etc.) that have to be issued in real time based on highly localized weather conditions."
No, large supercomputer systems are nothing unusual. However, large supercomputer _realtime_ systems are unusual.
By 'realtime' I mean not 'reacts almost immediately' but 'reacts always within M milliseconds'. Check this article: http://en.wikipedia.org/wiki/Real-time_computing for a good explanation.
> I suppose it's possible that humans just decided based on no new
> information that Accenture and Sam Adams had both lost 99.99% of
> their value, but it smells like a broken feedback loop to me.
> I understand the benefits of the liquidity that traders like this
> add, but we can't lose sight of the fact that these are basically
> automated feedback control systems where the control variables are
> not especially well understood. This sort of thing has me torn
> because my background puts me on both sides of the fence: I studied
> economics and computer engineering with an emphasis on signal
Humans didn't decide that the value of several stocks dropped to
zero. Significant selling triggered stops which turned in to market
orders with no price limit whatsoever. It was an irrational situation
that a short trading halt would have prevented.
As long as there are both fundamental traders and technical traders
things will work out. If technical traders or programs make prices
diverge from fundamentals, value traders will see the bargain and
bring things back in line. Exchange rules are in comment period now
that will create halts if things get out of line too quickly and give
time to value traders to spot the bargains and jump on them.
I am so tired of hearing people bitch about HFT. A trader tax wouldn't just affect HF traders, it would affect me, my family, and a good chunk of my friends.
You really shouldn't try to talk about things you have NO IDEA about. Your just not qualified to touch on this matter.