How To Profit From Planetary-Scale Computing 178
An anonymous reader writes "MIT physicist Alex Wissner-Gross and mathematician Cameron Freer have devised a technique for exploiting geographic location in high-frequency trading, reports FastCompany. From the article: 'We view this work as one of the first serious, credible justifications for covering the planet's surface with computers. [...] We've perhaps identified a new type of natural resources that sovereignties might take advantage of.' Physicist and hedge-fund manager Jean-Philippe Bouchaud says, 'This shows that the technological arms race to extract every penny from high-frequency mechanical arbitrage will soon reach its ultimate limits.'"
Another way to look at this: (Score:5, Interesting)
And to what extent is this latest proposal, while apparently to do with the distance between exchanges, also actually about putting resources into jurisdictions which have perhaps more elastic definitions of what constitutes legal trading?
On previous form, this will probably get moderated troll or flamebait. But it's actually two questions that I have never had adequately answered, except for the usual "you wouldn't understand" from the traders. If I, a graduate systems developer with further education in economics, can't understand them, what's the betting that our elected representatives can?
Increases liquidity at what cost? (Score:5, Interesting)
Some of the more involved trading strategies exploit price fluctuations between separate exchanges: traders construct complex automated financial instruments designed to seek out and exploit price differences between a range of different shares or commodities on these exchanges. The uncertainty of price movements means that individual transactions cannot guarantee a profit, but firms can make steady profits by making millions of transactions each day.
Maybe we should be exploring cheaper ways to create market liquidity without allowing firms to siphon off profits through pure arbitrage.
Highly Amoral (Score:5, Interesting)
HFT is done by the greediest scum of the earth. It is an approach that is highly instable and can do tremendous damage. It is high time this practice is outlawed. Considering that fast stock trading does not produce anything, but only serves to shuffle money around, tolerating such a destabilization risk is completely unacceptable.
Personally, I would add a mandatory random delay in the 15-30 minute range to each stock transaction. Or maybe even a few hours. This would curb speculation, while at the same time beneficial effects, like a company getting money to invest from an IPO would still work.
Wont work in Dark Pools (Score:4, Interesting)
Re:Limits? (Score:5, Interesting)
These guys go too far. One of these days we'll have botnets doing trading with funds from sniffed credit/debit info. They could even pay back what they took... then profits get dumped anonymously to campaign funds. Botnets do get free-speech rights don't they?? (they may have an opinion on capital gains taxes, or want to own broadcast stations)
If it makes anyone feel better, call that pile of computers a bank and lend it some "government" money
Trading in those strange mortgage death futures is too risky, botnet futures are the new thing.
Cylons and Skynet Terminators will have their own electronic religion making them tax exempt.
Re:Another way to look at this: (Score:3, Interesting)
Once long ago there was "a href="http://dssresources.com/history/sshistory.html"> vis-a-calc. Who would of thought today we'd be in the mess we are in.
Once long ago there was real and imminent fear that mutual self destruction would occur, and almost did, because the Nuclear C&C systems act out commands fast. Humans were inserted to cool things off.
Wow, now the Wall Street(s) have wired the financial & economic system together with less safeguards of global meltdown when the spreadsheets (now huge programs) start to ping pong (like in Forest Gump in china [youtube.com]) the markets. Its cool --- as long as you keep your eye on the ball.
However societies can't now rely on inserting humans into the chain. Anyhow Stock & Money Traders are not .mil hardy nor accountable like .mil
I cite May 6, 2010 [wikipedia.org] --- b !=m --- who programmed that billions of stocks could be sold without 2 person authorization?
I cite Jerome O'Hara, and George Perez [wikipedia.org], who worked on programatically cooking the system for Bernie Madoff
Re:Increases liquidity at what cost? (Score:5, Interesting)
A) Require 1/2 hour averaging for all trades. This will stop most arbitraging on two accounts: 1) it stops exploiting split second inefficiencies that can only be spotted by computers, 2) creates doubt to which price one is actually paying.
B) Tax all automated computer trades at 1%. Takes the profit motive out of computerized trading.
C) Charge all revocations of unused (non-expired) puts and calls a flat fee. This is to prevent flooding the market with option trades that people have no expectation of completing.
D) Tax profits made by short term traders at a higher rate than long term holder. I propose having several rates for capital gains based on how long a person holds a stock. Example (illustrative only) Less than two weeks @ 50%, less than six months@35%, 1 year @ 33%, 5years @25%, 10 years @10%, greater than 10 years @0%.
The problem isn't liquidity. Never was. Market is plenty liquid at 1/2 hour intervals. Low volume stocks need lower liquidity than high volume stock. The problem is exploitation of timing at split second intervals which can only be accomplished by computers, and has no basis in fundamental market principles. The goal should be to limit trades to people who actually hold stocks as investments, not in people making money off market fluctuations.
Re:Highly Amoral (Score:5, Interesting)
If HFT were to be legislatively controlled, it seems to me the most obvious way to do it would be by modifying the long and short term capital gains taxes to create a progressive system: the longer you hold the asset before taking the capital gain, the less tax you pay. If you had to pay 99% tax on a gain resulting from possessing an asset for less than 1 minute things would be a lot different.
This is not to say that I favor that solution, it's just one that occurs to me. I think there's a solution that doesn't require the use of force. If I were the CEO of a publicly company, I would not want to be listed on an exchange that allows HFT. If I were an amateur investor in stocks, I would not want to invest in companies listed on an exchange that allows HFT. As a result, there's clearly a market for a 'natural' exchange as opposed to one that is 'on steroids'.
Re:Another way to look at this: (Score:3, Interesting)
And to what extent is this latest proposal, while apparently to do with the distance between exchanges, also actually about putting resources into jurisdictions which have perhaps more elastic definitions of what constitutes legal trading?
Lolwut? Did you RTFA?
I'm making an educated guess, but I'd say the answer to your question is "zero extent."
If you want to trade on [exchange] you have to play by [exchange]'s rules.
Basing yourself in another jurisdiction will not keep [exchange] from locking you out for bad behavior.
If I, a graduate systems developer with further education in economics, can't understand them, what's the betting that our elected representatives can?
Knowing how stock markets physically work isn't necessarily the kind of thing they teach you in economics.
Maybe you should keep furthering your education and audit some relevant business classes.
Re:Increases liquidity at what cost? (Score:5, Interesting)
I don't get why we can't even just have one-day ticks. Every day an order book accumulates, and at 5PM the exchange executes everything at the price that generates the most volume. Priority is given to sellers who offer the lowest price and buyers who offer the highest price. Within a price orders are executed in random order.
The book is kept secret until after all trades are settled. So, you can't see if the price is trending towards a price you like and then put in a bunch of sells for 0.01 to get ahead of the line - if you put in that price you might just find your trades executing at that price.
With such a system ordinary investors can compete with investing houses. Bad news means that everybody loses out at the same time, and the insiders don't have nearly the same advantage (getting news 15 minutes early can make a HUGE difference today). You could even make the trade settlement time midnight or something like that so that it is well after the business day so that last-minute news has more time to get around.
You wouldn't need so many market-makers and other forms of arbitrage since the total daily volume of a stock will tend to guarantee that there will always be buyers and sellers. Market makers could still fill a niche in low-volume stocks making sure that there are always buy and sell orders in the book.
You could even go a step further and execute trades once per week/month/etc - that would start to make investing more of a long-term thing and less of exploiting market psychology..
So much misinformation (Score:2, Interesting)
Let me begin by saying that I'm posting anonymously because I'm a professional in the financial industry.
First, I'll explain the inter-exchange arbitrage being used here: it comes in a few forms. Imagine the Philadelphia exchange has orders on its book for a share of Apple Inc stock (ticker AAPL) as follows: 200 to buy for $308.12 and 300 to sell at $308.14. The New York exchange has 500 to buy for $308.13 and 100 to sell at $308.14 . There's nothing anyone can do to make a profit.
Then things change. Perhaps someone does a trade in New York, taking out the orders on the sell side of the book. New York is now having 150 to buy at $308.15 and 700 to sell at $308.16. Since $308.15 > $308.14, someone can buy in Philadelphia and sell in New York, making a penny profit on 150 shares.
The first company to notice this and send the necessary electronic order messages makes $1.50. Repeat as necessary.
More complex versions of this same trade involve the same stock traded in different currencies, requiring a currency hedge at the same time. But the idea remains the same.
This sort of arbitrage has ALWAYS taken place in the markets. The HF traders don't do anything differently from what has historically been done, in this or really any other strategy. They just do it faster. Market makers have historically held small positions, by the way. Other posters who assume they used to hold millions in inventory are exaggerating. I don't see HF as any kind of slimy behavior, and I say this as someone whose firm is on the other side (i.e. the HF firms take money from us, lots of it, every day).
Regulations designed specifically to remove the profits of HF firms are neither wise nor beneficial. Regulations should concentrate on ensuring the markets are fair, efficient and reliable. If those regulations have the emergent property of killing the HF firms, that's fine.
It's worth noting that fairness, efficiency and reliability are sometimes competing goals. To take a low-frequency example, insider trading is illegal in the USA but not in some other places, and economists generally consider that information propagation is more efficient in those other places, at the obvious cost of diminished fairness.
The valid concerns about HF trading are not the profits made by those firms (which are anyway estimated to have a ceiling of $21B industry-wide by a well-known academic paper -- a fraction of investment banking profits). The valid concerns about HF are market stability and fairness.
I view fairness as having decreased with the advent of HF, in that HF firms are now likelier to prevent Joe Blow from trading "on the bid" or "on the offer". That's a cost to Joe Blow. On the other hand, bid-offer spreads have narrowed considerably, due mainly to the savage competition in HF. On the whole, Joe Blow therefore buys or sells at a better price than he used to. I therefore think the loss in fairness has been more than compensated by this increase in efficiency.
The situation is a little vaguer for big traders like my firm, where we trade so many shares that we have to worry about being "detected" and having markets move against us. But even for us I think the tradeoff is worth it.
Now let's consider market stability. Complex dynamic systems are subject to occasional wild behavior. This is even true of ones involving humans, as with the Dutch Tulip craze or the recent real-estate bubble. Dynamic systems run by machines can enter undesirable states faster than humans can usefully respond.
May 6 2010 is cited as an example, though I'll note that the crash happened over many minutes, not in mere milliseconds, and therefore was actually well within the range of human reaction times. Many human traders, some at our firm, made big profits off those using machines. This in itself serves as an excellent correction and lesson to those relying too much machines to trade, and has helped put humans back in the loop at many places, I'm sure.
Re:Increases liquidity at what cost? (Score:4, Interesting)
All of the potential issues you have raised could be fixed using just two minor tweaks to the grandparent's suggestion:
- Trade every hour, on the hour (this is unaffected by timezones)
- Trade 24/7 (it's done by computers, after all, so it makes zero sense for an exchange to be "open" at only certain times of day)
- Every trade is logically split into some small unit (share, dollar, whatever), and randomized, or an algorithm is used that effectively trades simultaneously in a completely fair way.
Done.
Problem solved.
The current situation is not necessary and the alternative proposed by the gp has no real issues except that a bunch of HFT traders would lose their jobs. Boo-friggin-hoo.
Re:Another way to look at this: (Score:3, Interesting)
The Scandinavian case he mentioned is when the small bunch beat the robo traders, they go to jail.
When the robo traders screw up big time, the stock exchanges roll back their losses.
http://www.bloomberg.com/news/2010-05-06/nasdaq-to-cancel-trades-of-stocks-moving-more-than-60-in-market-plunge.html [bloomberg.com]
They claim it was "someone hits the wrong button", but I think most slashdotters can guess what really happened :). Bug in computer programs - failure to handle corner cases or exceptions.
I understand a rollback if it's a bug in Nasdaq's programs that caused the problem. But if it's a bug in a trader's robo-program, the trader should eat the loss.
Otherwise, why shouldn't they cancel my trades if I ever lose more than 60% on Nasdaq? Or poker?
That's where the real profit is coming from - these people cheat and they get away with it.
Otherwise all those billions they make would be lost during the times they screw up.
It's their metascheme that is profitable. Not their actual schemes.
Metascheme= win, keep your winnings. Lose big? Stock exchange rolls back your losses. Lose really big? Government bailout.
Even I could make money like that. I just have a conscience and I wouldn't be able to fool myself on what I'm actually doing.
All that talk of liquidity and making the market more efficient is BULLSHIT. How the fuck is it more efficient if you have to do massive bailouts?
Re:Silly flat-earthers! (Score:3, Interesting)
A tunnel bored directly between London and New York would be even faster and require less cooling. Only two points intersecting the center would be competitive with my Earth Chord Trading Tunnels!
Your idea intrigues me and I would like to subscribe to your burrito delivery service [idlewords.com].
Yes, lag is exploitable (Score:3, Interesting)
I'd thought of this a few months ago, after reading the detailed report on the 2010 flash crash. [wikipedia.org] Speed of light lag wasn't quite an issue, but it was close. Stocks are mostly traded in New York, while options are traded in Chicago. Round trip time between the two is at least 7ms. That's exploitable. Lag isn't just for video gamers any more.
Unfortunately, this isn't a joke. There is now special purpose hardware for high frequency trading. [stoneridgetechnology.com] General purpose computers aren't fast enough for high frequency trading. This 1U device contains FPGAs, and custom trading algorithms are written in Matlab, compiled into Verilog, and loaded into the FPGAs.
Vendors are advertising "8 microsecond average latency, wire to application". [redlinetrading.com] Not milliseconds, microseconds.
Well, I will give a tiny bit of credit (Score:3, Interesting)
It actually does produce something and that is a high deal of liquidity. Something that you have in the market right now that is nice is that you can buy or sell any stock any time you want. Because of all this day trading, HFT, there is always stock being shuffled around, and in rather substantial amounts, so you can always get in or out of a stock when you please. That is a benefit.
However it is not a benefit that is worth the instability HFT causes. We need to fix the system, either with a time based tax or random delays or something. But we do need to recognize that it does provide a benefit, just not one that outweighs the cost.
Re:Time for a rant... (Score:2, Interesting)