Aussie Telco Lays New Fiber For Microsecond Trading Boost 212
schliz writes "Australian data center and telecommunications provider Vocus has installed two new underwater fiber links across the Sydney Harbor in a bid for the lowest connection latency between the city's financial district and the Australian Securities Exchange's recently opened data center, north of the CBD. The project involved 1.6 kilometers of custom, 312-core single-mode optical fiber cable, and was expected to deliver a route that is 400 meters shorter than existing links. RTFA for pretty installation photos."
Not surprising (Score:5, Informative)
I have a friend who is a developer for a hedge fund where they pay him and a few others north of $250k each per year (it is NYC) to try and and shave milliseconds off transactions. They spend big bucks trying anything to reduce a transaction time from 4ms to 3ms or lower.
Re:looks like a.. (Score:3, Informative)
This is a privately owned cable.
Re:Not surprising (Score:4, Informative)
"The problem with the world is eventually you run into people like Thatcher". - MRe_nl
All money is other people's money: my expenditure is your income and vice versa.
Thatcher was a whip used on the lower and middle-classes in the UK. Her continued praising leaves me baffled.
Re:Not surprising (Score:5, Informative)
Re:If the NSA submarine cuts the line to tap it . (Score:3, Informative)
. . . does that add more latency to the line? Can you measure actual versus expected latency to see if your undersea lines have been tapped?
No - you can use something like a 1:99 optical splitter so they'll barely notice the signal drop, and will add about 5mm of optical fiber into the line, so they won't notice any additional latency (less than 20 picoseconds). Then run your 1% signal into an optical amplifier, say an EDFA [wikipedia.org], and snoop to your hearts content.
Re:High-Frequency Trading (Score:5, Informative)
Not really. Stock-exchanges should either just enforce once-a-minute matching (with lottery determining which trades to fulfill if there's several takers), or they should just set some minimal fee for every non-filled order which stands for less than a minute, 1% of the order-value would be plenty, probably even 0.1% of the order-value would be enough to stop HFT dead.
They're taking steps, some of them, but it's baby-steps. For example the Norwegian stock-exchange is adding a $0.01 fee for every trade - for those traders who file more than 70 orders for every *one* that goes trough, only orders which are withdrawn before 1 second has passed, are counted.
This is an *extremely* timid step. Make it $1, one in ten, and 1 minute rather than 1 second, and we're talking.
Re:High-Frequency Trading (Score:1, Informative)
I used to work for an algo-trader, and so AC. I should point out that I was tech-support, so not a trader, not an algo programmer and generally didn't get anywhere close to the "real" knowledge of how the company or it's various parts worked in detail. However...
The one I used to work for identified the "race to the bottom" and seemed to generally avoid it. They did co-locate in some exchanges (but by no means all - there are sometimes advantages to *not* being in the exchange), and generally did some of everything you can think of. They also did vast amounts of other things at the same time, so whatever your beef is with algo-traders, they did some of it, but most of their business was doing other stuff.
IMHO, the only algo-traders causing instability are the bad ones. The good ones have algos that specifically scale back when things go crazy, but the bad ones say their being responsible because they "turn everything off when things go crazy". Turning off is probably the worst thing you can do in a crisis, and actually, the good algo traders make money out of the most volatile situations. Turning off means you cancel all your open orders, which removes liquidity from the market, which means crazy prices for things can occur. If you have lots of liquidity (even shrewd liquidity) then you may not get what you think is a great price, but it won't be an off-the-chart bad one.
So how's all this good for the world at large? Well, I'm honestly quite mixed about it, but what I do know is: ;-)
- Algo traders provide liquidity. This really does help all of us that want to invest (long term) for our pensions or just to do something other than trust banks credit interest.
- No stock exchange in history has ever been "about long term investments". They have all, without exception been about a way for companies to raise money, and for everyone else to try to make some money from time to time. If you want to invest, then go buy some stock (maybe privately - no exchange required) and sit on your investment for a few years. What happens in the meantime is irrelevant, and thus so are the algo traders. If you think you're being smart by setting a stop-loss or whatever, then all you're doing is a speculative trade, and if you're doing that, you're no different from the algo traders, except they're better at it than you. Also, if your chosen stock doesn't pay dividends by policy, then you have to wonder what kind of trade this is - is it investment or speculation?
- Algo traders spend *vast* sums of money on technology. This is also "liquidity provision", but at Dell rather than the stock exchange. They push tech to new levels, which as we've seen from space exploration and F1, helps everyone eventually.
- Algo traders spend a lot of money on other stuff too - be it travel, food, office space, whatever. They're by no means a charity, but they do spend money in their local areas.
- Algo traders grow little-to-zero marketing or sales people. For the average techie, this probably seems like a good thing
One other thing I'd like to debunk is the myth that they hold onto things for microseconds - that's not really true (although there are times that they might). They might buy on one exchange and sell on another in almost the same moment, but that's subtly a different thing (and the electronic equivalent of buying things cheaply in another city and selling them in your own). There's no speculation in the buy/sell method, and as algo traders will tell you... don't take any risk you're not getting paid for. Buying something in the hope it'll go up and then selling it again doesn't make a great deal of money unless you hold onto it for a while, so microseconds make no sense.
Also, one last thing, for which I'll climb onto my soapbox. For anyone who's lost money on a stock exchange and blames the algo-traders: If you weren't up to the task, you shouldn't have played the game. I won't be at Wimbledon this year because whilst I can play tennis, I'm not very good at it. I might play aga