How Microwave Transmission Is Linking Financial Centers At Near-Light Speed

The L.A. Times has a short but compelling article about the state of the art (and coming state of the art) in dedicated networking technology in one of the applications where you'd expect the customers to care most about it: connecting financial trading centers. Milliseconds count, and the traders count milliseconds. From the article, one example: "[New York-based networking company] Strike, whose ranks include academics as well as former U.S. and Israeli military engineers, hoisted a 6-foot white dish on a tower rising 280 feet above the Nasdaq Stock Market's data center in Carteret, N.J., just outside New York City. Through a series of microwave towers, the dish beams market data 734 miles to the Chicago Mercantile Exchange's computer warehouse in Aurora, Ill., in 4.13 milliseconds, or about 95% of the theoretical speed of light, according to the company. Fiber-optic cables, which are made up of long strands of glass, carry data at roughly 65% of light speed."

Comments

But

[rossdee]

There was a story a few weeks ago about someone in Chicago that had a faster than light connection (when the Fed issued a statement about interest rates)

Re:But

[Anonymous Coward]

Time for somebody to put up some tall, shiny aluminum billboards on rural estate along a certain 734-mile path...

Re:But

[Z00L00K]

To avoid this the trade system should delay all transaction with a random factor between 10 and 20 minutes. That should make microtrading completely useless.

Re:But

[pepty]

Just accumulate all of the orders for one second. At the end of each second match the closest buy/sell orders to each other and execute the trades. Repeat.

Re:But

[Z00L00K]

No, the 10 to 20 minute random delay is to equalize for human traders,

Re:But

[SuricouRaven]

Ask where all the money to fund this is coming from.

HFT doesn't actually manufacture anything or provide a valuable service. It's purely exploiting the numbers for profit, out-trading slower traders. The profits made in HFT are ultimately at the expense of everyone else on the stockmarket who can't keep up. Which likely includes your pension fund.

Re:But

[Anonymous Coward]

People don't like HFT because it's profitable, and they perceive things like "it's stealing money from my pension fund". But it isn't. Your pension fund makes a trade every few weeks. Daily at most. If it's an index tracker, and most of them are, then they rejig the portfolio when an index rebalance happens - the day they decide which successful small players come into the index, and which fallen big players are now out. This happens every three months, six months or annually depending on which index it is. I promise you, your pension fund makes more money from your pension fund than HFT firms make from your pension fund.

HFT firms are generally market makers. These are people who have no particular view on the direction the market is about to move, and will give you a price to buy at, and a price to sell at. If they strike a sweet deal with the exchange, then they are obliged to do both and for a fairly high percentage of the time. If they chose, they can also just put a buy order and a sell order in exactly the same way, but the exchange won't give them the same discount.

Either way, with market making, you buy low and sell high, and try not to hold on to the thing long, since otherwise you end up with a risk that it goes up (if you've sold more than you had, known as shorting) or it goes down (if you've bought more than you've sold, like most people invest).

Therefore, the only way you make money is when people trade with you. You need to buy one and sell one a lot, and then the difference between your buy and sell price, known as the spread, is yours to keep. You therefore need two things: to have the best prices, and be at the front of the queue.

Speed is essential for both of these: if you are fast, you can give a good price, because when things turn against you, you can cancel the orders that would now hurt you. Getting to the front of the queue is obviously speed dependent.

So in practice, you end up competing against the other market makers, since you all only make money when someone trades with you. The pension fund, or whoever, actually trying to buy or sell their holdings in the market gets a much better price for doing so.

It's like in an RPG when merchants give you 50% of your loot's value, but sell it back to you for 200%. It's pretty clear where they are making their money, but you're only out of pocket if you keep buying and selling. If your wizard sells off the Stupidly Heavy Sword of Smiting because he doesn't want it, then he still pockets the gold. Likewise, when he buys the Nifty Cloak of Resist Goddam Everything, he gets the benefit from it, no matter what else the merchant buys or sells. If you sit there and keep buying and selling the same wooden bowl from the guy, then yeah, you can get rid of all your money fast. Pro tip: don't do that in the stock market either.

Obviously it's better for you if instead of the mark up being buy @ 50% / sell @ 200%, it becomes 90% / 115% (or whatever, doesn't have to be absolutely symmetrical). The HFT market makers are the dudes who let you get the better prices in the market when you / your pension buy or sell.

Re:But

[khallow]

Dealers profit from market inefficiency, inserting themselves as middlemen and profiting by pushing prices away from their efficient levels.

"Pushing prices away"? If you have a "dealer" trading away from the "efficient levels" and a "dealer" trading towards the "efficient levels", I can tell you who will have money at the end of the day and who won't.

Someone who pursues a dumb strategy is going to lose their wealth to someone who isn't.

Why not just let ultimate lenders and borrowers meet with technology, obsoleting the "market makers" raising prices because of their profit-seeking motives?

That's the stock markets in a nutshell. Yet they still end up with market makers.

Re:God knows

[EdIII]

Whatever you're smoking please share.

Improving the lives of ordinary citizens

[Anonymous Coward]

Every day Slashdot should have one story that features a technological advance which could improve the daily lives of ordinary citizens.

This isn't the one.

Re:Improving the lives of ordinary citizens

[Anonymous Coward]

Only if not used to take advantage over said ordinary citizens.

Can go somewhat faster...

[Thagg]

Sending a neutrino beam through the earth will be faster than taking the great-circle route across the surface of the earth.

Of course, one would have to send a ridiculously large number of neutrinos to be sure to have them detected, but that's just an engineering problem.

Dodgy Customers

[mrbluze]

here you'd expect the customers to care most about it:

Yeah, and what customers? Electronic trading supercomputers.

None of this makes a difference to honest trading except to ensure its loss making properties.

Re:Dodgy Customers

[Anonymous Coward]

Traders with ultra fast computers place orders at a fixed price. If no one nibbles immediately and they can't make a profit in a few milliseconds, the order is cancelled. The current system makes "traders" very rich because they "cheat" legally.

Re:Dodgy Customers

[girlintraining]

None of this makes a difference to honest trading except to ensure its loss making properties.

I think I see a flaw in your cunning argument: First, all trades are honest as long as it is not made under duress. And any macroeconomic teacher will tell you that in a trade both people get something they want, so the more trade you have, the better the economy is. People are operating under a wide variety of misconceptions regarding high speed trading, and only a few of the criticisms are valid. High speed trades mean that the value of a given stock or financial offering is much closer to the line where supply and demand cross. It results in less money being wasted either because the price is too high, or too low. What high speed trading does, in essence, is reduce the delta. Conceptualize a curved line, and then consider a number of equally-spaced rectangles under each approximating the volume of the curve. The more rectangles you have, the more accurate you can recreate that curve. High speed trading simply improves the delta of the supply and demand curves, so that the spread above and below the point where supply and demand cross is very small.

People get confused between high frequency trading and algorithmic trading. High frequency trading carries benefits for both buyer and seller. Algorithmic trading, on the other hand, can and has resulted in huge losses. Like most algorithms exposed to unexpected input, they behave erratically and in unanticipated ways, and once one algorithm goes off the rails, as it were, it can lead to a cascade failure where different financial agents within the system also see something that was unanticipated and then in turn fail. Because these algorithms control a lot of different financial products, these cascade failures can spread and crash entire markets, dozens of stocks, etc.

There are valid criticisms for algorithmic trading. I see none for high speed trading, however. And I do not know why people banter about about "dishonest" trading -- the trades themselves are public record, and only executed because the seller and buyer agreed on a price. There is no coercion or manipulation in the trade itself. The dishonesty in the system comes in over or under-valuation of a financial instrument, or from insider trading. This is external to the trade system itself, and comes from people using information not publicly available, or from deceptive accounting practices.

But here again, the algorithms themselves, nor the computers executing trades, are responsible, and using them is neither dishonest nor something that "only the rich" can afford to do; Sites like e-trade offer consumers a wide variety of tools which can execute high speed trades when various conditions in the market occur, such as the price rising above, or falling below, certain points, and these systems are available for use by the everyday person for reasonable fees. The dishonesty in the system is largely on the CEOs, senior management, and accountants, who collude to profit at others expense.

It has nothing whatsoever to do with the systems themselves, and I really wish people would stop spreading the idea that there's this mythical beast living in data centers in New York gobbling up poor people's money -- those live in the Penthouse, not the basement.

Re:Dodgy Customers

[Anonymous Coward]

Shilling hard for Goldman Sachs are we?

People get confused between high frequency trading and algorithmic trading.

A distinction without a difference. High frequency trading requires algorithmic trading otherwise it can't match the fluctuations in price.

I see none for high speed trading

Valueless arbitrage. Buyer and seller would have met if not for the middleman leeching off both.

There is no coercion or manipulation in the trade itself.

What about when the traders fuck up, why is it they can ask NASDAQ et. al. to roll the trades back but I can't?

Re: Dodgy Customers

[bazorg]

yes, but do they run linux?

"and the traders count milliseconds"

[superdude72]

The fact that the traders count milliseconds is proof that they are just scam artists. Parasites, producing nothing of value. High-speed trading is nothing but a mechanism for funneling money to the rich from everyone else.

Re:"and the traders count milliseconds"

[femtobyte]

but... but.. liquidity! Businesses need to be able to make decisions to respond to market needs on 4.13 millisecond timescales! Imagine how much economic damage there would be if the Job Creators had to wait a whole 6 milliseconds (at fiber-optic speeds) for their decisions to take effect!

Re:"and the traders count milliseconds"

[blue trane]

Quoting from Mehring's Notes for Lecture 10, in Economics of Money and Banking, Part One:

In financial theory, it is common practice to assume perfect arbitrage, and hence also
complete liquidity. Assets are assumed to trade at their fundamental value since any other price
would create an arbitrage profit opportunity. In effect, the world that the finance theorists
imagine is a world in which the VBT outside spread is very very narrow, so there is no room and
no need for dealers. In the real world, the outside spread is quite wide, dealers offer prices inside
that spread but the prices can deviate very far from fundamental value. That is the world Fischer
Black was talking about in his infamous presidential address to the American Finance
Association when he said that he thought markets were efficient, meaning price was usually
within a factor of two of true value.

We can understand what Black is saying by referring back to the Treynor model.
Suppose that fundamental value is the price that dealers would quote if their inventories were
exactly zero, so they are not exposed to any price risk. The Treynor model then shows how
market making by dealers pushes price away from fundamental value, on one side or another, by
more or less depending on the size of the outside spread and the dealer’s maximum long and
short position limits. Standard asset price theory abstracts from this effect, in effect treating the
outside spread as collapsed around fundamental value, so there is no need for dealers. Some
markets are close approximations to this, but others are not; some times are close approximations
to this, but others are not.

If efficiency is completely arbitrary, then the prices that high-frequency traders even out, which the post I was responding to hailed as the legitimate foundation of capitalism, are based on what? They're just pushing prices around to make a profit. Why is that "legitimate"?

Dealers profit from the general lack of liquidity. Wouldn't it be better if a "public option" existed, which provided liquidity without a profit motive? You wouldn't have to regulate dealers; just provide another option which worked in the public interest, pushing prices lower than dealers alone will.

Blimp in Middle attack?

[runeghost]

Step 1: Intercept the transmisson Step 2: ??? Step 3: Profit!

Re:Nothing "near" about it

[Pseudonym Authority]

Nope, nope, nope. The speed of light is c, the speed of light in various materials is denoted as factors of c (0.65c or 0.99999c).

News? Well, interesting perhaps...

[rusty0101]

Considering the Nova episode 'making things faster' with David Pouge as the host presented this over a month ago, and they have production lead time delayse amounting to months, I'm with the people suggesting that this is a repeat. And hardly qualifies as 'news'.

That said, cool.

Also noted in the Nova episode was that the physical path that the fiber-optic route takes is going to be longer than the path that the microwave route takes.

If they are not doing it already, I would suspect that the next step will be to move the repeater electronics up to the microwave dishes at the intermediary locations. My experience with telcom is that this is usually housed in the shelter at the base of the towers, however if they can locate it with the dishes it will eliminate a path distance of at least twice the sum of the elevations of the dishes, divided by the velocity factor of the media they use to connect those, (roughly 1 for dry waveguide, and usually between .6 and .9 for different varieties of co-ax media.) It's not a lot, but if they are looking to eliminate every possible delay, that's got to be in their plans.

Re:News? Well, interesting perhaps...

[Anonymous Coward]

Already done some years ago. Dish has redundant modules on the back that only need power and cat 5 for ip down. Cross connect across tower legs on same tower to the next dish and go

Re:Nothing "near" about it

[Mr Z]

Raises the question, maybe, but it certainly does not beg the question.

In any case, the speed of light in fiber optics is dominated by the glass or plastic, not any air that might be somehow still be in the fiber. So far as I know, that quantity is zero or close enough at least. For fiber optics to work, you need total internal reflection. To get total internal reflection over a decent range of angles (so that you can actually bend your fiber optic cable), you needs a sufficiently high index of refraction. It turns out that the higher the index of refraction, the slower the speed of light in the medium.

Re:God knows

[toygeek]

Its the latest literary craze: eat two cans of alphabet soup and then forcefully vomit it onto a high contrast flat surface, take a high resolution photo of it and then run it through OCR from 1999. Tada! Instant classics the whole family can enjoy.

Re:Nothing "near" about it

[davester666]

the next guy up will put up a series of dead-straight tubes, that they can create a vacuum in, then pass the microwaves through the tubes, just to get 97% of the speed of light.

and of course, he will also be able to claim a patent on transmitting the internet through a series of tubes.

Local silicon beats microwave everytime

[Anonymous Coward]

The high speed trading field is bein geviscerated by the placement of FPGA based devices on fiber leaving the building. There's no point ot a "high speed link", with all the insanity involved in that technology, when you can program a sophisticated monitor and sales control tool right outside the stock exchange.

Now, the center that analyzes the history of such data and builds your models of how to respond can be in another state and pre-program the FPGA. But these devices are already being used: It took me twenty minutes to stop laughing at a job interview with one such company: I was unsure just which of my talents they wee interested in until I noticed a display of such devices, and realized they wanted me to assist in their migration to a much, much cheaper part of the country for all their modeling and analysis tools, and just leave a box or two of Nallatech FPGA cards at the New York Stock Exchange.

It was amazing how fast they put the NDA in front of me when I explained the implications to their core business model, and started tying it to the repercussions of moving *away* from the human contacts that provide back channel, essentially insider knowledge, when the analysis boxes are no longer in the data center with your own staff on site. My speculative "if you need to tune these models manually, you need to secure the information, and who is allowed to change that model? how do you manage the provenance of such orders" discussion apparently set off alarm bells that urged them to get me under an NDA, whether or not they ever made an offer. Life got really adventuresome when I refused to sign an NDA without an offer on the table.

Mod parent up: it's called VELOCITY FACTOR, folks!

[storkus]

For some place that's supposed to be for nerds who, unlike me, finished college, this discussion is embarrassing. Parent post and 1 or 2 other posts have it right, and this is something that every radio guy knows as well.

Wikipedia references: http://en.wikipedia.org/wiki/Velocity_factor
More general discussion with heavy math: http://en.wikipedia.org/wiki/Group_velocity
The reason for it all: http://en.wikipedia.org/wiki/Refractive_index
This is straight from the horse's mouth: http://www.corning.com/WorkArea/downloadasset.aspx?id=39403

All this effort for parsitism

[dbIII]

It looks like another symptom of why China is going to kick our ass. All this effort put into getting crumbs from what other people do instead of doing something tangible.

Re:All this effort for parsitism

[blue trane]

What does China do again? Provide cheap labor? 3D printing will eliminate that advantage.

Hot Pocket in the Middle Attack

[Calavar]

Step 1: Rent an apartment across the street from the stock exchange
Step 2: Cook delicious hot pockets in the microwave while the interference messes with the traders' signals
Step 3: ???
Step 4: Eat delicious hot pockets

Re:Mod parent up: it's called VELOCITY FACTOR, fol

[Nemyst]

Light speed, without any qualification after it, is generally accepted to mean c, celeritas. Adding "in a vacuum" is a pointless waste of space and only serves to confuse people who do not know, nor care, about the difference.

It's amusing though how there's always at least one pedant "correcting" the article or summary in each and every article with that short-form term in it.

Re:Nothing "near" about it

[drkim]

no, the speed of light is always the speed of light. the speed up light in a medium is ... the speed... of the light.

You are correct sir.

It's those lazy-ass science folks that always forget to append:

"The speed of light ...in a vacuum. "

Re:What's next for fiber?

[putaro]

Line of sight communications means you need relays somewhere to go around the curve of the earth. We already cut huge latencies off intercontinental links by going from satellite links to fiber optics.

Re:Nothing "near" about it

[Trogre]

The speed of light in a vacuum is c. Otherwise you are correct.

Look up cherenkov radiation for a cool example of particles exceeding the speed of light in the local medium.

Re:Nothing "near" about it

[slick7]

That's why the big boys lay dedicated fiber.

When you say "dedicated" is that with or without NSA knowledge?

Re:Old News

[Mister Liberty]

progress comes from the financial world

Twisted definition of 'progress'.
Where have you been the last couple of years?

Re: Old News

[CurryCamel]

Cutting down the ping from 6.04ms to 4.13ms. How is this not progress?