Hacker News new | past | comments | ask | show | jobs | submit login
The Most Expensive Place To Put A Computer (larrycheng.com)
37 points by lwc123 on Sept 16, 2009 | hide | past | web | favorite | 34 comments



These high frequency traders typify the harmful Wall St. attitude with which America's twenty-somethings are enamored. It really needs to stop.

Let's compare business to football.

Some people become experts in training, coaching and playing football. Others manage teams and learn how to spot talent and work with dynamics between players to create winning combinations. They're trying to raise the bar and increase the level of competition within the game.

Then there is a class of people who don't really add anything to the game, but they want to profit off of it. These are the gamblers.

High frequency traders are like meta-meta-football players. They try to profit off transactions between the gamblers.

Frankly, I don't know how they can look at themselves in the mirror. I wish they would put their minds to good use and actually solve real problems instead of meta ones.


Your post is a jumble of invalid ideas that, to me, merely expresses your contempt of modern day Wall St. You're completely entitled to your opinion, but you should be more aware of exactly who it is you are pissed off at. You certainly should not typify anyone on Wall St as "harmful," especially when you do not know exactly what they are doing.

The article does not explain what the high frequency traders, specifically, are doing. But it is by no means necessarily evil. All they are doing is two things: they get their trades into the system quickly, and they get current market conditions quickly.

If there is anything that you should quarrel about, it is that their system is automated. But, still, at the end of the day they're the ones holding the bag. If they blow up, they're going to pay for it.

If they start packaging their complicated and automated trading algorithm as a "bond" and get it AAA rated, and then start selling it to pension funds, then you've got a reason to be pissed.


Your gamblers vs. players analogy doesn't really work. The high frequency traders are playing the same game that traditional traders are playing. They just pay for a bit of an advantage in how the game is played.

Whether or not this is an unfair advantage is up for debate.


But gamblers can't put a large bet out there on the sportsbook floor and just before an interception is caught declare "Whoops! That bet isn't valid anymore".


Betfair has realtime sportbet trading


it seems like arbitrage is often presented as somewhat negative. floating exchange rates between fungible commodities (and the arbitrage opportunities that accompany them) are a feature not a bug. they allow markets to correct for fluctuations in supply and demand very quickly ensuring more efficient resource distribution.

arbitrage being available exclusively to a small group who also control membership to their own group? corporatism, and obviously bad. this problem of the financial world having insiders and outsiders has nothing to do with high speed arbitrage, that's just an illustration of it.


actually, a lot of the best high frequency trading shops out there are small, private, prop trading groups. for example, tower research capital (http://www.tower-research.com/) is a pretty successful high frequency shop that isn't a big bank. it was started from scratch by mark gorton, the guy who created limewire. there are tons of high frequency shops making tons of money who are just like this.


Fluctuations in supply over milliseconds aren't causing an inefficient distribution in resources though. There is an inefficiency here though in the extraction of wealth from the financial system without a corresponding input simply by making high-frequency transactions.

What resources were you thinking of?

The only resource I can think of that needs instantaneous variation in supply is electricity. Logistics of distribution mean that supply of most other things will only be variable on the order of 6 hours or so. Or do you mean supply of shares to be traded?

As an outsider, trading of securities and derivatives appears to be a long way from the initial goal of optimising the market and providing capital.

Thinking aloud: If trades were pegged to a single time each day (per market) how would that affect the goal of supplying capital (I'm guessing it wouldn't really)? It would reduce liquidity by quantising it in step with the market but would that be harmful?

I'm sure someone will point out to me why, beyond making traders less wealthy, that's a bad idea?


The fact that server proximity plays are role in trading advantages seems... wrong. I'm not sure that it matters, but I think most people have a general aversion to gaming the system.


it's not really gaming the system. if you have money, you can buy server space close to the exchange's computers. that may seem unfair, but that's not really unique to trading. consider manufacturing: if you have more money, you can buy a bigger factory and leverage economies of scale to make your widgets for cheaper; that's a competitive advantage.


It's a bit different. Your metaphorical factory isn't making cheaper widgets by taking tiny pieces of your competitor's widgets away from them.

I'm not arguing for or against high frequency trading, just saying that finance actually is a zero sum game unlike manufacturing.


In this case, you may be right. Generally speaking, finance can have positive effects -- even commodity speculation can be argued to positively influence long-term price stability. But it's hard to imagine how competing for low-latency server space makes the world a better place, except perhaps to piss people off enough that someone might eventually implement an official timing methodology that can't be gamed by sub-half-second latency advantages.


Finance isn't zero-sum. Better finance makes market capital allocation more efficient, which makes the pie bigger for everyone.


I've never understood how this applies to arbitrage. Without arbitrage, the parties to the transaction would capture the money lost to arbitrage, and in a modern market, the same information would be learned. If I offer to sell 1000 at $3.03 and you offer to buy at $3.04, we can split the difference. If someone else gets the information of that spread faster than we can resolve it ourselves, they don't "lubricate the capital markets" or anything. They just get our money.


Arbitrage supplies liquidity at all prices, including prices where most trades might not otherwise occur. Liquidity is required for businesses to systematically eliminate risk, therefore it is highly valued as a source of stability for most companies who deal with financial instruments.

Knight has some thoughts on systematic risk in his book Risk, Uncertainty and Profit.


Did this answer my question? I'm really open to the possibility that it's just me, but I still don't get it. I see that some kinds of arbitrage could be useful to a market.

I don't see how these microsecond-timing types of arbitrage can possibly help any market. Every last one of these trades will happen anyway. (*nitpicker's corner - of course some wouldn't happen, e.g. at the precise close of trading.)


He did answer your question. The value of arbitrage is liquidity. Liquidity is extremely valuable. Here, I'll quote Harris:

Liquidity is the ability to trade large size quickly, at low cost, when you want to trade. It is the most important characteristic of well-functioning markets.

Everyone likes liquidity. Traders like liquidity because it allows them to implement their trading strategies cheaply. Exchanges like liquidity because it attracts traders to their markets. Regulators like liquidity because liquid markets are often less volatile than illiquid ones.

Market are not magical. Just because something may be fundamentally worth X does not mean you can automatically find someone to take the other side of a trade based on that valuation.


I still don't get it. I accepted that liquidity is valuable. This specific case adds no liquidity. None. Both sides are participating in the same market, and have enough information to make the trade. They will make the trade if no one interferes. Someone else comes in as a MITM, and collects the spread. What value did they add?


Capital allocation only works towards efficiency if you remove capital from failing businesses ... but no one would prop up a failing business with billions of $USD and negate the whole process would they ...?

Also, you need people to want what's good for them. Suppose we need better food supplies; food production companies need more capital to optimise their production. But, people want nicer trainers (aka sneakers) and so we allocate more money there for growth to recoup more profit from that id-ish desire for fashion. That's not efficiently running the system that's efficiently recouping the greatest profit. People's desire for better food needs to be improved but instead the profit from the shoe sales feeds-back and more money is spent on advertising and marketing and enticing people to want slick shoes in preference to healthy food. The feed back loop rolls on and more resources are allocated to shoes still at the expense of other sectors.

Capitalism is not really about optimising production, IMO. The essence of capitalism is to maximise profit and the two ideals are not co-terminous.


well, there is a limited demand for widgets. if i make cheaper widgets, i'm taking away sales from my competitor. isn't that the same?

your observation isn't something unique to finance. capitalism, or just business, is inherently competitive. guy A wants 1 widget; your factory and my factor can't both sell our widget to him. isn't that also a zero sum game?


But making cheaper widgets can increase widget demand to an extent, right? Maybe guy B, who couldn't afford a widget before, can now buy one. By producing more widgets for the same total cost, you are increasing the amount of wealth in the world.


I think it would be more like if I get to find out that Guy A wants to buy some widgets before you get this information. I also get to find out that you're selling widgets before Guy A gets this information. I can buy widgets from you for cheap because you don't know that there is a market yet. I can sell them to Guy A for expensive because he doesn't know there are plenty of widgets to go around.

On the face of it, there's really nothing wrong with this - too bad for you and Guy A that you don't have access to the information that I do. But I think the concern is that I can use my position to control the markets, or that everyone will lose faith in the game and stop playing.


"When billions are made on millisecond-level trade throughput, having your server right next to the server that executes your trade is worth a lot more than $4,000 per square foot, it’s priceless."

It's actually not priceless. If the space is truly that valuable, the price will keep rising to the point where the profitability decreases relative to the risk. The risk inherent in this is also key. If you are buying and waiting for the uptick, and then it goes down...


While you're definitely correct, you should be aware of a literary device called "hyperbole" :)


Of course, but a key implication of the article and many of the responses was that this was somehow unfair, in that connected companies had special access to these spaces. In fact, if the ability to make billions were predicated so highly on the precise location, people would likely be willing to pay quite a bit more. If the advantage were that great, one would expect the price to be higher.


If they could all agree on a third-party timing and signing methodology, this would be a non-issue in my opinion. Gaming it? Sure, but make the fines insane, and use the actual timing (whatever it is) for auditing.


Ah, the cheat computers! Where they get a little peek at their opponent's card before betting...

High frequency trading is cheating. Ban it.


How do you to propose that? Put time limits between receiving information and acting on it? This will just increase the arbitrage possibilities.


How do you propose banning it? Somebody will always get the news first... you can't stop them for reacting to it.

Or, is it the frequency that you think should be banned?


How important is physical proximity, really? An electrical signal can travel 186 miles in a millisecond. So locate the servers in cheaper real estate 10 miles away an you lose like about .1 millisecond round-trip. If that really makes a difference in trading success something is really screwed up.


>>How important is physical proximity, really?

First off, high frequency traders are making their money off of how quickly they react to the market conditions. The quicker they react, the more money they make. In this sense, if they trade at high enough volume, it's worth paying more for a lower latency. This is, of course, assuming they are good at what they do. If they are bad at what they do, a lower latency wouldn't harm.

Second, they are in a competitive environment. If they want to make money off of obvious arbitrages that their competitors are likely engaged in as well (very likely in the forex market), they need to react first.

I assume that physical proximity reduces latency in ways other than the time it takes for photons to travel through fiber optic cable, for example router hops.

>> On: Trading success based on latency

In Forex (the currency exchange), there are sometimes very obvious arbitrage opportunities. You can take your Dollar, and trade it into Euros. Then, go from Euros to Yen. Then, go from Yen back to Dollar -- and end up with more dollars than you started with! Of course, after a certain amount of money flows through, the market will adjust itself and the arbitrage will vanish.

Being the first to react to this arbitrage allows you to realize and profit from the arbitrage before anybody else does.


The latency in the network has more to do with the networking equipment than the transmission speed. Also routing can often cause longer return times and dropped packets.

Paying the high price ensures that their transactions don't leave the building and all of the equipment in the middle is optimised for that type of operation.

The only other solution would be 10 miles of dedicated fibre through a metropolitan area. Not a cheap solution.


Worrying about how "high frequency traders" are making so much money is like worrying about VCs making so much money. The best will make a lot, everyone else will break even or lose money.


Answer: In orbit.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: