Rimpinths

I work in the industry too and the things that people mislabel as "front running" is really aggravating. At worst, you could call it "order anticipating": using publicly available knowledge to figure out that if someone hit Exchange A and B, they're probably headed to Exchange C next. But they have no inside knowledge that the same party will in fact send an order to Exchange C next. They're taking a risk by anticipating that.

"Front running" as defined by the SEC has a more narrow definition. It basically means that you have a customer that has placed an order for XYZ and you aware of the order, but you placed your own order to be executed in front them, thus forcing them to buy it from you at a higher price than if their order was executed first. HFTs are not "front running" anybody.

Great point, and you understand the American stock market better than most Americans. This article was true about 20 years ago, but the author is completely wrong when he says this:

"A single market to trade. All stocks for Microsoft (MSFT), are traded on the NASDAQ exchange. All stocks for Ford (F) are on the NYSE."

MSFT and F both traded on 16 difference stock exchanges, not to mention countless "dark pools", each with their own book of bids and offers. But there's a national best bid-offer (NBBO) that all exchanges must respect, so it can behave like a single market. That's what Reg-NMS is about. The benefits of having several exchanges competing each other, while trying to retain the benefits of single market. This is also where HFT enters the picture with latency arbitrage and other trading strategies when prices on those markets get out of sync.

20 years ago, you could say that MSFT only trades on NASDAQ, but that hasn't been true since Reg NMS came into effect in 2005. Each stock has a primary listing market that controls things like halts and opening/closing auctions, but the stock can be traded on any exchange, each with its own dynamics.

Right. I love to listen to music while coding, but I couldn't listen to a podcast. I've tried, but I can't write words and listen to words at the same time. But music is no problem.

Some HFT strategies depend on fragmented exchanges. Look up "latency arbitrage" for examples.

I've seen this type of behavior described as "order anticipation", which I think is a much better term. "Front running" means that you are using confidential information that an order will be submitted. "Order anticipation" means that are you using public information and you are anticipating that an order will be submitted.

Does this remind anybody of the CueCat? Same story...subsidized hardware that hackers repurposed for other uses. I think the makers of CueCat tried to unsuccessfully sue the hackers:

https://en.wikipedia.org/wiki/CueCat

http://www.securityfocus.com/news/89

Do the Amazon Dash buttons come with a license agreement? Has anyone read it to see if it prohibits these kinds of hacks?

But they are comparing sales per employee, not market cap per employee.

What you should really admire is Amazon's shareholders' tolerance for failure. A lot more companies would be more likely to take big risks like Amazon if their shareholders would tolerate year after year of losses. Time will tell if they are stupid or admirable.

I'm reluctant to get involved in the politically incorrect side of this argument, but FWIW, that stat comes from "Intelligence: Knowns and Unknowns", which was a paper published in the mid-90s by the American Psychological Association (in response to the controversial book "The Bell Curve").

From the paper:

"The differential between the mean intelligence test scores of Blacks and Whites (about one standard deviation, although it may be diminishing) does not result from any obvious biases in test construction and administration, nor does it simply reflect differences in socio-economic status. Explanations based on factors of caste and culture may be appropriate, but so far have little direct empirical support. There is certainly no such support for a genetic interpretation. At present, no one knows what causes this differential."

They boasted that they signed up 1 million new Prime members in the 3rd week of December. I wonder how many of those people signed up for the free trial with the intention of cancelling before the end of the free month? I know that's exactly what I did last year when I still needed to buy a few more presents a couple of weeks before Christmas.

Gold become desirable because it is an ideal metal to use for jewelry. That came first and its use as a currency came second. In that sense, it has been used in industry longer that it has been a currency.

And so on, and so forth. The author makes a big point of arguing that they aren't financial instruments so much as they are real assets. There's some truth to this; if I own a bitcoin nobody is obligated to give me anything in exchange; it doesn't exist as a liability on anyone's balance sheet. And? The same is true of a gold ingot. Or, largely, or a share of Apple stock; nobody is obligated to give you anything for that share. By the same logic, the "fair" value of gold (or Apple stock) is zero.

Gold has some intrinsic value. If nothing else, it has some industrial uses. Bitcoin has no comparable intrinsic value.

Owning a share of AAPL common stock has value. You get voting rights in the company. You also have a claim to Apple's assets, although with a lower priority than any bondholders or other creditors. So owning AAPL shares does come with obligations from Apple.

Euros (or dollars) also have some value as fiat currencies. You can pay debts and taxes in euros. The government has an obligation to accept euros as payment for taxes. So essentially these currencies are your ticket to stay out of jail.

Bitcoin doesn't share any of these qualities with gold, AAPL, or euros. That was the author's point and it still stands.

I was wondering the same thing. The number of employees is around 2300, according to Bloomberg. I think you could run Twitter with a development/engineering staff of 100. What are all of those other people doing? Sales?

I think it's a bad sign when you see a company as dramatically overvalued as Twitter is. The primary function of stock markets is capital allocation, i.e. directing capital to companies that can provides the greatest return. When you see an IPO like TWTR today, that's not a sign of rational and efficient capital allocation; it's a sign of speculation gone wild.

Why is this bad? It can cause a couple of negative effects:

(1) Good companies that deserve the attention of investors may be starved of capital while billions of dollars gets directed to flashy overvalued companies.

(2) If it turns out to be another bubble, investors will feel burned and they'll become more risk adverse in the future. Investors will be reluctant to provide capital to companies that can make good use of it.

Twitter just made more money from selling stock than they have ever earned in revenue in their lifetime. Something is seriously wrong when companies start making more money from selling stock than they do from selling products or services.

Economist Joseph Lawrence of Princeton University in 1929: “The consensus judgment of the millions whose valuations function on that admirable market, the Stock Exchange, is that stocks are not at present over-valued. Where is that group of men with the all-embracing wisdom which will entitle them to veto the judgment of this intelligent multitude?”

Alan Greenspan in 1999: "To anticipate a bubble about to burst requires the forecast of a plunge in the prices of assets previously set by the judgments of millions of investors, many of whom are highly knowledgeable about the prospects for the specific investments that make up our broad price indices of stocks and other assets."

There was an interesting article on Seeking Alpha a few months ago comparing Wal-Mart with Amazon at the same stage of growth (i.e. comparable level of overall revenue and revenue growth). He compared Wal-Mart in 1990-92 and Amazon in 2010-12:

http://seekingalpha.com/article/1547622-is-amazon-com-the-ne...

He noted that Wal-Mart had about the same levels of CapEx spending (~4%), yet it was also highly profitable by that point. Amazon isn't spending any more on CapEx than Wal-Mart was at the same stage of growth.

So Amazon isn't on some unique path in the business world by choosing to divert its profits to CapEx spending. Its level of CapEx spending is comparable to Wal-Mart at the same stage of growth. It simply has such low margins that it has no profits leftover after its CapEx spending.

They're thin in some places but huge in others. AWS, for example, has a 50% gross profit margin, and their gross margins on digital goods (mp3s, ebooks) are also high.

How do you know this? I don't think they have ever released this information in their financial reports. Are you going by analysts' estimates?

Gene Weingarten of the Washington Post set this up and wrote a Pulitzer Prize winning article about it. It's a great article, "Pearls Before Breakfast":

http://www.washingtonpost.com/wp-dyn/content/article/2007/04...

It's an asset from the point of view of the bank or institution that made the loan. And they can transfer those assets from one investor to another. And the value of those assets can be based on unrealistic expectations about future default rates, so in that sense, student loans certainly can be a "bubble".

The real damage done to the economy by the mortgage bubble wasn't all of the people who lost money on homes when prices collapse. The most damage was due to the effect it had on banks' balance sheets when their assets (the value of debt-backed securities) lost value, which meant they had to call loans and restrict credit to make up for the difference. In that sense, a bursting of the student loan "bubble" could also have damaging effects on the economy.

It was a pretty big deal. I ran some queries against some intraday data of a proprietary market index (kind of a cross between the S&P 500 and Russell 2000) and it was the biggest one minute drop (at -0.56%) that I found within the past year.

For comparison, here are the top 5 that I found since 4/23/12:

(1) 4/23/13 @ 13:10: -0.56% (2) 8/1/12 @ 14:14: -0.45% (3) 11/29/12 @ 11:41: -0.44% (4) 6/20/12 @ 12:33: -0.38% (5) 12/31/12 @ 13:47: -0.35%

As best as I can tell from the news on those days, (2) and (4) were related to Fed announcements and (3) and (5) were related to the fiscal cliff.

If you're really interested in this topic, Michael Lewis's book "The Big Short" covers this in detail. Great book.

BATS Exchange in Kansas City is hiring C++ programmers. Needless to say, your background would be ideal.

http://bats.com/about/careers/

Speaking as a software developer at one of the exchanges, I thought these comments were accurate and insightful. A lot of people think that colocation is inherently unfair, but they don't realize what a huge improvement this is over the old system of a limited number of floor traders.

One slight correction: I think the movie about floor traders in Chicago is called "Floored". There is another another movie about floor traders in NY called "The Pit".

I thought what lrm242 said and what I said were pretty much the same thing. But there is no standard definition of 'latency', so it's hard to say what "they mean" unless they publish a definition on their website. But I know of at least three major exchanges (one of which I work for) that consider the starting and endpoints for latency measurements to be outside the firewall. As for what Turquoise means, I'm not sure but would love to see their definition to know if their numbers are comparable. Also would like to know if these measurements were made under load because that can also make a significant difference.

They are clustered together because it takes roughly the same amount of time to determine that an order can be matched (filled) or that it can't be matched (acknowledged, on the book.) Either way, it has to be processed by the matching engine.

I'm a software developer at an exchange in the US. The LSE has not released details of how it is measuring "latency", but FWIW, we measure latency as the time it takes to submit an order and receive an order acknowledgment or fill, with both the sending and receiving outside our firewall. The two biggest SW components involved are the FIX handler (the component used to process orders; FIX is an industry standard protocol) and the matching engine (the component used to match buy and sell orders). I could of course spend hours talking about our hardware and software customizations, but it's a very competitive industry. Our software is C++ running on Linux, and that's about as much detail as I'm willing to go into.

I've always really enjoyed reading your posts on HN. I saw that you were quoted in The Register with some identifying information, and I finally put 2 and 2 together and realized that you were the author of some of my favorite posts over at TMF too. Those were some hilarious stories. I hope you find the time in the future to share some more of them over there, but can completely understand what a time sink it is to write material for an audience of strangers on the interwebs. On behalf of those strangers, I would like to say that it was much appreciated while it lasted. Thanks for all of the laughs and knowledge that you've shared.

One other aspect of HFT that was not mentioned in the article is that HFTers often seek arbitrage opportunities. For example, the value of many ETFs such as SPY (i.e. an ETF tracking the S&P 500) are derived from the value of underlying securities. If the value of SPY versus the value of the underlying securities becomes out of sync, HFTers may go long one and short the other and then profit when they converge again. In this sense, HFTers only profit if the market returns to fair value. This applies to many ETFs, convertible securities, and options.