Hacker News new | past | comments | ask | show | jobs | submit login
Apple and the risks of trading 29,000 times per second (cnn.com)
76 points by processing on Mar 27, 2012 | hide | past | web | favorite | 47 comments

My fear as a pessimistic engineer is that unless there is more regulation of HFT, sooner or later someone big (a bank, an exchange, a country) is going to be financially wiped out by a HFT-gone-bad incident, and the resulting mess will take years to unravel while everyone else (pension funds, private investments, etc.) gets to suffer for the sins of their masters.

I propose an exponentially decaying tax on trades - the longer you hold a purchase, the less you pay when you sell it. If you WANT to trade at sub-microsecond levels, you can bloody well pay for the clean-up fund when your system goes bad.

> I propose an exponentially decaying tax on trades - the longer you hold a purchase, the less you pay when you sell it. If you WANT to trade at sub-microsecond levels, you can bloody well pay for the clean-up fund when your system goes bad.

It seems like any tax on trades at all (even a fixed tiny percentage per trade) would go a long way toward reducing HFT.

Marvin8 on that site had a great comment. Here it is:

What I'd like to know is what happens to customers who have resting orders near those "fat-finger" trades. If a customer had a sell stop-loss at $583, did he get stopped out of the market at that price or near $582? I'll bet he did. If another customer had a buy order in at $583, did he get filled on the way down? I'll bet he didn't. The broker wil ALWAYS come up with a bs excuse as to why a customer got a bad fill and never gets a good one. It's the way the industry works. That's why folks are complete suckers to get involved. The industry makes BILLION$ screwing its customers.

The guy who sold at 582 had to sell to somebody, so the buyer at 583 got his order too. That's why they're called trades. There's somebody on both sides.

That's the theory. In practice we know that's not necessarily true. e.g. stop-loss order and the stock gaps down. Did the stop-loss order execute? No, it may have executed as a market order.

And that's the point he was trying to make. He's saying that the broker can create any suitable excuse to fit the situation.

With HFT it may not even be an excuse and could very well reflect the reality of the situation i.e. the price has moved too quickly for the exchange to keep up.

If the claim is that a sell at 582 executed and a buy at 583 did not execute, I want to see evidence. I see variations on this claim with a frequency approaching high, always by a random internet commenter "betting" on some hypothetical.

the price has moved too quickly for the exchange to keep up. What does that even mean?

Edit: added link to show that orders don't necessarily get filled according to time priority ("an order clearly arrived later than ours with the same limit price, yet it was filled and we were not.")

It's only hypothetical until it happens to you.

Refer to the fleitz's comment about short squeeze. That's an example of orders that don't get filled.

In theory the broker is supposed to borrow shares to allow the trader to sell them short. What happens if the broker flouts securities law and does not follow the rules?


BTW, non-HFT'ers get consolidated market feeds which have higher latency than raw feeds that HFT'ers use. That's what a major part of the HFT debate is about.


http://www.tradeworx.com/TWX-SEC-2010.pdf [PDF] (refer to page 17, 18)

Thank you for the TWX pdf, but is now a good time to point out that the entire paper was dedicated to proving that HFT is a good thing? Their solution to the order priority is also removing artificial regulations, not adding more. I feel like I should have been the one to post that link.

The paper presents the case that HFT as it is practiced right now is bad due to regulations.

But we do not know if tweaking the regulations is the "right" move. Could be a case of "out of the frying pan and into the fire".

I've been trying to figure out how to keep my money as far away from Wall Street goons as possible. The hit on Goldman Sachs' reputation from a couple weeks ago is the latest signal that Wall Street's job is to steal customers' money, stay away. The loss in confidence will eventually get them, though that will probably just mean they get another bailout.

I just like how they invent lingo to cover up their screw-ups. False print? Come on - Now if I want to describe the problem there is a name for it. And they made sure to mention it three times + one in the photo so I will remember it.

Easy, "false print" comes from the days of ticker-tape

Ticker tape would literally be a paper tape. Trade info like last trade prices/qty would be sent over a telegraph link to be printed on the paper tape.

That's where the Print comes from.

The False bit is some error along the way means the print you got should be discarded. Historical reasons were operator error, dead rats on wires, cosmic rays etc

Right... because we don't have any jargon in the tech industry.

The jargon is for the insiders to speed up their communication.

Using the jargon with the outsiders is just short for BS.

"In 2011, BATS accounted for more than one in 10 U.S. stock trades, processing an average of 29,000 trades per second. Against that kind of computer power, retail investors don't stand a chance."

Correct me if I'm wrong, but isn't BATS an exchange? Why are retail investors "competing" against the exchange? Is the solution some sort of paper and pencil exchange? Or maybe we can go back to jumping up and down and flapping our arms?

"Why, these investors ask, do false prints and fat finger trades always happen on the downside"

Because anybody with money can take advantage of it. On the upside, only people holding the stock can do so (unless you short the stock, but I've never tried nano timeframe shorting.)

You're correct. It just means BATS is processing a lot of orders. I'm guessing LIFFE, LSE etc process far more. The articles conclusion is comepletey erreneous - this has nothing to do with HFT.

Edit: And wrt to retail investors "competing" agianst HFT - they're not. A retail investor is not interested in expoiting very short-lived pricing discrepancies between different contracts (or perhaps the same/equivalent contracts that trade on multiple exchanges). A retail investor is just that - an _investor_ - who holds the stock.

Retail 'investors' shouldn't have any issue with BATS if anything BATS provides liquidity, however retail 'traders' stand no chance because the computer is far better at technical analysis and pattern matching than the average trader. Retail traders never really stood a chance against institutions because retail trades the market rather than creating it like institutions do.

eg. A retailer can't execute a short squeeze but an institution can.

No one needs 29,000 trades a second liquidity. HFT serves no market making function. It is a drag on market operations and confidence.

No trader is doing 29,000 trades a second.

BATS is an exchange like NASDAQ. They're facilitating and executing trades for their clients. NASDAQ handled 70k per second in 2008, and could handle almost 4 times that load(http://www.forbes.com/forbes/2009/0112/056.html).

HFT has nothing to do with this.

I'm sorry, but this reads like some sort of PR hit job vs. BATS.

Given their self-inflicted disastrous launch, I don't think anyone needs to pay for a PR hit job against them.

Tom Clancy predicted this kind of thing years ago. Although in his story the catalyst was a foreign power messing with the system. Once things got rolling the automated systems went with it to the point of the whole system crashing. Clancy wrote it so that a retired money guy had to explain to the people in charge what really happened because they were clueless on how the system actually worked. The solution was to simply reset the clock back to what was before the problems began and they pretended it never happened. Some of the discussion on the economy and trading were really interesting in that book.

How is making 29,000 trades per second good for the market? It seems like trading at that frequency should be illegal.

I believe that number is all trades by all participants.

Why should trading at that frequency be illegal?

What is the optimal amount of trades per second one should be making?

Once per second? Once per minute? Once per year?

Even if there were such a number how could any bureaucrat ever arrive at the optimal frequency any particular market participant should be trading at?

Does the intrinsic value of a company change 29 000 times per second ? is this signal or noise ?

Trading once per day and randomizing the order of trades would work nicely I think. There was a recent very good example of this concerning Apple : When Steve Jobs died, they kept the information secret and agreed with the stock exchange to suspend trading for a day. The time for everyone to think about what it meant for the company instead of hysterically following a random trend. A big event, obviously changing the company's intrinsic value, was interpreted with a daily period. Everything I read about high frequency trading indicates that it is mainly noise.

I mean, stock exchanges close at night and during week ends. There are 48 hours without quotations. How does one concile that with nano-second trading.

As others have pointed out below, the value of the company may change at a fast rate, however that is far from the only reason that you might see many deals in a short period.

One of the primary benefits that market makers (which are often HFT firms) provide is the efficient transfer of risk. That is to say, the value of the company may not change thousands of times a second, but the willingness of existing holders of that stock to continue to do so may change rapidly. Or similarly some participant may suddenly need a hedge and buy this stock because it has the correlation their looking for. This is all to say that there are valid reasons for buying and selling a stock that have no direct (but some indirect) relationship with the value of that company. This is not a game played by retail investors, but it's not just played by HFT firms either. Institutional investors (e.g., my 401k money) will also use financial products in this way.

Or put a completely different way, I may sell my Apple shares because I need the cash back to help pay for a downpayment on my house. Did I sell those shares because I think the value of the company has suddenly changed? No. There are a myriad of reasons participants may move in and out of a position.

If it's mainly noise, why do you care?

What I find most amazing is that the stock market is relatively small, yet people have strong opinions about it while ignoring much larger markets. There's some adage about how if you can't measure it, it doesn't exist or something. If CNN doesn't run a ticker for some market, I guess it doesn't matter.

I care if all my 401(k) alternatives are in markets being parasitized by HFTs who have learned to exploit execution flaws and attack the confidentiality of the order book. Less selfishly, I care if HFTs are diverting society's resources away from the effective producers the market exists to support.

Less selfishly, I care if HFTs are diverting society's resources away from the effective producers the market exists to support.

Unfortunately, there's gold in them there microseconds.

For the same reason you minimize noise to prevent Larsen effects : It is noise that can have feedback effects.

The intrinsic value of a company is unknowable, the only thing we can know is the market value.

For some companies I do think the value could change 29,000 times per second, if it didn't the stock simply wouldn't move . You can see this in the juniors where HFT is rare if non-existant.

I do think that it's a good idea to have an exchange that doesn't allow HFT, or perhaps only allows trades every day, and for it to be able to compete with exchanges that do allow HFT. If this system is better then companies will list their stock on that exchange instead of exchanges that don't allow HFT.

However, what I disagree with is the idea that people should be put in jail for trading stock too frequently.

Even though it will be hard to determine the optimal amount of trading per second, you have to agree that the situation now is far from normal.

One of the most important economic benefit, and function of the stock market, is to allow companies to raise capital.

High frequency trading does nothing to promote this goal.

In some sense the stock market today is more like a gambling hall.

Well said. As in some casinos, in the world's stock markets some big players (probably including certain HFTs and hedge funds) take much more of the winnings than small ones, due to information asymmetry or pure competitive advantage (e.g. due to extremely fast trading and advanced algorithms).

The more money is sucked from the market by intelligent arbitrageurs/gamblers, the less there is for small(er) investors. That's a bad thing in itself and also increases the focus on financial gameplay (analysis of market trends and fads) to the detriment of investment (analysis of the worth of companies). Frothy, ever-rising stock markets attract investment away from other markets, usually for poor reasons: the traded companies haven't done anything that would justify such rapid and significant changes in their stock values.

The point of the stock market is not to let the smartest people profit from the rest; it's to allow companies to raise capital efficiently and fairly. If we tweak the rules of the market, by limiting trade frequency for example, we may be able to push the play in the market towards a form that aligns better with this core function of the market.

Finally, I want to reiterate that for a closed population, the stock market is a zero-sum game. Money made is won from someone else, always.

edit: i'd appreciate you mentioning the points you disagree with when you give your downvotes - thanks

I didn't downvote you, but you're wrong for a huge variety of reasons.

Firstly and most importantly, the stock market is not zero-sum.I don't know why you think it is.Equities in companies ideally (and historically) grow in real value.This is basic common sense.If your friend sells you a stake in his company, you have an ad-hoc stock market (a buyer and a seller for company equity). Is one of you destined to lose in this deal? Or can the company do well and you both succeed?

What's going on with HFT is not a novel way to exploit the system. Financial pros will always get the information sooner and be able to act on it more quickly/efficiently than retail investors. People who suggest otherwise are deluding people for political purposes. Before computers existed, people on Wall Street still got the information first and acted on it first.

What computers have done is made the whole thing quicker and more convenient. If you want to fill an order you are much more likely to be able to. Those advanced algorithms help make sure things are correctly priced which is good for buyers and sellers. The companies looking for investment benefit from this. The smaller investors aren't hurt by this at all from this, assuming they are trying to invest in value and not take advantage of arbitrage opportunity.

Contributing liquidity and pricing information to the market is valuable. Some people do it faster and better than others and so they profit from it. It's unclear to me why they need to do a worse job of it so that other people can share in this value. Should Google be limited to only providing so many search results a day so that other small(er) search engines can get some of the wealth too? What exactly is the problem, besides the general popularity of banking fear-mongering recently?

"Firstly and most importantly, the stock market is not zero-sum."

It is, unless you have another definition of zero-sum. The talk about companies "growing in value" ignores the fact that the "value" is purely what the market will pay for those companies. If the market is a closed system, i.e. companies are not being listed or delisted, then the net profit made by buyers & sellers if all transactions were to be closed out is zero, by simple summation. (Actually, it's negative, because of significant transaction fees.)

"If your friend sells you a stake in his company, you have an ad-hoc stock market (a buyer and a seller for company equity). Is one of you destined to lose in this deal? Or can the company do well and you both succeed?"

If the company does well and you sell the stake to a third person (C), then you've profited and C is left holding the bag. Every single trade that is closed out results in either a profit or loss to its participant, and it's only the fact that the more money is put into the stock market over time that obscures the fact that, if all extant trades were to be closed out at one point in time (i.e. everyone cashed out), the net profit of everyone involved would be zero.

The market keeps growing because new money and participants are flowing in, but if this stops (say, due to foreign investors wanting to put their money elsewhere, or a shrinking investor population, or a recession), then all you have is a moribund market into which existing investors dare not put more money. The average profit for trades goes to zero because the average stock price is not moving up or down.

This is practically illustrated whenever a bubble bursts and money leaves the market: many participants start taking the losses that had been hidden by the previously rallying prices. The winners are the ones who got out first and took profits from those inflated prices.

I argued this point with two friends of mine, one who was an investment advisor, and another who was an oil trader. The advisor said that it wasn't zero sum and the market was always expanding, but after a bit of discussion and graph-sketching, he came to see that it was zero-sum after all. The oil trader (a pretty savvy chap) said, "Yeah, that's obvious. That's the game I play every day."

(I think this can be extended further to encompass the various financial markets as alternative sectors of a giant investment market offering various classes of product. In the end, each market is zero-sum, and so the overall investment market is zero-sum, which is obscured by the increase in production as technology advances. But this part is more speculative so I won't push it.)

I'm a bit occupied atm so I'll reply to the rest of your message if there's still interest later.

No, your friend is not left holding the bag. He's left holding the money you paid him for the equity. You were the one in the red at first, but since the value of your stake could grow and pay you in dividends you overcame that loss. Both of you gained money - that's the point. He didn't even lose the opportunity for more money, because he needed the initial investment to even grow the company in the first place.

Here's the key point you're missing. In order for it to be zero sum, if you make a million dollars off of this company, he needs to LOSE a million dollars. Not the opportunity to make a million dollars, an actual million dollars. Clearly this does not happen. You are totally and entirely wrong here.

The market doesn't grow because new money flows in. This is mercantalism and your understanding of economics is hundreds of years old. The market grows because companies create REAL (not nominal) value that didn't exist before. I can explain this in more depth, clearly your financial friends aren't very good at their jobs (it's quite common).

The reason your oil trader friend agrees it's zero sum is because the futures market is zero sum. The stock market is not.

Okay, I accept that I could be entirely wrong. I had forgotten about the role of dividends in determining the value of a company, and that companies definitely do appreciate in real value on average. (I should also have clarified I was thinking only about a closed secondary market, as opposed to the primary market where funds flow into companies.)

What I'd like to understand is this: if the total real value of companies is expanding in the long term, then isn't the stock-trading/investment game about who manages to pick the fastest-growing stocks and capture the most price appreciation? The net gain in value would exist as long as people had invested in the first place, but its distribution is zero-sum in that the net gain in value is captured by someone or other. I guess this is what I was thinking of when I talked in earlier posts about "closing out" all trades - if all trades are closed out at one point in time, then all value in the market is captured by one player or another, and the result is zero-sum.

If the total gains in real company value are influenced by the trading game in the secondary market (e.g. total gains depend on the volume of trading in the secondary market, perhaps through new stock issues), then the question does seem a little more complicated.

Glad to have had this discussion. I guess I need to read up on economics more, although googling didn't turn up a lot of actual literature on stock markets and value-creation right away.

I completely agree. There are a fair amount of investors that are going after a percentage of a company, they are making investments that can actually propel a company forward. The rest of us are gambling on a popularity contest.

Just add a sales tax on stock.

You'll just make the spread larger by the amount of the tax. Why is this a good thing?

While it would make the spread larger, I disagree with your use of the word "just". It would have many other effects as well - for one thing, it'd make trades on small shifts in value (ie less than twice the sales tax) unprofitable.

That would make HFT much less attractive at the ridiculous frequencies it happens currently, as you'd need to hold on to stock for longer for it to shift enough for the gross gain to exceed the sales tax. I suspect that's the effect ars had in mind.

Also, you could always put the tax into a kitty for bailing out the financial institutions who're engaging in this lark the next time they mess it up ;)

That would make HFT much less attractive at the ridiculous frequencies it happens currently, as you'd need to hold on to stock for longer for it to shift enough for the gross gain to exceed the sales tax.

This is true if for some reason the HFT wants to take liquidity on both sides of the trade. In reality, the HFT will probably add liquidity on both sides of the trade, making a profit equal to the true spread plus any change in price. Moves of one tenth of a cent will continue to be massively profitable for the HFT, but the HFT's counterparty will have to pay a much larger spread.

One group of people who commonly take liquidity on both sides of the transaction are retail investors, so they would end up taking a great deal of the losses created by this tax.

I must admit to being a bit of a layperson when it comes to the dark arts of finance, so I may well be completely wrong.

I was mostly going off my recollections of my high school economics teacher saying something to the effect of "tax on a transaction reduces the market size - for example, tobacco & alcohol duties serve to reduce usage of both products".

Would that not apply for some reason when the product in question is stock trades?

It's probably true that less trades would occur, and HFT profits would probably be reduced by this, but they would be able to mostly avoid paying the tax.

What follows is full of holes but is a mostly correct explanation of what I was talking about before.

Each venue maintains an "order book" for each security that trades on the venue. The order book consists of a number of standing orders, either orders to buy at $k or orders to sell at $j (k<j, with many distinct values of j and k). These orders which sit on the book because they do not have a counterparty are called "add liquidity." When you want to just execute the trade and you don't mind the spread, you place an order that crosses the spread and transact with the person who has the best standing order (selling for the least money or buying for the most money) of all the standing orders. This is called "take liquidity." My impression, which may not be correct, but which forms the basis for my previous response, is that the HFT industry primarily plays the add side right now and that most firms only take liquidity to close out positions when they are unable to close them out by adding liquidity.

What difference does that make? I'm really not following you. When they sell, or buy, they pay the tax. It makes no difference if they place a limit order or a market order.

The spread is just the difference in prices, it's not an extra fee that some pay and some don't. The only reason it even exists is that where the prices meet trades happen, so the prices never actually meet for long.

The tax becomes part of the spread. The spread is paid (in aggregate) by the participants who take liquidity to participants who add liquidity.

It's not that market-makers and HFTs would not be assessed the tax at all, it is that they would be paid more than enough to cover the tax by the people who cross the spread.

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