The Casino Called Wall Street

May 15, 2010 3:11 PM

Maybe Microsoft is trading in London at a penny less than it's trading at the same moment in New York. A high-frequency trader will buy shares in London and wait for them to rise. Since the discrepancy lasts a mere fraction of a second, speed is key. [Tradework CEO] M. Narang boasts it takes only 15 millionth of a second for his computers to place a buy or sell order after detecting an opportunity. Or, as he puts it, "If you try to pick up the penny, we'll probably beat you to it."

While the exact cause of the stock market's recent "flash crash" remains uncertain, the event has renewed concern about the incredible growth of high frequency trading (HFT) in the market.

To put HFT in the broader context of how the complexity and speed of the market has changed, consider the recent remarks of SEC chairman Mary Schapiro (from the "remains uncertain" link above; all of her comments are well worth reading BTW):

Although developments in the markets and in technology may help speed access to market data, they also greatly complicate our efforts to analyze the complex web of trading arrangements and market dynamics that have developed since 1987. For example, the key day in the 1987 Market Break Study involved a trading session processing a little over 600 million shares in NYSE stocks. Last Thursday, the markets processed 10.3 billion shares in NYSE stocks alone. In addition, the interconnections among markets and among equity securities and derivatives have grown immensely more complex over the past few years. Orders in one stock directed to one market can now ricochet to other markets and trigger algorithmic executions in other stocks and derivatives in milliseconds. By contrast, in 1987, investigations could focus their attention on discrete transactions largely effected on only one or two markets. [emphasis added]

There's also the question about whether or not HFT increases the possibility for fraud. As one commenter on a HuffPo thread notes (how accurately, I cannot say):

certain stock exchanges, such as Direct Edge, will--for a fee--send a 'flash' of a upcoming trade to the firms that engage in high frequency trading. For instance, if a mutual fund is going to buy 1 million shares of IBM, Goldman Sachs or other companies with HFT operations will get a notice from Direct Edge that this is about to happen. Within milliseconds, the HFTs put in buy orders for IBM, then let the mutual fund buy their 1 million shares, which pushes the stock price up, and then the HFTs sell at a couple of pennies higher. As it stands right now, this is LEGAL. The SEC is looking into blocking the practice, which they should, because it's simply a high-tech version of insider trading.

Whatever the case about HFT and the role it plays in market volatility and/or manipulation, one thing is certain: some investors are preparing themselves for any future "flash crash," and no doubt hoping to exploit it.

posted by HP LaserJet P10006 (62 comments total)

I thought the NYSE and NASDAQ had imposed a 15 minute latency in trade reports precisely to prevent this kind of thing.

In the early 1990's there was a case where program trading went bozo, and the stock market dropped a huge amount one day for no apparent reason. It later turned out that all the program traders responded to the same situation in the same way and did a bunch of selling all at once. Over the next month or so, the market rebounded all the way.
posted by Chocolate Pickle at 3:20 PM on May 15, 2010

Chocalte Pickle, while I'm not an expert on thios, come of the remarks of Mary Schapiro (third link) may help here:

it is useful to describe the effect a certain feature of the NYSE had on market movements that day. In attempting to meld the traditional open-outcry floor-based auction model with today’s technology, the NYSE’s trading system utilizes what are known as “liquidity replenishment points,” or LRPs. LRPs are best thought of as a “speed bump” and are intended to dampen volatility in a given stock by temporarily converting from an automated market to a manual auction market when a price movement of sufficient size is reached. In such a case, trading on the NYSE in that stock will “go slow” and pause for a time period to allow the Designated Market Maker to solicit additional liquidity before returning to an automated market. This “speed bump” occurs even when there may be additional interest beyond the LRP price point.

On days of major market volatility such as May 6, stocks with significant and continual declines may frequently cause NYSE trading to go slow for many different stocks. Some have suggested that this practice caused a net loss of liquidity as orders were routed to other markets still offering automated executions. Others believe that the LRP mechanism served to attract additional liquidity that helped soak up some of the excess selling interest. We are focusing on whether the disparity in exchange practices can be addressed to promote more consistency in how orders are handled in the context of rapidly changing prices without undermining the benefits of individual market practices.

Another disparate exchange practice that occurred on May 6 is the use by certain exchanges of the “self-help” mechanism under Regulation NMS. When an exchange believes that another exchange is experiencing systems problems, the “trade-through” rule under Regulation NMS (described in section III below) permits the exchange to declare what is called “self-help” against the other exchange. This allows the exchange to exclude the quotations of the other exchange from its determination of whether the other exchange has a better “protected” price to which it must route orders for execution.

On the afternoon of May 6, just prior to the steep market decline, two stock exchanges declared self-help against another exchange, thereby excluding its quotations (and liquidity) from the two exchanges’ routing tables. The excluded exchange has asserted that it did not experience systems problems that would warrant the declaration of self-help. We are investigating these issues and whether there needs to be greater consistency in exchange practices with respect to the self-help mechanism.
posted by HP LaserJet P10006 at 3:25 PM on May 15, 2010

Broken systems are broken.
posted by hippybear at 3:30 PM on May 15, 2010

So it's like Core Wars? Except when it ends the losers need to sell even their kidneys to pay the winners? Shit, who needs science fiction anymore these days...
posted by Iosephus at 3:35 PM on May 15, 2010 [3 favorites]

The speed of trading isn't the problem. If you go back to the old ways of actual stock boys running from office to office with paper shares, life for large American businesses isn't exactly going to get better.
posted by Doug Stewart at 3:38 PM on May 15, 2010

Microsoft is trading in London at a penny less than it's trading at the same moment in New York. A high-frequency trader will buy shares in London and wait for them to rise.

So, When the prices rise and the stock is resold the HFT makes a profit.

Where does this profit come from actually? It seems like a numbers game for the HFT, but wouldn't that money eventually come from somewhere in reality?

Also, trading without the possibility of loss seems like I would be against the "spirit" of the market.
posted by Hicksu at 3:48 PM on May 15, 2010

still hammering on the HF trading eh? you are barking up the wrong tree if you are looking for explanations on "the crash of 2:45".
posted by H. Roark at 3:49 PM on May 15, 2010 [2 favorites]

The speed of trading isn't the problem.

Perhaps, but it's not just the speed that's a concern: it's also the lack of transparency, size, and complexity of what, according to the first link in the FPP, accounts for "two-thirds of all U.S. stock trading." If HFT makes fraud, manipulation or insider trading easier (and it's certainly to see how it wouldn't), and if the SEC is currently looking at HFT as one possible cause to the May 6th "flash crash" (and it is), then the question of the role that HFT plays in the market seems valid. Consider this recent exchange:

What about high-frequency trading, asked Faber.

Blankfein, declining to comment on speculation high-frequency trading was at the heart of the [May 6] collapse, “I think that business in and of itself is a good business for Goldman Sachs, and it’s an important liquidity generator for the market. I have no opinion, because I haven’t heard any connection made to the …” and then he trailed off as Faber moved to the next question.
posted by HP LaserJet P10006 at 3:51 PM on May 15, 2010

I thought the NYSE and NASDAQ had imposed a 15 minute latency in trade reports precisely to prevent this kind of thing.

Seriously? You can get real time trade information on almost any NYSE or NASDAQ stock on Google finance for free, today. And real time information has always been something that you could pay extra for, usually not much extra at all.
posted by delmoi at 3:51 PM on May 15, 2010

It's interesting to see the parallels from a distributed systems perspective. Some of the speculation (Paul Kedrovsky 2 and Kid Dynamite, particularly this comment) sound very similar to an eventually consistent (AP¹) system reaching equilibrium after a network partition.

It's not a perfect analogy because obviously someone had to be selling at $0.001 (which would be a byzantine failure not just a simple partition) but the HFT liquidity providers exiting the market due to latency is striking.

¹ Eric Brewer's CAP theorum shows that for any distributed system you can pick two of Consistency, Availabilty and Partition tolerance, never all three. Some systems (mostly Dynamo [PDF] based) allow the system to become temporarily inconsistent by allowing each side of a partition to write (trade) as if everything were fine. When the partition is resolved the inconsistencies are rolled back via some mechanism, often timestamp or vector clock.
posted by Skorgu at 3:53 PM on May 15, 2010 [2 favorites]

Well not today today, it's Saturday. But check on Monday.
posted by delmoi at 3:53 PM on May 15, 2010 [1 favorite]

Also, trading without the possibility of loss seems like I would be against the "spirit" of the market....
HF and all algo stuff trades with real potential for loss, that is frequently realized. Its not at all a sure thing, just like any other investment strategy. And I use the term "investment" carefully here - does anyone railing on the HF community understand that many of the popular investment vehicles for most people (notable ETF's) are heavily supported under-the-hood by HF algos? Everyone loves the fact that the loads on ETF's are ultra-low - thank your ETF's HF dept! Or would you prefer some fucker taking 3% off the top of your S&P tracking fund?

ah, but now carryon with the bashing...
posted by H. Roark at 3:54 PM on May 15, 2010

It's not a perfect analogy because obviously someone had to be selling at $0.001 (which would be a byzantine failure not just a simple partition) but the HFT liquidity providers exiting the market due to latency is striking.

Check out this anonymous askme question from a regular day trader playing around with $5k saw a stock price of 1¢ and placed a market order, and ended up buying $600k worth of stock naked. According the question he sold it right away and made $17k. Of course it could all be made up, but with how crazy the market can be who knows...
posted by delmoi at 3:56 PM on May 15, 2010

High frequency direct arbitrage would tend to push the price back up.

Suppose that MSFT drops by $1 in NYC, but not in London. The arbitrage opportunity is to buy in NYC, sell in London. Buying tends to push prices up, particularly as buy volume meets and then exceeds sell volume. London prices would be pushed down, but only to the point where the NYC prices meet the London prices.
posted by b1tr0t at 4:14 PM on May 15, 2010

Skorgu: anecdotally, the coders writing high speed transaction financial software definitely think a lot about parallel processing and distributed systems. Codeword: Event Processing
posted by honest knave at 4:15 PM on May 15, 2010

the technologies used for market oversight and surveillance have not kept pace with the technology and trading patterns of the rapidly evolving and expanding securities markets -- SEC chairman Mary Schapiro (from the the "remains uncertain" link above)

I can't see how this will ever change, unless perhaps you create an SEC test suite that all trading software must pass, and you make the software checksum frequently to prove that it's the tested version.
posted by honest knave at 4:20 PM on May 15, 2010

carryon with the bashing

Thank you for explaining. I certainly didn't want to sound like I was bashing HFT. The article made it sound like a "sure thing", so I was curious.
posted by Hicksu at 4:20 PM on May 15, 2010

Man, life sure was better in the 80s, back when there was a single market maker for each stock and he could simply pocket the $0.125 spread in bid/ask prices one very trade.
posted by Nelson at 4:26 PM on May 15, 2010

I've been lurking for several years but there is enough misinformation in this thread that it finally encouraged me to sign up.

certain stock exchanges, such as Direct Edge, will--for a fee--send a 'flash' of a upcoming trade to the firms that engage in high frequency trading. For instance, if a mutual fund is going to buy 1 million shares of IBM, Goldman Sachs or other companies with HFT operations will get a notice from Direct Edge that this is about to happen. Within milliseconds, the HFTs put in buy orders for IBM, then let the mutual fund buy their 1 million shares, which pushes the stock price up, and then the HFTs sell at a couple of pennies higher. As it stands right now, this is LEGAL. The SEC is looking into blocking the practice, which they should, because it's simply a high-tech version of insider trading.

This really isn't true. The way it works is that if your order is not marketable, (so either your order is really big, or you don't want to pay the going price), exchange members can receive a chance to fill your order with price improvement before it is routed to other exchanges. If your order is marketable it does not even get flashed. In practice there is not sufficient liquidity to execute a million share buy, and therefore a million share buy must be spread out over time. This notion that you can get instantly frontrun on a million share order is hyperbole designed to spread outrage. A properly executed million share order can not be distinguished from market noise. Algorithmic (not high frequency necessarily) trading, where computers are used for order execution instead of a physical trader, is slowly putting buy side trading firms out of business. Guess who have consistently been raising the most fuss about high frequency and algorithmic trading? Buy side trading firms. Much of the high frequency trading outrage is nothing more then a giant lobbying operation disguised as a grass roots main street investor movement.

Disclosure: I daytrade, but am not involved in HFT. HFT firms are usually on the other side of the trades I make.
posted by An algorithmic dog at 4:30 PM on May 15, 2010 [14 favorites]

I thought the NYSE and NASDAQ had imposed a 15 minute latency in trade reports precisely to prevent this kind of thing.

FINRA requires that trades be reported within 30 seconds (it was changed recently from 90 seconds).

The data is delayed because they want people to pay for data subscriptions.
posted by milkrate at 4:36 PM on May 15, 2010

I for one welcome our new overlords and admit I find the idea of these guys making a penny (ha!) off an imperfect system not deplorable at all.

I've been lurking for several years but there is enough misinformation in this thread that it finally encouraged me to sign up.

welcome. good to have you here.
posted by krautland at 4:37 PM on May 15, 2010 [1 favorite]

Much of the high frequency trading outrage is nothing more then a giant lobbying operation disguised as a grass roots main street investor movement.

Well the big firms appear to be making a killing through HFT, fwiw.

And while the controversy about HFT, which goes back at least six months now, may or may not be warranted, it certainly is educational.
posted by HP LaserJet P10006 at 4:39 PM on May 15, 2010

H. Roark, you seem to be conflating HFT with arbitrage. While some arbitrage operations may involve HFT, HFT does not encompass the universe of arbitrage. HFT is trading that takes place in spaces of time so small that physical proximity of computers to the exchange (for network latency) is a very real concern. The arbitrage mechanism that keeps the prices of ETFs in line with their underlying instruments is only that precise in the very liquid, very high volume funds such as SPY. Others, especially those for whom the underlying instruments are illiquid, can experience divergence from the price of the underlying for easily observable periods of time. Additionally, no load whatsoever is paid to acquire any ETF (you might be thinking of expense ratio, which low value is more due to the index-tracking nature of the fund and plain old free market competition than the arb mechanism).
posted by indubitable at 4:44 PM on May 15, 2010

Of course, making a killing does not necessarily mean that anything illegal or even unethical or even bad for the market is going on. (Indeed HFT Is largely necessarily to keep net asset value of ETF's inline with the underlying assets during the trading day. Without high frequency trading ETF's experience wild price swings (which is exactly what happened when HFT people turned off their algo's during the "flash crash").

Most of the high frequency trading story broke on zerohedge.com, a largly anonymous site. Most of the papers that have fanned the flames of the HFT hate have come from Themis Trading, a buy side trading firm. Some of the first interviews on Bloomberg on why HFT is bad have come from Themis. I don't know if I would go as far as to call it an astroturfing war, but the people speaking out against HFT have all sorts of things to gain by seeing it more heavily regulated. If you could go out and get a computer to place an order for you and execute a million share buy without human intervention you don't need companies like that anymore.
posted by An algorithmic dog at 4:47 PM on May 15, 2010 [2 favorites]

Market not fee enough. Needs more freedom.
posted by hamida2242 at 4:50 PM on May 15, 2010

the people speaking out against HFT have all sorts of things to gain by seeing it more heavily regulated.

Does that include Senators Schumer and Kaufman? Like derivatives, the subject of HFT need not be reduced to a simple "for/against" dichotomy (or "legal/illegal") in order to be worth discussing. Markets need to be understood before they can be regulated in any case. Certainly the current FPP is meant to stimulate critical discussion about a little understood topic that has been generating a lot of news, and not to bash HFT outright. That much should be clear from the way it's presented.
posted by HP LaserJet P10006 at 5:05 PM on May 15, 2010

Yeah, I don't think HuffPo is the place I'd turn to for information about market structure issues.

One thing to be aware of is how small today's orders are - most are less than 500 shares, because of how traders work orders and the way algorithms divide up trades. A "million share order" is not sent to an exchange in that way; it is a block trade - it gets negotiated.

Second, flash trading takes advantage of a loophole in Section 602 of the 1934 Act. And it is likely to close this loophole soon (at least on equity exchanges, options exchanges will probably be treated slightly different for a few reasons).

This is the general process for flash trading:

1. The exchange receives an order and attempts to fill as much as possible at the best bid-offer against its order book, both displayed and non-displayed.

2. If the entire order cannot be filled against the exchange’s order book, the order can be:
2a. Cancelled, if it is an "Immediate-or-Cancel"order
2b. Posted to the exchange’s order book, if it is not marketable at another exchange
2c. Routed to another exchange

3. If it is to be routed to another exchange, the order may be "flashed" at the initial receiving exchange at the best bid-offer. For a short period, the order is exposed to a group of traders (in Direct Edge's case, it is 25 "enhanced liquidity providers" who are competing against each other) who can choose to fill the order at the best bid-offer or better.

3a. The flash order will either be filled or time will expire, at which point the remainder of the order would be routed to another exchange to be filled.

BATS and Nasdaq introduced their flash trading programs about a year ago (BOLT and Nasdaq Only Flash Orders, respectively). They were shuttered within weeks after the controversy erupted, likely because they just wanted to cause an uproar to push the SEC into shutting down Direct Edge's program. It hasn't happened just yet.

But Direct Edge has maintained the controversial practice, though the use of the ELP program has fallen over 60% (down to 76m ADV in March 2010 from 201m ADV in March 2009) and they have managed to lose a ton of market share over the last 9 months (dropping from about 14% to 9%). The real reason why Direct Edge has maintained it through the controversy? Because flash orders can generate 7x more revenue for the sponsoring exchange than a standard order.
posted by milkrate at 5:19 PM on May 15, 2010 [3 favorites]

HP: I didn't see one article pointing out good things that HFT does while I saw a bunch of them spreading bad things. For instance:
A good FAQ from the WSJ
A good article on benefits of HFT
Schumer and Kaufman are primarily guided by a desire to look like they are taking on big bad wallstreet. Having said that, I can certainly get behind some things: more study is always a good thing, as long as the goal of the study is not just to justify taking HFT firms out behind the shed to shoot them. And I think that naked access as it stands now is incredibly dangerous and while it may not need to be done away with entirely, at the least it needs to be significantly more strongly regulated. So Schumer and Kaufman are accidentally right.

On the other hand, I think claims that HFT front runs orders are from people who are entirely misinformed on what front running is. And this is largely the fault of Quants and Traders. People on wall street seem to have completely lost the ability to explain what they are doing in plain English that people who don't work on wall street can understand (if you watched the GS hearings, that much was obvious). I think a good deal of the anger about HFT comes purely because no one is trying to explain why it is useful except by saying things about "liquidity." It is like the story about Feynman and text books on "energy". It may make sense to say that "HFT enhances liquidity," and take it as a given if you work in that field. For everyone who does't spend their whole life studying market microstructure, I don't think that it is obvious at all.
posted by An algorithmic dog at 5:21 PM on May 15, 2010 [3 favorites]

[Tradework CEO] M. Narang boasts it takes only 15 millionth of a second for his computers to place a buy or sell order after detecting an opportunity.

15μsec? Really? Light itself travels about 5 km in that time, never mind things like router lag and the like.

What this is saying is that the most important thing in the trading world is your physical network distance to the NYSE and NASDAQ trading computers. You want to beat this guy? You get 3.3μsec every km closer to the trade than he is.

The word for this, of course, is isanity. The fix is easy. Every trade has, upon entry to the system, a delay of 1-1000msec randomly added to it. The only thing I'm arguing, here, is the "milli" part. 1-1000 seconds would bring this back to human comprehensible speeds.

Indeed, the more I watch the automated market, the more I'm convinced that the correct answer is to remove the computers from the trade to reduce the velocity and gamemanship.

Remember: The point of an equity market was to provide capital to business. Nowadays, who the fuck cares about that, there's far more to be made gaming the markets itself.
posted by eriko at 5:26 PM on May 15, 2010 [2 favorites]

Eriko: if you add in a random delay you still want to run as fast as possible because you want to start the timer for the delay as soon as possible. There is no way to get around the fact that at the end of the day there is a race to place orders that can not be eliminated. You can introduce uncertainties, but then you just build models that take into account the distribution of the uncertainties when deciding which trades to place. If you remove computers from the trade all I do is go pay world class twitch gamers to enter my trades as a computer tells them what to do. There is no way to remove the speed factor from trading. Some of the first uses of telegraphs and telephones were to trade on information before it had reached everyone else by newspapers or other means. Trading and business has always been about speed and always will be.
posted by An algorithmic dog at 5:30 PM on May 15, 2010 [3 favorites]

I don't think HuffPo is the place I'd turn to for information about market structure issues.

Probably not; that's why I gave six links in the FPP and four in my thread comments.

I have no idea whether Schumer's proposed legislation regarding flash trades (i.e. HFT) from last summer (when the topic was also discussed on MeFi) is good or bad (or well intentioned but wrong, or ill-intentioned) in the final analysis, but looking at it might help frame the question somewhat:

I request that the SEC act to prohibit the use of so-called “flash orders” in connection with optional display periods currently permitted by DirectEdge’s Expedited Liquidity Program, NASDAQ’s Flash order program and BATS’s Bolt Optional Liquidity Program. Flash orders allow certain members of these exchanges to obtain access to order flow information before that information is made available to the public, allowing those members to use rapid trading programs to trade ahead of those orders and profit from advanced knowledge of buying and selling activity.

While pre-routing programs can benefit markets by providing additional liquidity, this kind of unfair access seriously compromises the integrity of our markets and creates a two-tiered system where a privileged group of insiders receives preferential treatment, depriving others of a fair price for their transactions. If allowed to continue, these practices will undermine the confidence of ordinary investors, and drive them away from our capital markets.
posted by HP LaserJet P10006 at 5:30 PM on May 15, 2010

Does anyone other than me have a problem with the fact that a large part of the market we're talking about consists of human beings playing shell games with meta-money, in ways that have less and less to do with anything concrete, productive or of any social value?

That we offer obscene rewards and high social status for playing these (often as good as rigged) games successfully?

That a lot of us these days seem content to live in, and on, this abstracted meta-economy, playing shell games for a living, and fewer and fewer of us know how to actually make or do real things?

(Just before I posted this, eriko seems to at least a bit.)
posted by namasaya at 5:31 PM on May 15, 2010 [3 favorites]

What this is saying is that the most important thing in the trading world is your physical network distance to the NYSE and NASDAQ trading computers. You want to beat this guy? You get 3.3μsec every km closer to the trade than he is

eriko, that's exactly what's happened. Lower Manhattan is now full of colo/datacenter facilities serving financial companies doing HFT.
posted by teferi at 5:48 PM on May 15, 2010 [1 favorite]

This is why I only invest in tulips.
posted by Legomancer at 5:49 PM on May 15, 2010 [6 favorites]

An algorithmic dog:

Great explanations!

I'd add that it absolutely isn't obvious that more liquidity is always better. Everyone agrees that completely illiquidity is bad - but for 99.5% of humans, if their trade went off within 30 minutes and was within less than a percent of the market, they'd be fine (and 30 minutes/1% is a simply HUGE illiquidity by trader's standards).

An algorithmic dog, I invite you to really think through what the consequences of a 0.1% tax on each trade would really be.

Certainly, it would cut out a lot of your life, and leave a lot of Wall Street people high and dry - but would it really prevent business from being done? Would the stock market collapse. I would say, definitely not! - and it would absolutely prevent a lot of this bullshit.

And I'm ABSOLUTELY not proposing this - though I might well be behind a 1% of 1% tax... but if you're a trader type, as I once was, it's useful to think through what such a world might be like.
posted by lupus_yonderboy at 5:57 PM on May 15, 2010

The data is delayed because they want people to pay for data subscriptions.

Another way to look at it is to think of data timeliness as a precious quality that's very valuable. Per-minute data, per-second data, and even "real time" feeds are available for increasing cost and are also priced by data granularity. 15 minute old data is considered obsolete and worthless from a floor trading point of view, so it's cheap enough to give away for free on the web.
posted by ceribus peribus at 6:04 PM on May 15, 2010

I could care less if there is a transaction tax. The major players will just widen the bid ask spread from $0.01 to 0.1% of NAV+1 cent. So for a $50 stock the bid ask spread will go to $0.06. Since I buy at the bid and sell at the ask, this will not effect me at all. It will effect all retail investors, who will now be paying an extra $12 round trip for their 100 share purchase.

However, the widening of the bid ask spread will probably also decrease liquidity. This will probably increase interday volatility some. From my perspective as a trader, more volatility (to a point) is awesome because it makes it easier to trade. From an investor standpoint, this is really bad.

Finally, if there was a transaction tax, I would probably just plan on moving to trading options or some other more highly levered security. A 3x levered ETF is still only taxed on NAV, so I can get the equivalent of 300 shares of exposure at the transaction tax of 100 shares, taking my break even point in terms of the underlying from $0.12 to $0.04. In short, I already have thought of plenty of ways to adapt to a transaction tax, and it might even make it possible to make even more money! And there are lots of other ways to avoid the problem that I haven't described.

On the other hand, a transaction tax really hits retail investors. It will probably serve to double the transaction costs for most retail investors. After you taken into account the widened bid ask spread, it will probably end up tripling the cost the commissions from maybe $8 to $21 if not more.

In short, trading taxes are not going to effect people who are essentially making markets. They will effect the retail investors and are essentially a stealth tax. They could kill unsophisticated retail daytraders.

In London they actually have a transaction tax on stocks:
http://www.lowtax.net/lowtax/html/offon/uk/uk_gotaway.html

All it has done is shift trading from stocks to derivatives.

On data feeds: You can get real time feeds for stocks, futures, and forex at the tick level for thousands of entities for around $200-$300 a month depending on the exact options. This is not outside of the range of a sophisticated retail trader. The internet has dramatically cut costs of data and continues to bring it down.
posted by An algorithmic dog at 6:19 PM on May 15, 2010

... is to remove the computers from the trade to reduce the velocity and gamemanship...

you have to be kidding. What, you yearn for the good old days when you would send an order to the AMEX and it would execute for 50 cents more than you paid?

heres the thing - there are a million ways to invest in the stock market. We've got Lenny Dykstra for crying out loud touting his DEEEEP in the money options - we've got IBD getting everyone into the latest momo stocks, money managers pushing "buy and hold and let me take 2% off the top", and whatnot. HF is just another strategy, and one with surprisingly little barriers to entry. What exactly are people worried about? Market impact? I think Bill Gross has a lot more impact when he talks his book, or one of the Cramerites that blindly follow him into topticking something.

What I think everyone is really worried about is they see goldman putting up 10Bil+ qtr's trading, and they assume its from HF. Trust me, its not. Want to know what its really from, and the real crime here? The fed's zero-interest rate policy lets banks let a 0% from you and I, and loan it back to the govt risk-free at say 2%, with 40-1 leverage. thats the easy, bullshit trade. fighting it out in the trenches of the HF world is nothing like that.
posted by H. Roark at 6:50 PM on May 15, 2010 [2 favorites]

15μsec? Really? Light itself travels about 5 km in that time, never mind things like router lag and the like.

What this is saying is that the most important thing in the trading world is your physical network distance to the NYSE and NASDAQ trading computers. You want to beat this guy? You get 3.3μsec every km closer to the trade than he is.

That's because they colocate their stuff in the same server rooms as the exchanges themselves.
posted by delmoi at 7:56 PM on May 15, 2010

no one is trying to explain why it is useful

HFT supporters have proposed that trading algorithms be loaded some specified time before they start trading, diluting the possibility of inside information being used.
posted by StickyCarpet at 9:17 PM on May 15, 2010

As was pointed out some time ago here: "...[W]hen robotics is mature, robots will operate on a completely different timescale."

Also from the same post..
posted by bz at 9:32 PM on May 15, 2010

There's lots and lots of problems in Wall Street, the biggest being the expectation of bailout in the case of disaster; with a more or less explicit guarantee, and essentially unlimited free money from the Fed, there's no limit to the risks those guys can run, no limit to how big a hole they can dig, leaving us mired in the pit while they skate away laughing.

That whole industry is a tumor on the real economy, but that's primarily because of the exotic financial structures and derivatives that they can use to get leverage and extract wealth from the system. And, with that guarantee, those derivatives just get weirder and weirder and more and more dangerous.

But I'm not real convinced high frequency trading is a problem. It's what we've always done, faster. We've been trading stocks for, what, two centuries now? If trading once per minute is okay, I don't see that trading once per second, or a hundred times a second, is fundamentally that different. If there's a bug in someone's system, and the market plunges or skyrockets in an unusual way, that's an opportunity for other traders to fix the problem, and profit by doing so.

As a human trader, you need to be very careful about placing 'market sell' or 'market buy' orders, in case you get caught in a weird program-trading crosscurrent, but beyond that, I just don't think it's anything to worry about.
posted by Malor at 10:31 PM on May 15, 2010 [5 favorites]

if you add in a random delay you still want to run as fast as possible because you want to start the timer for the delay as soon as possible

Yes, but your (say) 5ms edge on starting the timer is so small as to be irrelevant when the timer will run from a random value from 1 to 1000 seconds. Without the random delay eriko is suggesting, if A has an advantage of that size over B almost every time, then A will win almost every time.

But with eriko's artificial delay, A will win just barely over 50% of the time. Much more fair, right? Sure, people will still try to start the timer as fast as they, but it won't have nearly as big an effect.
posted by magnificent frigatebird at 10:32 PM on May 15, 2010

Oh, and: insider trading is still insider trading, no matter how fast it happens, and should be prosecuted as such.
posted by Malor at 10:34 PM on May 15, 2010

(As fast as they can. Whoops.)
posted by magnificent frigatebird at 10:35 PM on May 15, 2010

Nelson wrote: "Man, life sure was better in the 80s, back when there was a single market maker for each stock and he could simply pocket the $0.125 spread in bid/ask prices one very trade."

I agree. If only we'd also made it so the guys making the $0.125 can't do their own prop trading.

An algorithmic dog wrote: "It may make sense to say that "HFT enhances liquidity,""

I'm trying and failing to understand how someone doing front running enhances liquidity. Or even HFT in the general sense. Why? Because the entire point of HFT is to buy a stock and then sell it as quickly as technologically possible. So from one perspective, yes, you're adding liquidity (you have money and are looking to buy some stock). But look at it on a longer timescale and it does absolutely fucking nothing but transfer money from people whose trades are executed more slowly to people whose trades are executed more quickly.

Someone is trying to create a favorable frame of discussion for their side of the argument. I'm not ready to buy into it just yet.
posted by wierdo at 10:59 PM on May 15, 2010

I think a good deal of the anger about HFT comes purely because no one is trying to explain why it is useful except by saying things about "liquidity."

A good deal of the anger comes from the fact that the worlds financial system is obviously broken, but the people who run it pretend that this isn't true, and that outsiders don't understand the system well enough. All the people defending HFTs here are basically saying they are a good thing because they allow markets to run smoother (while conveniently making people shit loads of money off transaction fees). But I don't really care about the ability of the market to run smoothly, I want the financial sector to shrink as a fraction of our economy because the vast majority of financial activity is not useful in any way except making money for people in the financial industry, and it is a vast waist of talent and effort.
posted by afu at 11:18 PM on May 15, 2010 [2 favorites]

This isn't what's causing that problem, afu. It's the flood of wealth tokens that are being emitted from the central authority, along with the explicit guarantee of never letting anything blow up. So more and more of the economy moves toward manipulating wealth tokens, taking on higher and higher degrees of risk to extract more and more profit, instead of creating wealth.

HFT, in comparison, is basically the zit on the hippopotamus.
posted by Malor at 11:58 PM on May 15, 2010 [1 favorite]

I basically don't understand any of what you are saying. I'm so dense when it comes to the markets.
posted by Lleyam at 1:36 AM on May 16, 2010

Wierdo: Without people who trade on short timescales you could not sell, say 5000 shares without another long term investor who wanted to buy 5000 shares. Yes, you are right, you are paying people for doing nothing but buying and selling as quickly as possible. This is not a risk free activity, so some compensation is required. But you are in essence paying for the service of being able to buy and sell your stock when you want to. If you had to wait for someone to want to buy exactly 5000 shares of GE to sell your 5000 shares of GE, you could be waiting a very long time, and worse, the price of these transactions would be all over the map because the price would reflect the speed of which the seller wants to sell or the buyer wants to buy far more then the actual value of the company.

Afu: I'll tell you why I intend to go into the financial sector when I graduate. The financial sector is a true meritocracy. If produce great results, you get compensated extraordinarily well, but also you get treated (by other people) very well. It is the second one that is more important. Compare that to the R&D sectors of the economy where scientists consistently get treated like crap by everyone. If I went into academia I would look forward to doing graduate school, where I get treated badly. Then I have to find an academic job, when universities keep having their budgets cut. And if I manage to find a tenure track job and edge out all the other applicants, I still have to compete to 1) get tenure 2) get grants. And everyone knows that there is entirely too little grant money to go around and only conservative projects get funding. The reason talent and effort is going into wall street is because wall street has figured out how to recruit talent. The answer is surprisingly simple: if you treat talent with respect and at least pretend to value them, that is way more appealing then the R&D route. Society has so marginalized academia and R&D that no sane person would do research anymore. I completely agree, it is a major travesty that so much talent goes to wall street. Unfortunately this is a problem of our own making. If society actually cared about solving the problem they would start doubling and quadrupling university and research budgets instead of complaining about mathematicians on wall street.
posted by An algorithmic dog at 5:06 AM on May 16, 2010 [3 favorites]

An algorithmic dog wrote: "Wierdo: Without people who trade on short timescales you could not sell, say 5000 shares without another long term investor who wanted to buy 5000 shares. Yes, you are right, you are paying people for doing nothing but buying and selling as quickly as possible. This is not a risk free activity, so some compensation is required. But you are in essence paying for the service of being able to buy and sell your stock when you want to. If you had to wait for someone to want to buy exactly 5000 shares of GE to sell your 5000 shares of GE, you could be waiting a very long time, and worse, the price of these transactions would be all over the map because the price would reflect the speed of which the seller wants to sell or the buyer wants to buy far more then the actual value of the company. "

Thanks for the response. Why was this not a serious problem 30 or 40 years ago? And why do we consider it a problem today?

I thought traditional practice was to break up large blocks of shares to be sold and sell them in bits and pieces so as to avoid driving the price down, anyway.
posted by wierdo at 1:17 PM on May 16, 2010

Everyone loves the fact that the loads on ETF's are ultra-low - thank your ETF's HF dept! Or would you prefer some fucker taking 3% off the top of your S&P tracking fund?

I don't think there's any reason to ban HFT entirely, but the practice of front-running bids makes the market less transparent, not more, and is really just exploiting a loophole, IMO. The bid process should be transparent. Of course you can place contingent orders, but allowing front-running is not really benefiting anyone except the HFT at the expense of those who can't see it happening when they place an order.
posted by krinklyfig at 4:13 PM on May 16, 2010

I'll tell you why I intend to go into the financial sector when I graduate. The financial sector is a true meritocracy.

Ha!

Hey, I like trading the markets, too, but you're in for a rude awakening. The financial sector is full of sharks and opportunists, not to mention cronyism. That's why I don't work in the financial sector. If you're good at it and really enjoy it you may be better off on your own rather than trying to climb those ladders, or at least more fulfilled, but whatever floats your boat.
posted by krinklyfig at 4:18 PM on May 16, 2010

Thanks for the response. Why was this not a serious problem 30 or 40 years ago? And why do we consider it a problem today?

It was a problem, but it was considered arcane to those who didn't trade directly on the markets. The bid-ask spreads back then would be huge, typically a couple dollars on a lot of stocks. This means liquidity is low enough that the market hasn't determined a price between the two, so getting an order filled or sold is a lot more tricky if you have a particular price in mind, even if it's somewhere in the range. With high liquidity, you can get an order filled or sold at the price you want, as long as it's moving around that price, because the market has made a much more precise determination as to the current price, in other words a tight bid-ask spread. But the way it was done was just considered another aspect of the market, another potential inefficiency to be exploited if you knew how. It's still exploited today, just in a different way, in the direction of exploiting very minor inefficiencies with very fast trading. This is OK. I still don't think front-running should be allowed, because it's not taking advantage of an inefficiency so much as putting themselves in front of an already established transaction to shave a penny off.
posted by krinklyfig at 4:28 PM on May 16, 2010

OK, I guess what I'm not getting is that the markets seemed to work fine in the 60s and 70s despite the lower liquidity and the greater difficulty in selling at an exact price. Why would that system not work today? Is there a reason beyond cutting into trading profits?

Speaking of cutting into trading profits, why is it that banks are allowed to do their own trading? It seems like what we have now that Glass-Steagall is gone and the investment banks are now commercial banks is both dramatically unfair and dangerous, what with the banks being allowed to borrow directly from the Federal Reserve and do their own trading using those funds.
posted by wierdo at 4:46 PM on May 16, 2010

OK, I guess what I'm not getting is that the markets seemed to work fine in the 60s and 70s despite the lower liquidity and the greater difficulty in selling at an exact price. Why would that system not work today? Is there a reason beyond cutting into trading profits?

For the same reason we don't pick up the phone and go "Operator give me Murray Hill 54309".
posted by Talez at 7:00 PM on May 16, 2010

Krinklyfig: If you put in a nonmarketable buy order, and I take that to mean there is increased demand and thus I should be more aggressive when buying, that is not front running by any reasonable definition that I would use. I do not believe that front running of orders by HFT is a real problem. I think there are a lot of people who think that they should be able to buy a million shares of IBM without moving the price. But why should institutional investors be exempt from the laws of supply and demand?

Its been my experience that life is full of sharks and opportunists, not to mention cronyism. Trading has the really nice property of being able to argue pretty objectively about if you are right. In general the compensation structure (both nonmonetary and monetary) in finance is much more meritocracy based then anywhere else. I mean compare it to academia where all postdocs (or equivalent in industry) roughly pay the same no matter how good your research is.
posted by An algorithmic dog at 8:56 PM on May 16, 2010

OK, I guess what I'm not getting is that the markets seemed to work fine in the 60s and 70s despite the lower liquidity and the greater difficulty in selling at an exact price.

Actually, the markets were infamously bad in the 70s. At one point, some market makers were weeks behind in reconciling the orders they were managing. Many went out of business when they discovered that they were deeply on the wrong side of some transactions.

It would be interesting for a historian to go look at the mutual influence of financial markets in the '70s and '80s. I suspect that it would be found that the increasing power of minicomputers lead brought the markets back to reasonable efficiency, and the powerful micros (what we call PCs today) that followed drove experimental computational strategy. This, of course, drove computer hardware purchasing, which benefited the markets again. If I could have a beer with Michael Bloomberg, this is what I'd ask him about.

Liquidity is a good thing, and we couldn't handle the volume we transact today without computers managing the order books. We get additional liquidity when high frequency traders are moving in and out of positions faster than humans can think.
posted by b1tr0t at 9:12 PM on May 16, 2010

b1tr0t wrote: "we couldn't handle the volume we transact today without computers managing the order books."

That's not really an argument in favor of anything other than computerized trading platforms, which have nothing to do with what we allow people to do with those computers beyond that.
posted by wierdo at 11:37 PM on May 16, 2010

Once you have the computerized trading platforms, the speed comes naturally, because that's what computers do. Faster execution times means more liquidity, which means smaller bid/ask spreads and an easier time filling orders at a particular price, which is better for true investors and speculators, i.e. traders, alike. And as others have pointed out it's basically HFT which makes things like MeFi-favorite Vanguard's low-fee ETFs possible.

The "flash order" business seems a bit shady, but it doesn't seem to actually be insider trading and doesn't seem to be all that different from a lot of stuff that goes on with payment for order flow. So if you're going to go after flash ordering you might as well go after the whole business. However, there are some decent arguments for allowing PfOF on the basis of increased liquidity and narrowed bid/ask spreads, and most of the arguments against hinge on its effect on small or unsophisticated investors, which might be better solved by educating those investors rather than nerfing the market for their benefit.

There are a lot of things screwed up with the market and its major players, but HFT doesn't seem like it's one of them, or at least like it's anywhere near the top of the list. It seems suspiciously like a political red herring and a distraction from the real issues, which are the continued existence of TBTF entities and their implicit and continued taxpayer subsidy, and the cheap money policy pursued by the Fed alluded to above by Malor and others.
posted by Kadin2048 at 8:43 AM on May 17, 2010

I just want to point out that any retail investor can get paid for his order flow. There are plenty of brokerages that are open to the retail investor that will let you unbundle the brokerage commision from the exchange costs, and then if you offer liquidity (use nonmarketable orders) you get money back from the exchange. While usually you still have commisions because of the cost of the broker, they can be subpenny per share on a roundtrip basis. This is usually not worth the effort even for a casual retail daytrader though.

Many of the things people complain about are available to retail traders who care to spend the effort to do them. The problem is that people want the market to be something where you put money in and get significantly more then the risk free rate over time while taking no risk. There is a reason its called the risk free rate.
posted by An algorithmic dog at 11:52 AM on May 17, 2010

Thanks for your insight, algorithmic dog. It's been helpful, but I have to say, your comment about research vs. Wall Street made me a little sad, but good luck.

For other people going into this field because it's a meritocracy, I'd ask this: If you want to do work in which you are lauded for your merit, does it make sense to spend your days tricking a system for short term gain? Certainly, you will be praised for your ability, but is there really any merit, in the original sense of the word, at all in what you've worked so hard to do?

And remember, there's other directions to go in besides trading and academia. You could be an engineer or business person, among many other professions in which will respect your ability and skill.

I can definitely understand wanting to make a shit-ton of money, though.
posted by ignignokt at 11:35 AM on May 18, 2010 [1 favorite]