Hacker News Reader

Comment by ChristianMarks | Hacker News Reader

I'll believe that HFT adds liquidity to the market when the typical start of employment to retirement period is 15 milliseconds.

View on HN

An expected and uneducated statement. It's simple really: any participant who rests an order on the book is adding liquidity. Period. Many HFT strategies are largely passive (market making, for example) and therefore entirely rely on resting orders (as Chris' article points out). Therefore, they add liquidity.

Now, if you want to get into a discussion as to the quality of that liquidity then you have to go educate yourself a bit more. Once you do you'll find that the liquidity being added is not toxic nor fleeting. However, what you and most others don't understand is that it is also significantly more informed passive liquidity than it used to be.

This may lead folks to believe that the liquidity is fleeting because they can cancel away before you can execute. However, that simply means they are better and faster than you, not that they are doing anything wrong.

Why, then, don't we simply require them to rest their orders for a longer period of time? Simple: you want tight spreads. As spreads shrink market making becomes less profitable. When market making is less profitable market makers have to be more risk averse to avoid getting adversely selected and losing money. More intelligent models (they become more informed) and less latency (they can leverage that informedness to avoid being selected) is how market makers stay profitable in a market where the average spread is 1 penny.

How about rather than adding regulations, just discretize the market's clock? Would a market that performed a trade-resolution tick once every 100ms lose anything important? I believe arbitrage improves efficiency, but I'm less clear on the benefits of extremely high temporal resolution. Currently that's what the incentives encourage optimizing for, among other things. If you predict a market movement 10ms ahead of time, you can profit from that brief 10ms temporal-arbitrage window. And like any arbitrage, that does indeed improve the pricing signal in an absolute sense, in this case by taking a price change that was 10ms later than it "should've been", and moving it up in time via your trading. But it's not clear that 10ms-level pricing inefficiencies are actually something particularly important to smooth out via arbitrage, when you can just define them out of existence by going to a discrete-time market (which are fairly well-studied in the mathematical literature).

I think if you consult the literature you'll find that the prevailing trend is for markets to become more continuous rather than more discrete. Continuous markets help with true price discovery. Our markets are continuous in time right now and we've pretty much squeezed the spreads to their maximum, so while prices aren't continuous they are about as close as we can get w/o eliminating the profitability of the primary market participant (market makers) that make it function.

You can scheme up any number of possible microstructures that sound interesting on the surface. There's a reason why none exist. Remember, there is nothing to stop you from implementing a time-discretized market place. Current regulations (Reg-ATS) allow for you to do so. You'll find it hard to compete. Some markets do perform large scale discretization for block orders. Read up on POSIT and related ATSs and dark pools.

I read somewhere that non-hft tend to trade much more at the "second" tick. This may have been b/c they were dumb (hard to believe) or because converging at the same time (15:30:10 instead of 15:30:10.175) gives them better chances to finding a match to their sell/buy position

FYI - I wrote the successor to posit. Point in time matching goes out of fad when volatility increases. Why take on the risk of executing a block at a single point in time when you can spread that risk out over the day?

Evidently you have childlike trust in the capacity of the second-tier Russian mathematicians and physicists to get the boundary conditions right.

what exactly is 'true price discovery' ?

can a price be false?

With insufficient trades, you cannot know if a certain price is really what the 'market' wants to pay, just what that one person at one point in time wanted to pay. Overgeneralized, the more trades, the more accurate the price gets.

Discretization often leads to instability, particularly if not done carefully. The prototypical example is naive discretizations of stable ODEs. Of course, all we have are toy models and speculation. The only real way to know is to try it.

If I remember right, a couple of Asian equities markets have some sort of "auction on clock tick" mechanism, so we might be able to look to their experiences.

> Discretization often leads to instability ...

> Of course, all we have are toy models and speculation

These two statements could be construed to be in conflict, especially as I haven't the faintest idea what naive discretizations of stable ODE's is, and Google isn't of much assistance in providing an explanation that I can digest.

I didn't intend them to be in conflict - I listed a toy model, and acknowledged the limitations of it.

An explanation of ODEs and instability (some math background is obviously required): https://en.wikipedia.org/wiki/Stiff_equation

You lead with a fairly strong statement presented as a known fact and then later indicate it's just a model.

To me the idea of slowing things down sounds vaguely sensible, but I also know that I don't know enough to really judge. So it sounds like no one else really knows either, but the entrenched players like things the way they are, which makes sense, because the ones who have the money to be at the table running things are probably also the ones who have the money to throw at HFT and come out ahead.

It is a well known fact about dynamical systems in general.

I.e., if you ask me about the stock market, a flight control system, an electric circuit, a biological system and a computer network, I'll tell you that continuity gives you a better shot at stability than discretization in most of them.

I.e., if you don't know what you are talking about, lean towards continuity. If you do understand things, then explain the mechanics and back it up with empirics.

The hypothesis being advanced by people proposing point-in-time clearances is not necessarily that it will make the market smoother. It's that it will free up a lot of resources for other purposes.

For what it's worth - and not to distract from your real point - Exponential (or the more general Lyapunov) stability are a better explanation of instability. Stiffness is more a property of the method used to solve the ODE, rather than of the ODE itself.

Stiffness is a property of certain dynamical systems, which certain ODE solvers are better suited for. That said, "stability" is an ugly word in numerical methods, particularly for differential equations, as it could mean a number of things. I think this is the source of the confusion here.

It seems that investors trading on an exchange operating at discrete time intervals would be at a arbitrage disadvantage if other exchanges offered continuous trading in the same shares.

So a demand for markets that are liquid and efficient rewards trading that is continuous in time over time-discrete trading, and firms are competing hard to shave milliseconds from latency with co-located hardware and faster trans-oceanic cables.

On the other hand, pricing is limited by regulation to penny increments, with some pressure to telegraph smaller increments.

Are we likely to see a push for trading with offers and trades denominated in milli-cents or micro-cents? Would smaller pricing increments be welcomed by traders or regulators or neither?

Are any markets in the world trading at tiny denominations? (NB: assumes pricing in integer quantities of tiny denominations)

I don't see a reason that "normal" investors would be at an arbitrage disadvantage if trading on an exchange that was discretized to, say, 100ms. Arbitrageurs would still have an incentive to keep the prices on the "slower" exchange in line with the "continuous" exchange's prices, trading at the discrete ticks to exploit price differences. So, price differences past an epsilon shouldn't persist for more than a few ticks, which is more than fast enough for most non-HFT investors, who don't typically make trading decisions with anything close to sub-second precision anyway.

I think a bigger question mark with my admittedly offhand proposal is whether discretizing might actually increase gaming opportunities, since the change could do non-obvious things to the strategy space. On the other hand, it might also remove existing market-gaming opportunities (which are fairly poorly understood, and axiomatically assumed not to exist by the idealized equilibrium analysis economists typically use). Alas, current game-theory solvers don't scale up anywhere close to well enough to give solid answers in either direction (real markets are a bit more complex than 4-participant games iterated over 10 timesteps...).

>I don't see a reason that "normal" investors would be at an arbitrage disadvantage if trading on an exchange that was discretized to, say, 100ms. Arbitrageurs would still have an incentive to keep the prices on the "slower" exchange in line with the "continuous" exchange's prices, trading at the discrete ticks to exploit price differences. So, price differences past an epsilon shouldn't persist for more than a few ticks, which is more than fast enough for most non-HFT investors, who don't typically make trading decisions with anything close to sub-second precision anyway.

Sure, but what you're missing is that the slow exchange's prices are always going to be worse. Not a lot worse - probably only a penny either way on some ticks, and zero on other ticks - but a little. So why would any investor with the choice ever choose to trade on the slow exchange rather than the continuous one?

"Worse" in what sense? Won't the prices deviate pretty randomly around the true price, sometimes behind a penny higher than the continuous exchange, and sometimes a penny lower, making it basically a wash?

No. If the discrete exchange makes it possible to withdraw an offer in between ticks, then the HFT guys would do that every tick - so you actually wouldn't get a price (or you'd get a very wide spread), and it would basically fail to be an exchange. So let's assume any order you have on the book stays there until the next tick's auction. In that case the HFTers are going to give a much higher spread, because if new information comes in at 3.2s and the stock is suddenly worth more than it was at 3s, but they can't withdraw their sell order until 4s, then obviously they lose money. So occasionally you'd get a better price on the discrete exchange (when the market moves further than the spread in the space of one tick) - but the HFTers would be terrified of this situation, and make their spreads wide enough that they think it's impossible. So 99% of the time, you'd get a worse price.

Yep. One daily cross would be in some ways more fair. But it would also wound price discovery significantly.

SecondMarket is a start up pushing in this direction. The company whose shares are trading set the trade frequency. Sometimes once per quarter, once per day, whatever they want. Obviously, this is much less liquid.

There's no counter argument here nor does this come across as informed or educated. All you've done is talk about the process with as many technical words as you could without actually saying why 15ms liquidity gives us anything.

Apart from mini-crashes. And massive profits for companies that create nothing and actually were the cause of the recent financial meltdown.

Which is why many of us sit here scratching our heads or make sarcastic comments like the op.

I read his comment without any trouble. I don't work in finance but I'm perhaps a little more familiar with it than the typical HN reader. It certainly didn't come across as an attempt to snow the thread with jargon.

A careful reread generates this list of domain-specific terms:

resting orders, liquidity, HFT, passive trading, cancel, execute, spread, market making.

Most of these mean what they sound like they mean. If "liquidity" is complicated, it's also the topic of the whole thread. "Passive trading" is probably the closest thing to jargon in the whole comment.

It would be better if nobody made sarcastic comments, wouldn't it?

His complaint was that the post failed to explain why 15ms liquidity has any value, not the terminology. I didn't detect any sarcasm there.

I think if you read his comment just a little more carefully you'll see what I was responding to. I don't have a "side" in this particular debate.

I totally don't understand what you're on about, sorry. Paul summed up my point perfectly.

You seem to be forming a habit of ascribing things to me I didn't say. It's getting a little embarrassing/annoying.

"All you've done is talk about the process with as many technical words as you could"

> Apart from mini-crashes.

The Flash Crash was caused by the NYSE's faulty time-stamping system. It essentially broadcast false information that caused traders to think there was an arbitrage opportunity. This was entirely the NYSE's fault. See here for details: http://www.nanex.net/20100506/FlashCrashAnalysis_Part3-1.htm...

> And massive profits for companies that create nothing

They prevent you from paying even more to corrupt exchange specialists. This creates billions of dollars a year in value. They also have a neat ability to help the market recover from irrational crashes in 20 minutes, instead of lingering there for days or months while everyone is too scared to provide liquidity.

> and actually were the cause of the recent financial meltdown.

HFT is completely unrelated to the financial meltdown. The financial meltdown was caused by a housing bubble. HFT market makers don't even trade mortgage-backed securities. Those were traded over-the-counter (basically via phone calls) by investment banks.

If you don't know what these words mean, then look them up instead of complaining about it. If you want to make a point, it is your obligation to understand the subject matter. Don't argue about things you don't understand.

Apart from mini-crashes. And massive profits for companies that create nothing and actually were the cause of the recent financial meltdown.

Could you explain how low latency trading caused either of these things?

I'm particularly curious what relation you see between low latency trading in various computerized exchanges, and a financial meltdown primarily involving OTC trading in heterogeneous derivative-like equities.

By the way, to answer the question, 15ms of liquidity is just as good as 100ms or 1 sec of liquidity. The ordinary trader doesn't care about latency, latency is purely a game between HFT's to see who will receive the profits from selling liquidity.

He's not imputing a direct connection between HFT and the derivatives meltdown. He's implying guilt by association, since HFT systems were deployed by the big investment banks. Of course, so was Powerpoint, which itself probably played a bigger role in the meltdown, so maybe we should ban that too.

"By the way, to answer the question, 15ms of liquidity is just as good as 100ms or 1 sec of liquidity. The ordinary trader doesn't care about latency, latency is purely a game between HFT's to see who will receive the profits from selling liquidity."

I think this is the crux of why people think HFT is a net harm to the economy. They're devoting many millions or maybe billions of dollars to this stuff, and presumably making a hefty profit for it to be worth it, and it actually makes zero useful impact on the market? Yeah, that pisses me off.

I have no problem with the financial sector being overpaid if they actually provide value to the economy. Capital allocation is a necessity and nobody's claiming that soviet-style planning would be superior. But HFT looks to me like they're just sucking little pieces out of everyone else's trades, while providing no value to society.

Without getting into whether we're paying too much of a premium to the financial industry to make the economy more efficient (for instance, by enabling you to get a 15 year mortgage on a house with 20% down instead of a 5-year mortgage with a balloon payment):

HFT makes a profit off the trading spread. The spread is "friction" in the market. It's the gap between what one person wants to spend and another wants to take in for the same good. Every time you make a trade, you cross the spread and thus pay a fee; you're "buying liquidity".

HFTs compete with everyone else trying to make a profit from the spread. The people being harmed most directly by HFTs are themselves trying to extract a fee from normal market participants. Moreover, by competing very effectively, the HFTs are narrowing the spread. They are effectively bidding the price of liquidity down.

It would make sense to be mad at HFTs if the participants they were replacing were mom's and pop's just trying to make a reasonable living while looking out for the good of the whole market. But that's not who HFTs are displacing. The human traders who profited from spreads in the 1980s were often crooked as a carton full of fish hooks. The markets before automation (and the OTC markets today) are riven with people running strategies to skim money off real buyers and sellers.

The more you know about how the markets work and what the incentives of all the players are, the harder it becomes to feel any outrage about HFT.

Yeah, but you'll notice the quote at the top of my post was basically saying that that friction doesn't matter at small enough intervals.

I get that you don't want too much friction in the market, you definitely don't want excessive surpluses or shortages of actual goods and to some extent that applies to securities.

But I see zero reason to care about the price of IBM in even a 1-minute interval. A little friction? Who cares, we're not impeding shipments of goods here, and even if we were, the 1-minute spread on a 6 month lumber future is similarly unimportant, to me at least. That's without even getting into seconds and milliseconds.

Basically, I'm saying that "but, liquidity!" gets less and less convincing as an argument as the time intervals shrink. At a certain point, we're just funding a very expensive and stupid zero-sum game with pennies out of all of our 401k accounts.

Doesn't it make more sense to make a firm argument for "why not liquidity"?

Who exactly is funding this expensive and stupid zero-sum game? Exactly what effect does this game have on your 401k? Isn't your 401k invested in index funds? Don't they trade infrequently and in huge, huge blocks? A no-load S&P 500 tracking fund is not HFT'ing the components of the S&P 500.

Doesn't it make more sense to make a firm argument for "why not liquidity"?

Sure. Why not strip clubs? And blackjack? I acknowledged that liquidity has value it's not an end -- it only has value in as much as it provides value.

I don't fully understand who's funding it, but I know the money comes from trade volume, and a big chunk if not the majority of the money on the market is there from institutional funds, 401ks, pensions, etc. Those that are actively managed get nibbled away at. Those that hold for longer (index funds do rebalance) get nibbled away at less frequently.

What's your hypothesis for where the trade volume that they make money from originates? They're just taking money from the rest of the financial industry?

I don't know where to start. I'm not saying you're wrong, just, I don't know where to start responding. And please feel free to mentally append "as I understand it" to each of these.

1. Funds of actively traded securities are nibbling away at you with or without HFTs. Actively traded funds are evil.

2. HFT market makers aren't nibbling at actively-traded funds. They're nibbling at other market makers. Like I said previously: an HFT market maker is bidding the price of liquidity down, not up.

3. Your 401k is almost certainly not invested in a fund whose strategy is passive trading and selling liquidity. Your funds have positions in the market. Passive traders do not take positions. Your 401k benefits (very marginally) from the tightened spreads created by HFTs. The people who don't benefit are day traders.

4. I have no idea what "strip clubs" and "blackjack" have to do with any of this.

Mostly agree, except:

> Actively traded funds are evil.

Actively traded funds are not "evil." Actively traded funds are the only reason you can buy an index fund and actually expect a decent return. Without active traders, the market would be inefficient, i.e. things would not be fairly priced.

"We" aren't funding anything. We (401k owners) benefit minutely from the reduced friction, but it doesn't cost us anything; if hedge fund A gets their latency down to 1ms and so they get our buy/sell rather than hedge fund B that's running at 5ms, so what? It's not like A charges us any more than B did, and why should I care whether A or B gets the profit?

What's happening is commoditization, which you should be happy about - the only value the HFTs are destroying is their own profits. It's just like if PC maker A can sell hard drives 5 cents cheaper than PC maker B by building a faster robot - sure, I didn't really care whether my hard drive was $120 or $120.05, but I don't lose anything - why do I care whether A or B gets the profit?

"But HFT looks to me like they're just sucking little pieces out of everyone else's trades, while providing no value to society."

This suggests that market participants of any type provide "value to society". Their is only one reason any entity participates in the market, and it has nothing to do altruism, and everything to do with making money for themselves or their clients. Capitol allocation to companies going public is over once the shares are issued via the IPO auction. No transaction on the secondary market results in any money going back to the company that issued the shares. So how does the length of time a firm holds onto a stock before attempting to realize profits define whether or not they provide value?

For mini crashes, I think an argument can be made about HFT potentially making the stock prices, etc… less normal and more fat-tailed, and thus less well handled by statistical models (e.g. http://www.bankofengland.co.uk/publications/Documents/speech...). Quantifying how much, of course, is another matter.

The financial meltdown was just caused by a deviation slightly larger than the one regulators had accounted for.

Regulators were trying to juggle lots of different policies... relaxing underwriting and reserve capital requirements while pumping up mortgages to help make joe sixpack feel rich so he'd support the war on terror. Such widespread regulatory corruption is unprecedented in the first world.

It is/was a horrible mess, but HFT has nothing to do with it. The bizarre indignation people feel about HFT shows just how gullible people can be to "news" stories.

> However, that simply means they are better and faster than you, not that they are doing anything wrong.

So it sounds like "the only winning move is not to play" if you are a small fish, because you're going to get screwed by people who invest a lot in this stuff?

At least that's what I'm reading in that sort of message. I'm quite willing to be proven wrong, given I don't know much about the subject.

By "not play", let's be clear that with respect to HFT, the game is "active speculative trading". If your intent towards the market is to buy and hold a position, or to liquidate a long-term holding, the fact that there's a billion dollar game of Core Wars happening at the match engines is irrelevant to you except that you will probably pay a little less to execute your trade.

So where's all that cash being thrown at HFT coming from and going then?

Also, just out of curiosity, you seem fairly defensive of HFT - any particular reason? I feel fairly ambivalent about it. I don't like the idea of trying to regulate stuff like that away, but it also appears to be fairly worthless / zero sum after a certain point, to the untrained eye. I'll freely admit I don't know a great deal either, though.

Presumably from many of the same places GETCO's (significant) revenue comes from: by selling liquidity services. Market makers aren't a new concept.

I find financial tech fascinating, have had a lot of projects that involved attacking them (sometimes at a financial-domain level; ie, constructing technology-centric frontrunning schemes by leveraging software bugs), and so have had a chance to learn a bit about the field. Not as much as others on this thread.

I am defensive, it's true, but not about HFT (if exchanges adopted technical countermeasures to prohibit HFT, that'd actually be a win for me: one more thing for my team to test!) What bugs me is the knee-jerk comments, often from people with severely broken mental models of how trading markets work, piling on with "HFT flash-crashed the CDO meltdown" stuff.

The strategic and technical games that go on in the markets are engaging to watch, but they're basically ways of winning the most money (from the investor's/speculator's viewpoint). The more complex they get, and the higher the bid prices for talent which can win these games, the more the talent and energy is diverted from fields that could advance society more (say, basic research, or new product/tech development). So you get more smart kids going to biz school with an eye towards a place on Wall Street, instead of a broader distribution. That's my view on why financial jobs, and possibly large-scale investment, in general should be less profitable - though how to achieve that is a very open question (not sure how well capital gains taxes would work).

The cash comes from traders/investors who are impatient. If you want to trade immediately instead of waiting and hoping for someone else to hit your order, you have to pay the spread. Market-makers make money by allowing impatient traders to transact with them, and then holding the inventory until other impatient traders want to buy/sell it.

This requires the market-maker to take a risk. The spread compensates them for the risk. The size of the spread is (currently, with HFT) set by competition between market-makers. Before HFT, there was much less competition between market-makers, and spreads were hence much wider. Back then, this resulted in the transfer of significant wealth from investors to exchange specialists. In modern markets with HFT market-makers instead of specialists, these transaction costs are much lower, which saves you and your pension fund money.

Example:

Suppose the national best bid on stock ABCDE is $15.17 and the best offer is $15.18. The "spread" is $0.01. (These limit orders were almost certainly placed by market-makers using HFT.)

If you want to buy ABCDE, you can do one of two things: - You can place a limit order to buy at 15.17 and wait and hope that someone sells some to you. - Alternately, you can place a market order that will cross the spread and buy at 15.18 instantly.

The market order protects you from the roughly 50-50 chance that ABCDE prices start increasing and your order never gets filled. It costs $0.01 per share, which is basically paying the HFT market-maker for liquidity (the ability to trade immediately).

Before HFT, the spread might have been $0.05 or even $0.10. You would still have the same two choices, but instead of $0.01, you would have to pay some human specialist $0.05-0.10+ if you wanted immediacy. His father and grandfather would have also been specialists, and his bonus would have been several million dollars that year. HFT market-makers simply out-competed these parasites. There is no longer a monopoly on market-making, so liquidity has gotten cheaper.

Correspondingly, it is cheaper for you to trade (ditto for mutual funds, pension funds, hedge funds, etc.). This allows you to keep more of your investment profits.

I'd like to get a better understanding of this liquidity.

Assume the subset of market participants whose holding period is always > 1 day ("slow"). Assume another subset of market participants whose holding period is always < 1 minute ("fast"). Ignore everyone else in the market.

Do these "slow" participants see any practical liquidity benefit from the "fast" participants? I don't see how.

Financial guys envision these "slow" traders as grandmothers, but remember that most entrepreneurs have holding periods of > 5 years.

The entire point is that there are intermediate holding periods. Think about intraday speculators and statistical arbitrage funds who trade on scales of minutes to hours. They bridge the gap. Further out you have fundamental hedge funds, individual traders, pension funds, etc.

Thanks for the comment - this was the question I most wanted to understand. I realize that durations are on a continuous scale. I personally know people who operate across the full range of durations.

But it's still far from clear to me that grandmothers (aka entrepreneurs) see any liquidity benefit from folks with short holding periods. I'm not saying it's a problem if grandmothers like me dont see a liquidity benefit from HFT. I'm just asking if it exists.

> It's simple really: any participant who rests an order on the book is adding liquidity. Period.

If a stock is trading at $100.00, and I put a buy order on the book at $0.01, how have I added liquidity?

> This may lead folks to believe that the liquidity is fleeting because they can cancel away before you can execute. However, that simply means they are better and faster than you, not that they are doing anything wrong.

The question is not whether it's right or wrong, but whether it is beneficial. I don't buy that HFT is intrinsically evil, but there seems to be little convincing evidence that it's actually beneficial, either. The system as you've described it basically involves scraping pennies off transactions that would have occurred anyway. That's not intrinsically evil, but it doesn't seem to actually benefit anyone except the HF trader.

> Why, then, don't we simply require them to rest their orders for a longer period of time? Simple: you want tight spreads. As spreads shrink market making becomes less profitable. When market making is less profitable market makers have to be more risk averse to avoid getting adversely selected and losing money. More intelligent models (they become more informed) and less latency (they can leverage that informedness to avoid being selected) is how market makers stay profitable in a market where the average spread is 1 penny.

What's missing here is an explanation of why we should build a system that enables HFT to be profitable. That's not something I care about. If lower profits for HFT means a better market overall, that's fine with me.

Frankly, in markets where the spread is 1 penny, it seems like there's enough liquidity already.

The system as you've described it basically involves scraping pennies off transactions that would have occurred anyway.

Could you explain how to "scrape pennies" (as in specific mechanics, your explanation should involve limit orders or whatever), and why this strategy doesn't benefit the speculators the HFT is trading with?

> Could you explain how to "scrape pennies" (as in specific mechanics, your explanation should involve limit orders or whatever),

Are you under the impression that you're assigning homework to students? This comes off as really rude.

And no, I'm not going to write up a treatise on HFT for you. I don't pretend that this is an area of expertise for me. FTA:

Most HFTs run a market making strategy. What this means is they play both sides of the table - they take no position on whether a stock will go up or down. Instead, they try to offer securities both to buy and sell. If you want to buy, they will sell to you at $20.10. If you want to sell, they’ll buy from you at $20. As long as their buys and sells match don’t get too out of whack, the HFT will collect $0.10 = $20.10 - 20.00.

What is that if not "scraping pennies"? The entire goal is to jump between a buy and sell in order to skim a bit from the transaction.

e.g. Alice wants to sell at $10.00. Bob wants to buy at $10.05. They could simply trade with each other, but if Eve gets her way, Eve will buy from Alice at $10.00, and then sell to Bob for $10.05, making $0.05 on each share that would otherwise have gone to Alice.

Engaging in this is not evil, but I fail to see how it benefits anyone else.

> and why this strategy doesn't benefit the speculators the HFT is trading with?

Because when a HFT makes money, other parties involved make a little bit less. HFTs don't print money. They extract it from the market.

Also, because stock markets are not exclusively limited to speculators and HFTs.

Are you under the impression that you're assigning homework to students?

I'm asking for an explanation that goes beyond "they make money, and I don't understand how, they must be ripping people off!"

What's missing from your explanation is time. If you actually wrote out a specific explanation involving limit orders (like I suggested), you would immediately have seen that.

If Alice wanted to sell at $10.00 and Bob wanted to buy at $10.05, they would have already traded. The millisecond Bob's order hit the market, he would have filled Alice's order (or she would have filled his, depending on which was bigger). Eve never had any ability to participate.

If Eve played any role, there was a time delay. At 12:20, Eve put an order BUY(price=$10.00, quantity=100, time=12:20) onto the market. Alice chose to sell to her at 12:30 because she wanted to trade immediately, with no risk of her order being unfilled. After buying her stocks, Eve turned around and placed a SELL(price=$10.05, quantity=100, time=12:30:01) order. At 12:40, Bob comes along and places an order BUY($10.05, quantity=100, time=12:40), which is filled by Eve's order.

If Alice wanted to hold out for $10.05 she could have. She chose not to, because she didn't want to run the risk of Bob never showing up. Eve didn't "scrape pennies" from Alice and Bob, Alice paid her $0.05 to take the risk of Bob never showing up.

> I'm asking for an explanation that goes beyond "they make money, and I don't understand how, they must be ripping people off!"

I didn't say that anyone was ripped off. There's nothing inherently wrong with being a middle-man in a transaction. I just don't see that HFTs are adding any real value.

> What's missing from your explanation is time.

You're pretending that Alice and Bob placed their orders 20 minutes apart. That's not the game HFTs play. HFTs are playing games of milliseconds. It's not 12:20 and 12:40. It's 12:20:00.200 and 12:20:00.450. If Eve had taken the day off, Alice would have traded with Bob, and it would have looked just as immediate to both of them. And if there were a high risk that the market would tank immediately after Alice's sell order were placed, then Eve wouldn't have left her buy order on the book anyway.

If your 20-minute wait scenario were realistic, then sure, HFT would be adding meaningful liquidity. But then, it wouldn't be called high-frequency trading, and you wouldn't have written about how "speed matters".

If Eve had taken the day off, Alice would have traded with Bob, and it would have looked just as immediate to both of them.

This comment makes sense only in the context of a message board hypothetical, because it presumes foreknowledge on Alice and Bob's part. Here, Bob showed up. Eve looks like a genius. But it was equally likely that he wasn't going to show up.

A lot of the discussions about liquidity and HFT here seem a little innumerate. They appear to work from a scale where the hypothetical Alice's ask price is "absolute zero". That's not the real scale. Obviously, instead of Bob showing up at $10.05, you're equally likely to end up with Chuck at $9.95.

If you're not equally likely to get Chuck instead of Bob, why are you selling?

> This comment makes sense only in the context of a message board hypothetical, because it presumes foreknowledge on Alice and Bob's part. Here, Bob showed up. Eve looks like a genius. But it was equally likely that he wasn't going to show up.

Equally likely? So HFTs are flipping coins blindly and just happen to make a lot of money because they can flip quickly?

> If you're not equally likely to get Chuck instead of Bob, why are you selling?

I don't understand this question at all. I'm selling because I want to sell my stock. Maybe I'm liquidating assets to buy a house. Maybe I'm speculating that the market is going to tank. Maybe I'm just adjusting my asset allocation. I could be selling for any number of reasons, and I don't care about Bob or Chuck. I just want to sell and get the market price.

If you need liquidity but want to retain your upside exposure, there's a whole class of tradable instruments that does that for you.

If you need liquidity and aren't confident enough in your upside to want to be exposed to the downside, you're happy to have Eve.

Note well: your hypothesis is that Alice should get something for nothing. Alice wants liquidity (ie: no downside exposure) and immediate access to the next significantly better price to hit the market. It must be nice to be Alice! :)

As for your first question: HFTs do not have crystal balls. If they did, Chris Stucchio would be a billionaire.

For sure, HFTs don't have crystal balls. They certainly are able to leverage their market access to give them an advantage, though.

I feel like I need to reiterate that I don't think HFT is evil. I'm just not sure that HFTs really add that much liquidity to the market, and there is evidence that they contribute to volatility (such as the flash crash).

The "Flash Crash" was caused by a single, manually-initiated large block trade:

At 2:32 p.m., against this backdrop of unusually high volatility and thinning liquidity, a large fundamental trader (a mutual fund complex) initiated a sell program to sell a total of 75,000 EMini contracts (valued at approximately $4.1 billion) as a hedge to an existing equity position.

(From the SEC link Chris posted earlier).

I don't know whether I believe Chris that HFTs caused the market to correct much faster, but it seems clear that HFT didn't cause the crash.

The SEC seems to disagree, and says that HFTs added to the drop. HFTs also apparently burned through about half of the trading volume just trading with each other.

> The combined selling pressure from the Sell Algorithm, HFTs and other traders drove the price of the E-Mini S&P 500 down approximately 3% in just four minutes from the beginning of 2:41 pm through the end of 2:44 pm. During this same time cross-market arbitrageurs who did buy the E-Mini S&P 500, simultaneously sold equivalent amounts in the equities markets, driving the price of SPY (an exchange-traded fund which represents the S&P 500 index) also down approximately 3%.

> Still lacking sufficient demand from fundamental buyers or cross-market arbitrageurs, HFTs began to quickly buy and then resell contracts to each other – generating a “hot-potato” volume effect as the same positions were rapidly passed back and forth. Between 2:45:13 and 2:45:27, HFTs traded over 27,000 contracts, which accounted for about 49 percent of the total trading volume, while buying only about 200 additional contracts net.[9]

The SEC paper doesn't even use the word "liquidity" properly, and the authors appear to be biased.

http://www.nanex.net/aqck/2977.html

HFT didn't cause the flash crash, but neither did that mutual fund manager.

The primary cause was a delay in when incoming orders were time-stamped by the NYSE. Instead of stamping the orders when they arrived at the queue just before entering the market, the NYSE servers time-stamped them when they _left_ the queue and were placed in the book.

Since the queue was delayed by extreme volume (NYSE has always lagged on technology), stale prices were posted to the NYSE feed. However, it was impossible to tell that they were stale from the timestamps.

Since the market was falling rapidly, this resulted in the NYSE quoting higher prices than every other market.

Since the NYSE was quoting higher prices than every other market, arbitrageurs massively sold at the NYSE and bought on other exchanges.

Since the queue was delayed, however, the sell orders at the NYSE took a while to actually show up in the book. Meanwhile, more sell orders were placed.

Want to see something scary? Here's the result: http://www.nanex.net/20100506/FlashCrashAnalysis_Part3-1.htm...

HFTs are playing games of milliseconds. It's not 12:20 and 12:40. It's 12:20:00.200 and 12:20:00.450.

Or it could be 12:20:00.200 and never. You are assuming Bob will show up, but at 12:20:00.200, neither Alice nor Eve know if he will or not.

If Bob shows up, Eve makes $0.05. If Bob never shows up and the price drops to $9.50, Eve loses money. Alice paid Eve $0.05 to take that risk because she felt it was worthwhile.

If Alice didn't feel this risk was worthwhile, she would have placed an order at $10.05.

But then, it wouldn't be called high-frequency trading, and you wouldn't have written about how "speed matters".

Please go reread the section on why speed matters. As I said, speed matters to Eve and Eddie (both HFT's) - because Eve placed her order at 12:20:00.000, and Eddie placed his at 12:20:00.030, Eve trades before Eddie. Alice's timing is irrelevant in this part of the game.

> Or it could be 12:20:00.200 and never. You are assuming Bob will show up, but at 12:20:00.200, neither Alice nor Eve know if he will or not.

Sure. It could be never. But it's not 50/50 or Eve wouldn't be playing. Eve buys from Alice because she expects to immediately sell at a higher price.

> Please go reread the section on why speed matters. As I said, speed matters to Eve and Eddie (both HFT's) - because Eve placed her order at 12:20:00.000, and Eddie placed his at 12:20:00.030, Eve trades before Eddie. Alice's timing is irrelevant in this part of the game.

It's not irrelevant. A price-increasing event occurs and Eve jumps on Alice's sell order before Bob's buy comes into the system. Bob pays the same, Alice gets her asking price, and Eve pockets the difference. This is just exploiting unequal market access.

But it's not 50/50 or Eve wouldn't be playing.

Of course it's not. Alice has a lower appetite for liquidity risk - that's why Alice chooses to pay Eve to take on this risk.

A price-increasing event occurs and Eve jumps on Alice's sell order...

First of all, there was probably no event. It's most likely that Eve had passive orders - buy at $10.00, sell at $10.05 out in the market (HFT's usually don't take liquidity, that costs too much money). Alice came along and chose to fill Eve's order.

Ignoring that, you also haven't explained how Eve's "unequal market access" plays any role in this. Why does Eve have to be a machine in this process? While being a machine helps Eve beat Eddie (another machine), if no machines were in the game then Eve could easily be a human. In fact, Eve was a human until fairly recently.

From what I can see, the "best price" rule basically becomes meaningless in the face of HFT. Alice is pretty much always going to get her min only, right? This prevents Alice from setting her min lower to manage risk. e.g. In a world where matches are executed immediately, but market makers are competing without an advantage, Alice could set her min at $9.50. If Bob comes along and buys at $10.05, great. If Chuck comes along at $9.55, not as great, but okay. But in a world where HFT will pop up and buy at the lowest possible price, setting a low min stops being a reasonable option, because you're basically capping your sales price at that point. So maybe HFT helps Alice get $10 instead of risking $9.55, but it also stops her from getting $10.05.

I know that market making could theoretically do this anyway, but it's a different situation when market makers have such a speed advantage. If you've got to sit on your position as long as the typical eTrade user, leaving a passive buy for $0.10 under market price picks up more risk, because you might not be able to cancel if the market shifts downward by $0.30.

You're using terms like "best price rule" but asking questions like "if Alice sets her 'min' at 9.50 she can sell to Bob at 10.05". This doesn't make sense. Alice has a limit order on the book that says she's prepared to sell at 10.00. When Bob comes along saying he'll buy at 10.05, the market fills the order at 10.00.

There are some things I'm not entirely clear about here. I'm not sure how the best execution rule (best price rule is apparently a bit different) plays out when there's a spread. When someone bids 10.10 and someone else asks 10.00, how should that be resolved. Either someone takes the whole spread or it's split between them, and I'm not sure what the SEC says should happen. A "minimum" price doesn't make any sense if it's the only price, but then neither does a "maximum" price.

In any case, though, the spread would theoretically go to the existing participants, rather that an HFT. The "market making" of the HFT still results in extracting money from the market. This could be a beneficial thing in illiquid markets, but I'm not sure it's beneficial in markets that already have high liquidity.

You're not clear on how order books work, which, respectfully, suggests that your reasoning on this stuff is a bit suspect. The standing limit order prices the trade.

I can understand how upsetting HFT must have sounded to you (although to be fair, we're still shifting the good outcome from Bob to Alice in your best case) given that misunderstanding, but, no: to capture the 5 cents (more likely: 1 cent) between Alice and Bob, the HFT had to accept Alice's downside risk exposure. There was no (simple) outcome where Alice could have it both ways, scalping Bob for 5 cents in the best case but getting out at 10 cents in the worst.

I came to my understanding of this topic in a weird way (see downthread) but one resource I found extremely helpful was Larry Harris' _Trading And Exchanges: Market Microstructure For Practitioners_. It is the TCP/IP Illustrated of markets. Very well written, and well written in a way easily appreciated by programmers. Highly recommended. When I first started reading it, I literally didn't want to put the book down.

Respectfully, I never claimed to be an expert of any sort, and I said so earlier. A large part of why I participate in these kinds of discussions is so I can learn.

I still get the feeling that you think I'm attacking HFT. It's not "upsetting" to me. The only questions for me are whether there's more value in HFT than cost, and whether the same value could be had with lower cost. It's good to know that my Alice scenario is invalid, though. That means HFT isn't breaking what I thought was a useful scenario for sellers.

Thanks for the book recommendation. I've added it to my list.

The last sentence of your second graf is what I was trying to capture with my "upsetting" sentence; sorry for the poor choice of words.

It's a great book.

Here's the thing though: ostensibly, financial markets aid in allocating capital for real world investment.

Real world investment doesn't work on a millisecond basis. It doesn't even work on a per-minute basis - at best, it works on a daily basis.

So if we were talking about how the financial markets benefit the real economy, we would say that Alice wants to sell at price X on Tuesday, and Bob wants to buy at price Y on Tuesday. And in that case, the go-between is unnecessary.

Nobody in the real economy needs instant order fulfillment when even regular bank transfers take a day to clear.

I get that more frequent dealings can help with finding "right" prices. It certainly reduces the spread. The problem is that HFT makes market access unequal.

You claim that Alice could have held out for $10.05 if she had wanted to. This is incorrect. Alice most likely does not have the infrastructure in place, and the fixed costs would be far too high for her to participate in that game. So in fact Eve did not provide any kind of useful service to Alice. Eve exploited unequal access to the market for her personal gain.

From a libertarian perspective you may say that's fine, granted. But the argument that Eve provided a service to the real economy is not as clear as you make it seem.

So there is a tradeoff there. HFT reduces spreads and perhaps helps price finding. However, for most market participants form the real economy, intra-day fluctuations dominate the spreads by orders of magnitude anyway, so reducing the spread isn't even that much of a useful service to the real economy - at least not to the extent it happens today. On the other hand, HFT makes access to the market unequal, limiting market competition.

It seems unlikely to me that the current situation is a proper balance within this tradeoff.

You claim that Alice could have held out for $10.05 if she had wanted to. This is incorrect.

Unless the "5" key on her keyboard is broken, she could have typed 10.05 into eTrade instead of 10.00.

At the point of IPO the capital for a company has already typically been allocated. The financial markets allow the owners of a company to sell their shares to the public and realize profits on their prior investments. After that markets are about betting whether the price of the company will go up or down and people trade accordingly.

The need to trade instantly is very important for risk mitigation. If unexpected news comes out today that a company is the target of an adverse event like a DOJ lawsuit many market participants will want to react as fast as possible to the news. Say it takes 30 seconds for the price to fall by $10. If you could trade at the 15 second mark you would only lose $5. To an investment firm (non Market Maker or other HFT) this can be crucially important and many firms have automated systems of their own to trade.

A key point of the article that most people miss is that someone has always fulfilled the role that HFTs currently occupy. In the old days floor traders occupied this role as pointed out by the article. Floor trading was a club where you had to buy a seat to be allowed in. Depending on the market the cost of the seat might be hundreds of thousands of dollars. The traders controlled access to the markets and for some markets like Oil the only way to trade was through these floor traders. Now these floor traders took full advantage of this and kept the spreads on the instruments they traded much wider than they are today. In the early 90s you could easily pay $.20 per share to the market maker to get a fill on a liquid stock.

The realization (as pointed out by the article) that computers could do this better and faster is just like any other industry. The people most hurt by this development were the floor traders themselves. There is a movie called The Pit that explores the effect that computerized trading had on floor traders in Chicago. Today if I want to trade MSFT the spread on BATS's BZX exchange was just $0.01. To me, another market participant I consider it to be much better to pay a penny to Getco rather than $.20 to a floor trader.

The colocation business as it has evolved has made market access more equal, not less. Today anyone can pay to colocate a server at NASDAQ, NYSE, BATS, etc... I know longer have to buy a floor seat to be able to trade there. Now there are real market barriers to entry but they are not structural. Just as you can't go build a Google competitor given that you don't have all the historical search traffic data and the infrastructure to compete with them, HFT firms have made significant investments in infrastructure of their own. If you had sufficient capital for technology development you could compete with any other HFT firm.

Yes there are people hurt by HFT, just like anything else when new players with greater efficiency emerge. However the average investor who is not competing with these firms is not harmed. The floor traders were, just like telephone operators before them.

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".

This type of risk mitigation is important to whom, exactly? When the type of unexpected news you mention comes in, somebody is going to take a hit, yes. Who should take that hit? What are the criteria according to which you make that decision in the first place?

In the scenario you outline, say the price falls by $10, somebody is going to take that loss no matter what. Without rapid trade, the original holder of the paper is probably going to take the full loss. In your scenario with rapid trade, they found some "fool" who was prepared to buy the paper at an intermediate price, and so the original owner lost less. So from the perspective of the original owner that is certainly a benefit. But is it a benefit for society? That is an entirely different question.

One could argue that the only reason they lost less is that they were faster in reacting to a certain piece of news, on a timescale of minutes, which is entirely irrelevant to the real economy.

Your point about HFT replacing floor trading is a valid point, but all it really proves is that barriers to entry into the market must be low. There is nothing inherent requiring high frequency trading. The job could just as well be done by lower frequency algorithmic trading, with a matching algorithm that runs on e.g. a 5 minute heartbeat.

Yes, the spreads are going to be bigger. Then again, intraday movements dominate spreads by orders of magnitude anyway, so from the perspective of the real economy, why should I care about how small the spreads are?

What you are really proposing here is to take away competition from the market. This would mean that you need another way of assigning the winner(buyer or seller, in case there are multiple ones wanting to buy the same stock); Some ways of doing it: randomly, alphabetically, shoe size, networth etc. You get the drift... The question is, will it be fair?

In capitalist systems winners are assigned based on open and fair competition, but then again there are other political views and systems.

Arms races are fair, but they waste resources. Splitting money randomly may be better for society than encouraging an arms race.

I think this is the important point and one that seems to be lost on a lot of traders: Yes, what you are talking about is logical, but we as a society have the right to say: No, this is not beneficial to everybody.

The way many traders argue, they seem to have little shame saying that they want the system to give them a hand when they have made a bad decision.

> If unexpected news comes out today that a company is the target of an adverse event like a DOJ lawsuit many market participants will want to react as fast as possible to the news.

No, I think the first thing you want to do is kick yourself in the butt for investing in a company that went down like this. If you invest in an oil company and it has a huge spill, the stock will go down and you will lose money. Don't like that? Don't invest in oil companies with a lousy safety record. Or invest yourself in making sure that the company that you gave your precious liquidity doesn't mess it up.

I would further claim that there are very few truly "unexpected" news. Traders are just a little too much in love with not really caring about what they invest in. At least not as much as they are in love with the money they make or the image that they are the "market makers" who provide the liquidity that we all need.

A DOJ lawsuit happens for a reason. Oil spills happen for a reason. If you have made a bet, you are the one who has to provide reasons for your bet. I cannot believe there are that many people who argue they should be able to rip off as many people as possible once their bet has gone sour. That the one thing they really need is to be able to rip others off as fast as possible. And that all this is somehow reasonable and useful.

I understand nobody really likes being the loser, but that's how capitalism works - filtering out the bad apples by having them go down in flames. Well, that's what it's supposed to be like, anyways.

> Arms races are fair, but they waste resources.

Amen.

Arms races are fair, but they waste resources.

Thank you for making my point in such an awesomely succinct way. This is exactly what I was trying to get at.

Capitalist competition is inherently wasteful - advertising, lawsuits etc. - but overall it allocates resources more efficiently than any alternative that's been tried.

> financial markets aid in allocating capital for real world investment.

this is only one purpose of financial markets. the other purpose is to transfer risk. any capital that you hold is subject to risks, and as a capital owner, you really only want to be subject to the risks that you know about and can compute well. you'd like to transfer other types of risk to other parties.

> It doesn't even work on a per-minute basis - at best, it works on a daily basis

is this a joke? if AAPL releases earnings at 10am and they come in 20 cents under expectations, i can guarantee you that regardless of the presence of computers, by 10:01am, AAPL is going to trade down a lot.

> It certainly reduces the spread. The problem is that HFT makes market access unequal.

lower spreads mean that less profitable strategies can survive in the marketplace. there will actually be more market participants in this type of scenario

> You claim that Alice could have held out for $10.05 if she had wanted to. This is incorrect

what you're describing here is a limit order, the most basic type of order that everyone has access to. she puts in 10.05 in etrade, and gets filled if the market moves up to 10.05 (but not if the market moves down towards 9.95 and does not come back up).

> for most market participants form the real economy, intra-day fluctuations dominate the spreads by orders of magnitude anyway

this is exactly why latency is important (and also contradicts your earlier statements). you want your orders filled now, before the market moves away from you.

Real world grocery shopping happens on perhaps a bi-weekly basis. But I want to be able to check out soon after I get to the register; I don't want to have to wait for hours or days for a cashier to decide that they're ready to execute the other side of my purchase.

Investment happens on a timescale of months or years. But when I make the decision to exchange one investment for another (I include "cash" as an investment) I want to be able to make that exchange rapidly. If someone wants to jump on my offer in mere milliseconds, I'll take it. Maybe I could've gotten a few more cents by waiting for an offer minutes or hours later, but that's not the game I'm playing. If I cared about those last few cents I'd price my asset a little higher and wait for a better deal. Since I don't think it's worth the wait, I'll let the HFT have those few cents for the convenience of letting me clear my trade immediately so I can move on with my day.

I see your point in theory, but in practice it doesn't seem to apply.

Even regular bank transfers, i.e. transferring money from one checking account to another checking account, usually take a day to clear over here. I am talking about regular bank accounts in Germany, by the way. The economy manages to run just fine with this system.

As long as something as basic and fundamental as simply moving money from point A to point B is that slow, I really don't buy the argument that the real economy needs transactions on financial markets to clear that quickly. What's the point of having a trade clear in milliseconds if it takes me a day to transfer the money to a different account?

Trades don't clear in milliseconds. Trades are agreed to in milliseconds, just as bank transfers are. I.e., once you push the "make transfer" button, the wheels are in motion, it just takes a day for the money to actually move.

The same applies to equities - once the trade is made, the wheels are in motion. But actually transferring the securities takes up to 3 days.

> "Trades are agreed to in milliseconds"

One might ask "why is this important if the transfer itself takes days?"

The main reason is that it gives you certainty and closure. You try to make a trade for + or - X shares of Y at $Z and you get an essentially instant response. You don't have to wait around all day just to know if you've got a deal; you don't have to poke around looking for a slightly better price; you just make the offer and see it accepted immediately, and can move on to other things. (This is the sense in which I meant the word "clear" a few posts up: the trade is cleared from your to-do list, even if the securities take some time to actually move.)

There is a cost for the service of being able to move on: the HFT will likely capture a tiny bit of profit which could have gone to you with more effort. You've traded money for time and effort.

Question of understanding: If what you say is true, and it actually does take up to three days to transfer the securities, how can you immediately sell an asset that you just bought a few minutes ago? My understanding of HFT (including from your description) was that this type of thing, i.e. holding an asset for only a few minutes, happens all the time. How can that be possible if the transfer takes as long as you claim it does?

You don't actually have to have the securities physically in hand in order to enter into a contract to transfer them.

But then yummyfajitas' comparison doesn't apply after all. When transferring money, the recipient has to wait a day until he can order the next transfer. This is not the case with trading, so this whole "three days" claim seems spurious to me.

I think his point was that the time it takes to be agreed to and the time it takes for the trade to clear are two different things, and that it's OK for those to happen on different timescales.

Having the trade agreed to quickly is a big deal even if it could take days to have any tangible assets to show for it.

You're fighting a losing battle. These people are personally involved in HFT or the financial industry in general, so naturally they'll try to justify what they're doing with whatever distraction-bullshit they happen to think of.

Sure, I can't know this, but it's a safe assumption. Whatever. Feel free to downvote and flag my post to bury it again, HN. Stay classy.

The army of PhDs from elite schools working for the financial industry are too smart not to realize that it's full of shit. How could they not see the relentless greed, and the sociopath douchebags in charge for what they are?

They'll be aware of bad/immoral/illegal things being done all around them, but hey, the salaries in finance are pretty fucking ridiculous and they get to work on challenging problems.

http://www.zerohedge.com/news/high-frequency-minutes-hft-exp...

Just like you said, in the real world, someone trading in the real economy does not make trades thousands of times per second. And I bet HFT bringing spreads down to $0.05 doesn't help anyone without a HFT machinery of their own.

But here they are, on Hacker News, bullshitting/distracting us convincingly time and time again.

One thing that hasn't been mentioned is the practice of placing a large block order and then canceling it milliseconds later to take advantage of the price movement this causes. So placing a large order does not automatically create liquidity--it can create the momentary illusion of it. (This observation has to be pompously dismissed in gratuitously insulting terms--professional rational discourse will not suffice to explain indefensible practice. I must have touched a nerve.)

Alice wants to sell at $10.00. Bob wants to buy at $10.05. They could simply trade with each other, but if Eve gets her way, Eve will buy from Alice at $10.00, and then sell to Bob for $10.05, making $0.05 on each share that would otherwise have gone to Alice.

If this is what HFTs were doing, it would be understandable why they were so upsetting. But of course, HFTs can't do this, because Alice and Bob's trade is executed instantaneously in the match engine.

Obviously HFTs can't wait until the match occurs before jumping in the middle, but they can and do jump in the middle of a match separated by milliseconds. Granted, they're assuming risk by doing this, but the risk is low, especially given that they can also cancel in milliseconds if the market shifts.

I can't understand the point you're making here. Earlier, you suggested ALICE SELL @ $10, BOB BUY @ $10.05. Those orders can't rest in the book; the match engine will evict them immediately by executing the trade. The is no time interval between Bob's order and the trade execution.

Could you perhaps explain the specific scenario you're talking about here? Alice, Bob, and Mallory perhaps?

I know that they won't sit on the book. The entire point of HFT is to jump between BUY and SELL orders as they arrive, though. e.g. This could be the sequence.

12:00:00.000 - Alice - SELL $10.00

12:00:00.100 - Eve - BUY $10.00

12:00:00.200 - Eve - SELL $10.01

12:00:00.300 - Bob - BUY $10.05

(Eve could have issued her BUY order before Alice's order arrived, SELL after Bob's BUY, etc.)

If Eve stayed home for the day, Alice and Bob would have happily traded with each other and Alice probably wouldn't have minded the extra 200ms delay but would have appreciated the extra $0.05 per share. But instead Eve decided to jump in and extract a bit of money from the transaction. And again, there's nothing wrong with that. But I fail to see how Eve is adding any value to the market here. There was no lack of liquidity.

    12:00:00.000 - Alice - SELL $10.00
    12:00:00.100 - Eve - BUY $10.00
    12:00:00.200 - Eve - SELL $10.01
    12:00:00.300 - Chuck - BUY $9.95
    12:00:00.400 - Chris - BUY $9.90
    12:00:00.500 - Charlise - BUY $9.80

Oops. Compare:

    12:00:00.000 - Alice - SELL $10.00
    12:00:00.300 - Chuck - BUY $9.95
    12:00:00.400 - Chris - BUY $9.90
    12:00:00.500 - Charlise - BUY $9.80

Eve took the day off. Alice is screwed.

Alice isn't screwed in the second case. She just gets lucky in the first case.

Alice is holding a security that is decreasing in value, and her book order isn't going to clear.

Assuming no HFTs, maybe Alice could set min_price=9.9$, and still sell for $10.05 in the better scenario? It seems that HFT's kill alice's sell spread option, forcing her to sell at min_price?

Huh? If Alice places a limit sell at $9.90, Bob's limit buy at $10.05 order will execute at $9.90. The limit price of the standing order is the one that prices the trade.

Oh, thanks for correcting my misunderstanding. So in general, sell "min_price" really just means sell "price"? Or are there circumstances that it would sell above that price?

As explained above, if Bob's $10.05 buy order was in the order book first, then Alice's $9.90 sell order would trade with Bob's buy order at $10.05.

It may be helpful to think of Alice trying to unload ten different stocks each at $10, every day M-F.

With HFT, some trades will go according to your first scenario; others will go according to the scenario you responded to. On net, Alice sells all her stocks to Eve and gathers in $100 every day.

If Eve stays home, instead Alice unloads perhaps six of her stocks to Bob at $10.05 each, and is still holding on to four of them. She doesn't want to end the day with $60.30 and four stocks she doesn't want, so she unloads the remaining stocks to Chuck and nets $100.10. Except that Chuck takes Mondays off, so on Monday she might end up selling to Chris and only finish with $99.90.

So what Eve (HFT) really accomplishes here is she grabs 10 cents of Alice's potential profit on most days or 10 cents of Alice's loss on Monday. Alice is happy with this arrangement because she gets the certainty of getting her $100 each day for the minimum effort. Eve is happy because she's grabbed an average profit of 6 cents per day (40 cents for T-F, -10 cents for M) in exchange for taking on a little bit of risk.

The real loser in this scenario is Freida, who is trying to do the same thing as Eve but is a little bit slower.

If Eve stays home, and Alice places a standing limit sell order at $10.00, and Bob comes along with a limit buy at $10.05, Alice's is the standing order and will set the price of the trade. The trade will fill at $10.00, not $10.05.

This comment demonstrated the fundamental point of your misunderstanding. Your ordering of events is not how it works. Eve doesn't "jump in the middle" What actually happens is this:

  12:00:00.000 - Eve - BUY $10.00
  12:00:00.000 - Eve - SELL $10.01
  12:00:00.200 - Alice - SELL $10.00
  12:00:00.300 - Bob - BUY $10.01

Eve was there before Alice & Bob. This is KEY to understanding what is going on here. Even is NOT jumping in the middle. Eve provided a service to both Alice & Bob and got paid for it.

Here's another way to think about it:

Alice wants the ability to sell her stock whenever she wants for a "correct" price. Right? That doesn't come for free. Someone has to figure out what the correct price is. That takes effort and costs money. Humans used to do this (they were called market makers). Now computers do it because they're better/faster/cheaper at it (just like they're better at a lot of things humans used to do).

They're also cheaper at it. The cost to Alice of getting to sell whenever she wants at a good/correct price has actually gone down from what it used to be. You can see this in the shrinking spreads. It used to be we only had price accuracy to a dime or a quarter. Now it's generally down to only a penny! Even though Eve is making money Alice is getting a way better deal than she used to before someone figured out how to program Eve to do it instead of the slow and expensive Harry the Human.

It's like in Superman III.

It should be obvious how it is beneficial. HFTs are in competition with themselves, not with investors like you and I. Prior to HFT market making, manual market makers (brokers on the floor) were making huge profits on large spreads. This means worse prices for the average investor.

Without HFT there would be no "1 penny" spread.

> why we should build a system that enables HFT to be profitable.

because HFT's are more efficient. paying a guy to make a market in one stock is expensive. paying for a computer to do it is cheap. when the market maker's costs are lower, they can afford to quote tighter spreads and still be profitable. lower spreads => lower tcosts for "normal" (low frequency) market participants

If you cancel faster than the market, you can stand first in line for everything and leave the line for those occasions where the trade isn't profitable. I don't see the difference between jumping to the head of the line for Good and standing in every line but jumping out for Bad. And I don't see how line-jumping adds liquidity. A bid I can never hit because it will be gone any time I'd actually want to hit it isn't much of a bid, at least as far as I'm concerned. If it's gone before anyone can hit it, then I think there's a definitional question of whether that's really a 'bid' for the intents and purposes of liquidity provision.

Now I'm not saying such bids are Bad. They could be helpful in all sorts of ways. But I'm not sure that provision of liquidity is one of those.

>"A bid I can never hit because it will be gone any time I'd actually want to hit it..."

Why should they let you hit them when they expect to lose? The HFT firms have some pretty decent traders. They're not going to sit there idle and let you pick them off. If you want to win this game, you have to be better than the competition.

They shouldn't, and I don't expect them too. The point is that someone who is effectively jumping to the head of the line for profitable trades isn't really "providing liquidity".

When I think of liquidity providers, I'm thinking of market makers who post prices at which they either buy or sell, offered long enough to allow other traders to hit either.

I'll believe that Amazon's network engineers (and whoever else is involved in making their page fast) add value when it takes 200ms to ship products to my home.

Reply on news.ycombinator.com