I'm no stock expert, but I still don't see how people are getting "screwed" by hft. If you are concerned about a price fluxation just place a limit order. Simple as that. If you don't care, then place a market order. There's always been someone out there with more information, smarter people, and now faster connections than you. It's never been a level playing field and will never be.
The cynical answer to this is good marketing and savvy backers with a vested interest in getting retail customers to trade at a disadvantage.
The less cynical answer is that trading has always been a back and forth between sophisticated actors, some of which make money providing liquidity and some of which make money others ways but buying liquidity is a cost. The cheaper the latter can get from the former the better it is for them.
IEX is a dark pool setup directly for the benefit of savvy hedge fund types to move large blocks of trades and pass on those costs to others. Nothing wrong with it, but their moralizing in books is a little off putting.
Well, you can disable front running with your own exchange which is what he is doing. And it seems to be working (he is getting liquidity, their average market share has tripled from .4% to 1.1%)
This is much better than the government stepping in, so I am very glad he is doing it. We'll see if he can keep up the growth.
So is the police killing someone on a "nickel ride" like what recently happened in Baltimore. But if there's no enforcement of the laws, or if the regulators disagree with the sentiment of the rest of the population, the notional illegality of such activities is of little practical value.
Before you try and suggest that the financial authorities are 100% competent, I would remind you that Bernie Madoff did in fact dupe many people for many years. Financial authorities included. In fact it was a non-governmental person who blew the whistle and starting the whole thing tumbling down.
So I'm not saying that it's rampant and everyone is looking the other way. But it's definitely a possibility. What people are SUPPOSED to do and what they ACTUALLY do often don't overlap anywhere near enough.
Do you think it should be illegal to look at the order book for a stock (public info), formulate a prediction, and trade on that prediction? Seems much different than a broker trading ahead of his client.
Large informed traders and intermediaries like specialists/locals/market makers have been in a cat & mouse game since the dawn of time. Back in the day, big hedge funds would split their orders through multiple brokers or try to make their intentions. It's natural that one seeks to hide his intentions and buy/sell without moving the price the other wants to get the hell out of the way or trade with the big order. That's just competition.
I guess in modern terms "front running" really just means getting arbed. If someone fast sees prices move in one market, they can buy at a relatively unfair price elsewhere, even sometimes outrunning the person who moved prices in the first place. That's not really a bad thing. It makes prices move together and spreads out liquidity across markets. It's also not risk-free money. Even a pure arb has legging risk.
A lot of dark pools have orders "pegged" to the mid-market of the best bid and offer. For example if a stock is $100.01 bid offered at $100.02, the order would float at $100.015 and adjust as the price moves. If the market moves rapidly, and a fast trader sees it go $100.02 bid offered at $100.03, sometimes they can trade with the pegged order at $100.015 before the system knows the price even moved. Make a half-penny, rinse, repeat a million times. At least that's the story.
IEX prevents some of this by computing their reference price in real-time that's used to determine where orders can execute and where pegged orders are priced, while delaying incoming orders by 350 microseconds so nobody can outrun their calculation of the best bid and offer to pick off stale orders. Mind you, this isn't even a real issue in markets today. There are a million ways to trade orders that are statistically mispriced before the market actually moves, so everyone will be taking them with these predictions with little left for the guys doing a pure arb. If you blindly peg your order to any benchmark, you are going to lose.
Also, there was already a market solution for this. Off-exchange markets have discretion on who they allow in and who can interact with one another. If you do nothing but arb slow orders in their pool, they'll either kick you out or restrict who you can trade with.
Too close to secret sauce territory, but if you can predict the market is more likely to tick up than tick down or vice versa, then buying or selling at mid-market is +EV. Here are some publicly-available examples that are too well-known to work anymore, basically various features of the order book: http://www.orie.cornell.edu/engineering2/customcf/iws_events...
The fundamental idea of algorithmic trading doesn't really fall into the category of screwing-people-over. However, some techniques employed by institutions engaged in it can easily be seen as outright cheating. One common example is front running [1] where a player with access to real-time order information exploits the knowledge of pending trades from other players to turn a profit at the other players' expense. These techniques can also be used across exchanges.
The issue is a consequence of the lack of market transparency. The exchanges are only required to make order information public with a significant delay. However, they sell the real-time order information to a few market players enabling them to engage in the dubious tactics at the expense of other players.
Note that tactics like front running used to be employed by brokers before algorithmic trading and have since been banned.
Front running is already illegal, so if any group is doing that they are already breaking the rules of the game, and should be punished/stopped if found out.
What some HFT traders are doing is economically equivalent to front-running, but is not legally front running since they are not front running the orders of their own clients. All of this is explained in the article, so if you disagree, it would make more sense to read it, find the parts you disagree with, and argue against them, rather than skim the article and then argue with our collective rehashing of the article's argument.
That's a huge difference though. A broker owes a duty to his client not to screw him over. He's supposed to be helping the client. A prop trader has no duty to others in the market. He's their adversary. They're both competing to buy underpriced shares and sell overpriced ones.
For the purposes of this discussion, it's not a difference at all. The question is not what should be legal or not, but what customers would prefer, and what makes a better market.
If you ask a customer "would you like an exchange that gives people enough information to front run your order" or "would you like a broker who front runs your order" the answer in both cases would be a resounding no. The claim in the article is that you can create a better exchange by providing an equal footing for all participants.
Everybody wants to trade with no price impact. If you asked me if anyone else should be allowed to colocate their servers, I'd say no, just give me a monopoly please. One participant's desires don't make a good market.
Fair and non-discriminatory access to exchanges (not free, but anybody can pay to compete if they wish), fast trading engines and deep order book information are what let traders put up such tight spreads for customers in the first place. Vigorous competition between all market participants keeps prices fair.
EDIT: I think you are right, although I would express things a bit differently. I'm writing a top level post with my thoughts.
ORIGINAL POST:
When you make a trade, there are some information rents associated with that trade. If someone knew you were about to trade, they could capture all of those rents. When there is a monopoly on exploiting these information rents (as in frontrunning), they will be captured by the frontrunner. When there is perfect competition (as when a number of HFTs can trade in another exchange before your trade reaches it), these rents will be all wasted on technology to maximize speed, i.e. no one gets them, they become deadweight loss.
I understand that there are other issues, such as knowing whether a trade comes from a single seller or multiple smaller sellers. But this is a separate issue from simply trading ahead of other people. I don't understand why it's so controversial that an exchange would advertise the fact that traders can rely on their orders reaching the exchange before anyone else knows about it as a good thing.
The original trader still has the ability to capture his information rents in a single trade on one market. Nobody can see it and step away or react before the execution goes through. That's all anyone is guaranteed. HFTs face the same problem. If they're trying to cross the spread to put on a trade or hedge, they also have market impact costs and slippage.
I think it's controversial because a lot of their marketing implies all HFTs do this as their primary business and thus add very little value. It creates this narrative of a parasitic activity that intermediates trades which would have already occurred where no risk is taken. I don't think that's the case in reality. All HFT strategies are looking at order books in the products they trade, and they definitely make money (or avoid losses) when someone executes inefficiently across exchanges, but there's more to it than that. HFTs are very successful in futures and non-US markets where products trade on a single exchange. How do they do that if they aren't good at predicting prices and taking risk?
>The original trader still has the ability to capture his information rents in a single trade on one market... That's all anyone is guaranteed
Yes this is the essence of it. Based on other posts, I would guess that if you really wanted to make simultaneous trades in multiple markets, then you could. But even if HFT really did render this impossible, it's not a bad thing, since it trading in two exchanges simultaneously actually gives more rents to the large trader than the theoretical ideal of a single continuous double auction. And if you want to trade in A then trade in B based on the results of A, then you certainly don't deserve to be the first to make it to B.
The problem is that front running requires 2 things:
- the ability to see orders ahead of someone, which this article implies you can, but you can't.
- the fiduciary duty not to use someone's intent to trade against them.
HFT traders do not fall into either camp. They can't see your order before it goes to an exchange and they don't have a duty not to change their prices on one exchange in reaction to price moves on other exchanges.
The article claims that by colocating in one exchange, HFT can see you order arrive at exchange A, and send orders to exchange B before your order reaches exchange B. I don't know if this is true, but this is the most important part of the article. Given the fixed latency they introduce, you can send your trades to IEX before sending to any other exchange, safe in the knowledge that no other exchange will react to that information before your order.
No one is talking about the fiduciary duty. Fiduciary duty is just a way to define a default contract into law. E.g. brokers could write a contract saying that they promise not to front run. But since this is something that almost every client would want, it's just written into law instead.
There is a major caveat to what they are reporting that gets to the heart of this.
HFT can't see your order as it arrives on exchange A before it arrives on exchange B, they can see the results of an order. That is, they can see a large trade happened on price X on A. With that information they may want to change their prices on exchange B and they may be able to do that before you can also place an order on B.
There are lots of ways to mitigate this other than using a dark pool like IEX. All the exchanges offer exotic order types to make this easier for instance, and large block traders have for years had execution algorithms that spread their orders around venues and over time to minimize price impact.
That IEX offers another option for large block traders does not bother me at all, that they market it via underhanded moralizing i find distasteful.
Finally, in this context front-running has a specific meaning and it requires a fiduciary duty for it to be front running. By using that term incorrectly you ARE bringing fiduciary duty into the conversation, whether you want to or not.
Frontrunning is a crime because it's acting on privileged information. It's in the same boat as insider dealing. They have that privileged info because of their fiduciary relationship so it is relevant.
What the HFT guys are doing is not frontrunning because anyone can get hold of the same information they have (for a price). That information is not privileged (assuming deep pockets != privilege) so it's not frontrunning.
They're getting to the trade first because they're technically quicker than the routers the banks are using.
>The exchanges are only required to make order information public with a significant delay. However, they sell the real-time order information to a few market players enabling them to engage in the dubious tactics at the expense of other players.
It's easy to see why you view this as unfair.
But it is fair. Anyone can buy the real-time information if they have the capital.
Obviously markets are going to be more friendly towards you the more money you throw at them.
Just like everything else in capitalism, the fairness comes from anyone being able to have money. If you think this is unfair, you're essentially opposing capitalism.
The idea that HFT is front running any more than a human market maker is a fallacy. What HFT does, is to properly price the liquidity in the market. Previously, large institutional or more likely hedge fund orders would be priced very tightly, because it was in the market maker's interest to know the direction of the flow. By "winning" the trade he/she would have an information advantage on the subsequent likely move in the market (a good assumption here is that flow directions are auto-correlated, with a higher autocorrelation to important customers, which is why this makes economic sense). Armed with this information, said market maker would then pre-position against some of the slower "less in the know" flows who are inevitably triggered by price action, with such price action of course influenced by the market maker himself who can "dribble" out the big position at his leisure. He is "armed" with a position that he can move the market with. Such pre-positioning was profitable as smaller investors reacted to price action and sold. It was basically a game of "I pay for your directional information through tight bid offer for the privileged few".
HFT says the following. If you start selling, then I will move the price downwards in accordance to my best guess as to the size of your flow. If I see serial sell orders, I will move the price subsequently lower on the evidence. I am pricing your liquidity correctly. Thus, if you are a huge hedge fund shorting the market, you face higher transaction costs because the liquidity you are demanding of the market is high, and the market maker can no longer offset your liquidity cost against ripping off smaller investors who will be stopped. This latter is the key point. In fact, for smaller investors, HFT means bid offers are much tighter. If you put a small order through the market you are instaneously filled almost at mid, thanks to HFT, whereas in the old days, retail would pay the same bid/offer for his tiny flow as huge hedge fund. The converse is true for buy orders, obviously.
Have you wondered why HFT is most hated by large institutional funds? It's because suddenly they have to pay a fair price for their huge liquidity requirements.
HFT has short-circuited the old game where large influential funds obtained tight bid offer at the expense of smaller players, and were thus able to dominate markets by trading size at unrealistically tight prices. Now we all pay for our liquidity at a fair price. It's actually really democratic (keeping in mind that competition is mega-fierce).
Ultimately, those who accuse HFT of front running are 100% correct. But there is always "statistical" front running (autocorrelation remember), HFT or Human. The difference now with HFT is that the cost of this front running is proportional to the size of your flow, and thus for smaller/retail investors it is neglibeable (because pricing is now much tigher than under the old "human" regime), and therefore positive, but for the spoiled insitutional investors / hedge funds, used to being privileged, it is back to a level playing field. Need to sell a yard? Need to pay for that buddy. Not good for them.
(obviously I am schematizing to a certain degree. Even humans price large orders wider. It's just they didn't price those large orders anything like wide enough. They sponsored the big client for the info advantage thus earned, info advantage wielded against the rest of the market: read smaller, and/or less in the know, investors).
I don't follow what you say here - my understanding has it that spreads have grown tighter since computerizing things (and generally that a tighter spread is more efficient).
That's my point. For the vast majority of (smaller) flows HFT makes bid/offer tighter. TLDR on the above: HFT makes you pay a fair price for your liquidity requirement. If you're retail, your liquidity requirement is negligeable. If you're big your liquidity price is high. It's fairer.
In the old human system big funds' artificially high (cheap) liquidity is sponsored by the small player's artificially expensive liquidity and/or market move against him/her before he/she trades.
It's more complicated than you suggest: If informed trading can't capture a good portion of the difference between the "true" price and the current price then there isn't an incentive to do it.
The spread is tight for small orders, but if you keep hitting them in one direction, market makers quickly get wise to this and start pulling back. The cost to trade for large funds isn't just the bid-offer spread, but also how quickly the market reacts to their flow. Modern market makers are very good at pricing a small spread to uninformed flow while making informed traders pay.
This is also why big funds want rules like trade-at (all orders must trade on-exchange) which eliminate the ability for market makers to explicitly segregate order flow. If all retail orders went to exchanges, market makers would give tighter prices on average to win the retail flows. Basically retail traders would subsidize institutional ones.
Modern market makers are very good at pricing a small spread to uninformed flow while making informed traders pay.
My understanding was that it's mostly the opposite, that MMs profit from uninformed order flow (so much so that they'll buy access to it from brokerages), and that informed traders were their predators in the market food chain.
At any rate: cost of trading has by every account I can find plummeted since the introduction of automatic trading. As recently as the 1990s, spreads were quoted in eighths of dollars, and there was widespread collusion to fix prices at quarters. In 70s and before, the markets were worse; basically organized fleecing operations.
Yes I mean if I think the fair price of a stock is $100, and I knew who I was trading with, I might bid 99.99 offer 100.01 to a retail trader, but bid 99.95 offer 100.05 to a bank or hedge fund. The retail trader probably doesn't have short-term info and he's not going to keep trading in the same direction so I don't need to make as much spread.
On exchange markets you can't really tell who you trade with so you have to guess.
In order front-run someone, you must have a fiduciary duty to them. HFTs do not.
When you think about this broader proposed definition of front-running, where any anticipatory trading is "front-running", you quickly realize that virtually any informed trade could be described that way. Do extensive research on the supply chain of a billion dollar industry, notice a spike somewhere, and trade on that? You're front-running!
Exchanges make it difficult to know the current price. They sell fast data feeds to HFT shops allowing them to know about price changes before the rest of the market participants.
Exchanges have tons of order types, not just limit and market orders. There are order types that specifically benefit HFT shops by allowing them to pick off limit and market orders.
If you put in a limit BUY order for $10 and the current price is $9.90. There is a big difference between executing at $10 vs $9.90 on large orders. If a HFT shop is able to push the price up to $10 (there are a few ways to do this), you just lost a lot of money and probably made the HFT a nice paycheck when the price reverts back to $9.90.
There are also order types the specifically advantage large block traders at the expense of HFT market makers. IEX only offers those interestingly, while speaking out about the others.
There is also a big difference between pricing something at 9.90 when you think there is very little demand for it, and being prevented from changing the price when the demand changes.
if you have continuous trading, you are going to have complicated interactions (such as outlined in the article.)
instead, an occasional crossing (once a day? once an hour?) would provide much more "fairness" since everything happens at once, at the expense of "timeliness".
back when I worked in this industry (a decade ago) POSIT provided something a lot like this
Their system is not fundamentally different from discrete trading, since they have delay lines going into and out of their servers. So while everything is continuous, if you want to observe the market, then do a trade, then observe the result of that trade, etc. there is maximum rate you can do this.
This is what prevents front running[0], according to my understanding. Once you submit an order, no one can react to that information within a time greater than round trip latency to the servers.
The problem with crossing at discrete intervals is that it introduces new kinds of strategic considerations that make it much more complex for people to play the "game".
EDIT
[0] For the sake of the pedantic, I mean front running in the sense of the article at the top of this page, not the legal sense. The article defines a notion of front running that maybe not everyone agrees with, but that is what I am referring to.
This is what prevents front running, according to my understanding. Once you submit an order, no one can react to that information within a time greater than round trip latency to the servers.
That's not correct. Front-running is the crime by which your broker trades ahead of you after you state your intent to trade. Your broker does, in fact, see your trades before they reach the market (the broker is actually the one pushing them to the market). It's mainly a crime because your broker has a fiduciary duty to act in your best interest, and because the broker occupies a privileged position as an intermediary, not because there is something inherently wrong with demand anticipation.
Of course, it is correct that no one besides your broker has the physical ability to react to that information.
But if someone, somehow, developed a legal technical means to eavesdrop on the order you sent to your broker, and traded based on that, wouldn't it be fair to say that this is, in some sense, similar to front running? That is what the article is claiming, although the information is the information available at one exchange before it reaches another.
And going on with this analogy, wouldn't it be the case that we would want to prevent this eavesdropping if possible, because it has a similar effect to actual frontrunning, namely that the intermediary captures the information rents associated with your knowledge of the order you are about to submit.
The purpose of laws against frontrunning is to enable you to have a trusted relationship with your broker. There is no benefit to creating a trusted relationship with your adversaries.
Similarly, doctor/patient confidentiality laws are there to created a trusted relationship and make you comfortable admitting to your doctor that you had unprotected sex with lots of heroin addicted hookers in sub saharan Africa and you now need all the STD tests. In contrast, passing laws against journalists reporting on who visits hookers has no such benefit.
I really don't care about the law. The point is that if front running was legal, brokers would try to find ways to guarantee to their clients that they wouldn't do it. Similarly, the guy in the article advertises his exchange as preventing the equivalent of front running at the exchange level.
Their ATS can only exist by free-riding on the public price discovery in displayed US markets under Reg NMS. They protect their customers from trading at "wrong" prices by letting traders on faster exchanges determine what the "right" price is for them.
As a thought experiment, imagine IEX captured 95% of market share as a displayed exchange. The 350 microsecond delay wouldn't matter since it's applied to everyone and there would be no reference markets for them to price against. It'd just be another market with a slightly slow matching engine. Relative speed still matters so fast traders would have an advantage.
There's two sides to every trade though. Somebody wins, somebody loses. The orders on the other exchanges either belong to market makers or other uninformed traders.
If you structured the market so market makers couldn't see order flow as quickly, they'd charge a larger spread to everybody. Right now they charge a tight spread so long as the incoming order is small because they know at least some of the time they can get out of the way or try to dump their shares when a big order comes in. That's good for the small guy. They wouldn't even be there for the big guy to hit if you changed the rules. Instead the sales trader would only see half the shares at the best price and end up paying multiple ticks just like he does now.
If the orders on other exchanges are uninformed slow traders, they're going to lose either way, so we're really arguing over whether the savvy hedge fund or the fast aggressive HFT should get their money. If you're playing the game of making fractions of a tick, you're in the same execution game HFTs are in, so you better get fast.
The problem with their system is that they are basically always lagged, but the instruments are also traded on other exchanges that aren't. Why would anyone go there?
'Safe' markets are valuable even for insiders, because they attract more investors and volume. People naturally want to invest where they won't be cheated.
That has been forgotten in the (largely manufactured, IMHO) anti-regulation outrage. Even the big Wall Street firms should benefit from regulation that makes non-insider investors feel the market is safe enough to invest in.
That said, I'm confused and my theory fails: Despite the long run of incompetence and fraud on Wall Street, its reputation as the leading place to invest and for expertise seems to persist and it has resisted regulation to a great degree. You don't hear people say, 'don't invest in the stock market because you'll be cheated'; or 'don't hire (some major Wall Street firm) because look how they cheated these other people, and they demonstrated complete incomptence in events X and Y'. Their reputation seems immune.
It's pretty silly to describe IEX as curbing "abusive" practices.
What IEX is attempting to do is build a platform where large traders can move lots of shares while their smaller counterparties are stuck absorbing the price impact. This is potentially useful for large traders (e.g. Goldman, JP Morgan and Citi, as mentioned in the article) but bad for small traders.
What's actually kind of "abusive" is IEX's marketing - they are encouraging unsophisticated investors to direct liquidity to them rather than having brokers route for best execution.
This letter directs your broker to route your trades to IEX, rather than the best available venue. This means you may be stuck paying IEX fees - which could be greater than other venue's fees - and of course, you are providing liquidity to sharks who want to make sure you absorb the price impact of their trades.
I think what you're saying is, in essence, IEX's current incarnation of handling stock transactions does not exist in a vacuum -- because they are only a trading technology platform instead of an exchange like NYSE & NASDAQ. As a technology platform, they are really just "clients" of other exchanges and therefore, they're the tail trying to wag the dog.
If they become a full-fledged exchange with stocks listings exclusive to IEX, then those stocks would have better "fairness" characteristics that Brad Katsuyama claims.
Therefore trading AAPL through IEX may incur a financial penalty for being on the "slow" IEX system but trading YC2019Unicorn stock exclusive to IEX may not.
Actually, what I'm saying is the following. Don't put resting orders out onto IEX - a big player will take your liquidity in a big trade and you'll suffer the inevitable price impact. I.e., putting your orders on IEX rather than the open market is just a way to funnel your money to JP Morgan.
It's far better to mingle your orders with HFT orders - that'll keep the big guys honest. You'll probably also pay lower fees.
In 2006, if I saw 100,000 shares of AMD offered, and I wanted it, I could go out and buy it. It was a simple as that. In 2007, if I tried to buy 100,000 shares, I would get 80,000. Then in 2008 I would get 60,000. The market is showing me a volume at a price that I can no longer buy or sell at. I can’t buy or sell what I see on my screen.
Another way to put this:
"Once upon a time, I was paid a fortune by a giant investment bank to move large blocks of stock on behalf of their spectacularly wealthy clients. As recently as 2006, if one of those clients needed me to move a block of 100,000 shares, I could do that at literally the click of a button. This despite the fact that I earned a commission on the trade; it must have been a very expensive button my bank owned! And this despite the fact that my 100,000 share order was bound to move the market, and so I was in effect acting on inside information. But that's just how this is supposed to work, right?
"Anyways, the markets evolved, and my giant investment bank could no longer earn massive commissions just by pushing a single button. Even though I knew my giant hedge fund clients were going to dump vast numbers of shares on the market, depressing prices for all the other investors, the markets no longer allowed me to trivially profit from that information! My 100,000 share orders get broken into small numbers of lots just like everyone else's. No fair!"
"So I started a new exchange to rewind the markets back to the glorious, equitable, fair, transparent days of 2006."
whose money do you think is in those hedge funds? it's all a question of what game you want to play. the players are always the same. it's capitalism after all.
Whose money do I think is in those hedge funds? I don't know, a whole bunch of millionaires?
If the direction you're heading is "modern trading is scalping profits from pension funds and mom-and-pop mutual fund retirement plans", you'd probably want to be prepared to refute Vanguard's chief investment officer, who says HFT has in general improved outcomes for one of the world's largest and most trustworthy fund managers.
They may scalp profits from hedge funds that trade actively, but guess what, they're playing the same game: trying to trade at attractive prices. With two pros competing with one another, why does one side deserve sympathy? I don't see Burger King whining to the press when McDonalds outearns them.
For the average passive indexer who buys some ETFs or mutual funds every quarter it's meaningless at worst, and probably a net benefit since these products have tighter prices now.
Cute. Except the ones suffering the most from the HFT machinations were most likely pension and mutual funds, the ones most folks parents rely on. Not rich fucks or their catered hedge funds.
Vanguard says exactly the opposite is true. Which makes sense: passively managed funds and funds with simple value-based investment parameters don't trade aggressively. On the infrequent occasions when they do trade, they benefit from the reduced cost of trading.
If you have your retirement fund parked in an actively-managed fund where you pay an annual fee to have someone who makes $500,000/yr base continuously push "buy" and "sell" buttons on their computer screen all day, you have bigger problems than HFT.
> They could see my order at BATS, race me to the next exchange, and cancel all their sell orders and buy whatever is left, buy everything up, then turn around and try and sell stock back to me at a higher price. So that was the game.
This is the heart of the problem with IEX's explanation of the markets, and why I think most people view HFT as unfair. But its not how cross exchange market making actually works. The HFTs are not rushing to buy up existing inventory from others who are selling it, in an attempt to screw others. They are the ones who are originally offering to sell it in the first place. They aren't rushing to buy, they are rushing to change the price of their inventory.
The metaphor I like to use is that of a chain of gas stations running down a highway, owned by the same people. If a tanker truck pulls up at the first two and buys up all their gasoline, it would be perfectly reasonable for those first 2 to call ahead to the rest down the highway and tell them to raise their prices, as they clearly were being used to supply someone else's industry at lower prices than they should.
That phone call is what IEX prevents, so that their customers (guys with big tanker trucks) can buy gas without impacting the price of gasoline.
Whatever the realities of the market, people have an expectation of "fairness", and are generally against the idea of a party or parties being able to insert themselves into a situation where they can essentially get a speed jump over the rest of the market by virtue of buying a place that enables this (microwave links, colo in the exchange etc.)
Why would a HFT maintain a huge inventory of stocks? If you make money on the ability to execute trades faster than others on the movement of stocks, you're losing money by holding them. It feels akin to a car maker holding an overlarge inventory of parts - it's just tied up capital that's not making you money. In HFT, don't you maximize the profit by holding as few stocks for a little is possible?
The short answer is that they have to. Resting orders from yesterday are always faster than microsecond orders from right now.
So a market maker that was super fast but didn't rest orders, would be competing with one that is as fast as them and willing to rest orders. So, the resting order ones win and drive out the non-resting order ones.
Whether or not HFTs are rushing to buy the stocks or they have the stocks already, doesn't change the fact that proximity to the data center allows HFTs an unfair advantage. It's unfair either way.
Someone always has and always will have closer proximity to an exchange. You'll notice IEX is hosted in a large exchange centric data center and allows colocation as well.
This proximity is actually cheaper, more standardized and easier to access now than it's ever been.
Read "Flash Boys" for a better insight into how much difference variable latency between exchanges and clients makes to the profitability of fast execution of trades.
Patterson's _Dark Pools_ which tells the story of Island and the ECNs and the advent of automated trading. Patterson is more ambivalent about HFT than Kovacs, and does a good job of explaining the Core Wars phenomenon of modern trading from '98 to the mid-'00s.
Dan from IEX here. For folks genuinely interested in US equity market trading dynamics, this paper out of Columbia is an awesome primer. More objective than flash boys or what you'll read in the news.
80 comments
[ 3.1 ms ] story [ 143 ms ] threadThe less cynical answer is that trading has always been a back and forth between sophisticated actors, some of which make money providing liquidity and some of which make money others ways but buying liquidity is a cost. The cheaper the latter can get from the former the better it is for them.
IEX is a dark pool setup directly for the benefit of savvy hedge fund types to move large blocks of trades and pass on those costs to others. Nothing wrong with it, but their moralizing in books is a little off putting.
This is much better than the government stepping in, so I am very glad he is doing it. We'll see if he can keep up the growth.
So I'm not saying that it's rampant and everyone is looking the other way. But it's definitely a possibility. What people are SUPPOSED to do and what they ACTUALLY do often don't overlap anywhere near enough.
Large informed traders and intermediaries like specialists/locals/market makers have been in a cat & mouse game since the dawn of time. Back in the day, big hedge funds would split their orders through multiple brokers or try to make their intentions. It's natural that one seeks to hide his intentions and buy/sell without moving the price the other wants to get the hell out of the way or trade with the big order. That's just competition.
A lot of dark pools have orders "pegged" to the mid-market of the best bid and offer. For example if a stock is $100.01 bid offered at $100.02, the order would float at $100.015 and adjust as the price moves. If the market moves rapidly, and a fast trader sees it go $100.02 bid offered at $100.03, sometimes they can trade with the pegged order at $100.015 before the system knows the price even moved. Make a half-penny, rinse, repeat a million times. At least that's the story.
IEX prevents some of this by computing their reference price in real-time that's used to determine where orders can execute and where pegged orders are priced, while delaying incoming orders by 350 microseconds so nobody can outrun their calculation of the best bid and offer to pick off stale orders. Mind you, this isn't even a real issue in markets today. There are a million ways to trade orders that are statistically mispriced before the market actually moves, so everyone will be taking them with these predictions with little left for the guys doing a pure arb. If you blindly peg your order to any benchmark, you are going to lose.
Also, there was already a market solution for this. Off-exchange markets have discretion on who they allow in and who can interact with one another. If you do nothing but arb slow orders in their pool, they'll either kick you out or restrict who you can trade with.
Can you talk more about this?
The issue is a consequence of the lack of market transparency. The exchanges are only required to make order information public with a significant delay. However, they sell the real-time order information to a few market players enabling them to engage in the dubious tactics at the expense of other players.
Note that tactics like front running used to be employed by brokers before algorithmic trading and have since been banned.
[1] https://en.wikipedia.org/wiki/Front_running
If you ask a customer "would you like an exchange that gives people enough information to front run your order" or "would you like a broker who front runs your order" the answer in both cases would be a resounding no. The claim in the article is that you can create a better exchange by providing an equal footing for all participants.
Fair and non-discriminatory access to exchanges (not free, but anybody can pay to compete if they wish), fast trading engines and deep order book information are what let traders put up such tight spreads for customers in the first place. Vigorous competition between all market participants keeps prices fair.
ORIGINAL POST: When you make a trade, there are some information rents associated with that trade. If someone knew you were about to trade, they could capture all of those rents. When there is a monopoly on exploiting these information rents (as in frontrunning), they will be captured by the frontrunner. When there is perfect competition (as when a number of HFTs can trade in another exchange before your trade reaches it), these rents will be all wasted on technology to maximize speed, i.e. no one gets them, they become deadweight loss.
I understand that there are other issues, such as knowing whether a trade comes from a single seller or multiple smaller sellers. But this is a separate issue from simply trading ahead of other people. I don't understand why it's so controversial that an exchange would advertise the fact that traders can rely on their orders reaching the exchange before anyone else knows about it as a good thing.
I think it's controversial because a lot of their marketing implies all HFTs do this as their primary business and thus add very little value. It creates this narrative of a parasitic activity that intermediates trades which would have already occurred where no risk is taken. I don't think that's the case in reality. All HFT strategies are looking at order books in the products they trade, and they definitely make money (or avoid losses) when someone executes inefficiently across exchanges, but there's more to it than that. HFTs are very successful in futures and non-US markets where products trade on a single exchange. How do they do that if they aren't good at predicting prices and taking risk?
Yes this is the essence of it. Based on other posts, I would guess that if you really wanted to make simultaneous trades in multiple markets, then you could. But even if HFT really did render this impossible, it's not a bad thing, since it trading in two exchanges simultaneously actually gives more rents to the large trader than the theoretical ideal of a single continuous double auction. And if you want to trade in A then trade in B based on the results of A, then you certainly don't deserve to be the first to make it to B.
- the ability to see orders ahead of someone, which this article implies you can, but you can't. - the fiduciary duty not to use someone's intent to trade against them.
HFT traders do not fall into either camp. They can't see your order before it goes to an exchange and they don't have a duty not to change their prices on one exchange in reaction to price moves on other exchanges.
No one is talking about the fiduciary duty. Fiduciary duty is just a way to define a default contract into law. E.g. brokers could write a contract saying that they promise not to front run. But since this is something that almost every client would want, it's just written into law instead.
HFT can't see your order as it arrives on exchange A before it arrives on exchange B, they can see the results of an order. That is, they can see a large trade happened on price X on A. With that information they may want to change their prices on exchange B and they may be able to do that before you can also place an order on B.
There are lots of ways to mitigate this other than using a dark pool like IEX. All the exchanges offer exotic order types to make this easier for instance, and large block traders have for years had execution algorithms that spread their orders around venues and over time to minimize price impact.
That IEX offers another option for large block traders does not bother me at all, that they market it via underhanded moralizing i find distasteful.
Finally, in this context front-running has a specific meaning and it requires a fiduciary duty for it to be front running. By using that term incorrectly you ARE bringing fiduciary duty into the conversation, whether you want to or not.
What the HFT guys are doing is not frontrunning because anyone can get hold of the same information they have (for a price). That information is not privileged (assuming deep pockets != privilege) so it's not frontrunning.
They're getting to the trade first because they're technically quicker than the routers the banks are using.
It's easy to see why you view this as unfair.
But it is fair. Anyone can buy the real-time information if they have the capital.
Obviously markets are going to be more friendly towards you the more money you throw at them.
Just like everything else in capitalism, the fairness comes from anyone being able to have money. If you think this is unfair, you're essentially opposing capitalism.
HFT says the following. If you start selling, then I will move the price downwards in accordance to my best guess as to the size of your flow. If I see serial sell orders, I will move the price subsequently lower on the evidence. I am pricing your liquidity correctly. Thus, if you are a huge hedge fund shorting the market, you face higher transaction costs because the liquidity you are demanding of the market is high, and the market maker can no longer offset your liquidity cost against ripping off smaller investors who will be stopped. This latter is the key point. In fact, for smaller investors, HFT means bid offers are much tighter. If you put a small order through the market you are instaneously filled almost at mid, thanks to HFT, whereas in the old days, retail would pay the same bid/offer for his tiny flow as huge hedge fund. The converse is true for buy orders, obviously.
Have you wondered why HFT is most hated by large institutional funds? It's because suddenly they have to pay a fair price for their huge liquidity requirements.
HFT has short-circuited the old game where large influential funds obtained tight bid offer at the expense of smaller players, and were thus able to dominate markets by trading size at unrealistically tight prices. Now we all pay for our liquidity at a fair price. It's actually really democratic (keeping in mind that competition is mega-fierce).
Ultimately, those who accuse HFT of front running are 100% correct. But there is always "statistical" front running (autocorrelation remember), HFT or Human. The difference now with HFT is that the cost of this front running is proportional to the size of your flow, and thus for smaller/retail investors it is neglibeable (because pricing is now much tigher than under the old "human" regime), and therefore positive, but for the spoiled insitutional investors / hedge funds, used to being privileged, it is back to a level playing field. Need to sell a yard? Need to pay for that buddy. Not good for them.
(obviously I am schematizing to a certain degree. Even humans price large orders wider. It's just they didn't price those large orders anything like wide enough. They sponsored the big client for the info advantage thus earned, info advantage wielded against the rest of the market: read smaller, and/or less in the know, investors).
In the old human system big funds' artificially high (cheap) liquidity is sponsored by the small player's artificially expensive liquidity and/or market move against him/her before he/she trades.
This is also why big funds want rules like trade-at (all orders must trade on-exchange) which eliminate the ability for market makers to explicitly segregate order flow. If all retail orders went to exchanges, market makers would give tighter prices on average to win the retail flows. Basically retail traders would subsidize institutional ones.
My understanding was that it's mostly the opposite, that MMs profit from uninformed order flow (so much so that they'll buy access to it from brokerages), and that informed traders were their predators in the market food chain.
At any rate: cost of trading has by every account I can find plummeted since the introduction of automatic trading. As recently as the 1990s, spreads were quoted in eighths of dollars, and there was widespread collusion to fix prices at quarters. In 70s and before, the markets were worse; basically organized fleecing operations.
On exchange markets you can't really tell who you trade with so you have to guess.
When you think about this broader proposed definition of front-running, where any anticipatory trading is "front-running", you quickly realize that virtually any informed trade could be described that way. Do extensive research on the supply chain of a billion dollar industry, notice a spike somewhere, and trade on that? You're front-running!
Exchanges make it difficult to know the current price. They sell fast data feeds to HFT shops allowing them to know about price changes before the rest of the market participants.
Exchanges have tons of order types, not just limit and market orders. There are order types that specifically benefit HFT shops by allowing them to pick off limit and market orders. If you put in a limit BUY order for $10 and the current price is $9.90. There is a big difference between executing at $10 vs $9.90 on large orders. If a HFT shop is able to push the price up to $10 (there are a few ways to do this), you just lost a lot of money and probably made the HFT a nice paycheck when the price reverts back to $9.90.
There is also a big difference between pricing something at 9.90 when you think there is very little demand for it, and being prevented from changing the price when the demand changes.
instead, an occasional crossing (once a day? once an hour?) would provide much more "fairness" since everything happens at once, at the expense of "timeliness".
back when I worked in this industry (a decade ago) POSIT provided something a lot like this
you can't have it both ways, though.
This is what prevents front running[0], according to my understanding. Once you submit an order, no one can react to that information within a time greater than round trip latency to the servers.
The problem with crossing at discrete intervals is that it introduces new kinds of strategic considerations that make it much more complex for people to play the "game".
EDIT [0] For the sake of the pedantic, I mean front running in the sense of the article at the top of this page, not the legal sense. The article defines a notion of front running that maybe not everyone agrees with, but that is what I am referring to.
That's not correct. Front-running is the crime by which your broker trades ahead of you after you state your intent to trade. Your broker does, in fact, see your trades before they reach the market (the broker is actually the one pushing them to the market). It's mainly a crime because your broker has a fiduciary duty to act in your best interest, and because the broker occupies a privileged position as an intermediary, not because there is something inherently wrong with demand anticipation.
Of course, it is correct that no one besides your broker has the physical ability to react to that information.
And going on with this analogy, wouldn't it be the case that we would want to prevent this eavesdropping if possible, because it has a similar effect to actual frontrunning, namely that the intermediary captures the information rents associated with your knowledge of the order you are about to submit.
Similarly, doctor/patient confidentiality laws are there to created a trusted relationship and make you comfortable admitting to your doctor that you had unprotected sex with lots of heroin addicted hookers in sub saharan Africa and you now need all the STD tests. In contrast, passing laws against journalists reporting on who visits hookers has no such benefit.
As a thought experiment, imagine IEX captured 95% of market share as a displayed exchange. The 350 microsecond delay wouldn't matter since it's applied to everyone and there would be no reference markets for them to price against. It'd just be another market with a slightly slow matching engine. Relative speed still matters so fast traders would have an advantage.
If you structured the market so market makers couldn't see order flow as quickly, they'd charge a larger spread to everybody. Right now they charge a tight spread so long as the incoming order is small because they know at least some of the time they can get out of the way or try to dump their shares when a big order comes in. That's good for the small guy. They wouldn't even be there for the big guy to hit if you changed the rules. Instead the sales trader would only see half the shares at the best price and end up paying multiple ticks just like he does now.
If the orders on other exchanges are uninformed slow traders, they're going to lose either way, so we're really arguing over whether the savvy hedge fund or the fast aggressive HFT should get their money. If you're playing the game of making fractions of a tick, you're in the same execution game HFTs are in, so you better get fast.
That has been forgotten in the (largely manufactured, IMHO) anti-regulation outrage. Even the big Wall Street firms should benefit from regulation that makes non-insider investors feel the market is safe enough to invest in.
That said, I'm confused and my theory fails: Despite the long run of incompetence and fraud on Wall Street, its reputation as the leading place to invest and for expertise seems to persist and it has resisted regulation to a great degree. You don't hear people say, 'don't invest in the stock market because you'll be cheated'; or 'don't hire (some major Wall Street firm) because look how they cheated these other people, and they demonstrated complete incomptence in events X and Y'. Their reputation seems immune.
Maybe there are no better options.
What IEX is attempting to do is build a platform where large traders can move lots of shares while their smaller counterparties are stuck absorbing the price impact. This is potentially useful for large traders (e.g. Goldman, JP Morgan and Citi, as mentioned in the article) but bad for small traders.
What's actually kind of "abusive" is IEX's marketing - they are encouraging unsophisticated investors to direct liquidity to them rather than having brokers route for best execution.
http://www.iextrading.com/insight/letter/
This letter directs your broker to route your trades to IEX, rather than the best available venue. This means you may be stuck paying IEX fees - which could be greater than other venue's fees - and of course, you are providing liquidity to sharks who want to make sure you absorb the price impact of their trades.
If they become a full-fledged exchange with stocks listings exclusive to IEX, then those stocks would have better "fairness" characteristics that Brad Katsuyama claims.
Therefore trading AAPL through IEX may incur a financial penalty for being on the "slow" IEX system but trading YC2019Unicorn stock exclusive to IEX may not.
It's far better to mingle your orders with HFT orders - that'll keep the big guys honest. You'll probably also pay lower fees.
Another way to put this:
"Once upon a time, I was paid a fortune by a giant investment bank to move large blocks of stock on behalf of their spectacularly wealthy clients. As recently as 2006, if one of those clients needed me to move a block of 100,000 shares, I could do that at literally the click of a button. This despite the fact that I earned a commission on the trade; it must have been a very expensive button my bank owned! And this despite the fact that my 100,000 share order was bound to move the market, and so I was in effect acting on inside information. But that's just how this is supposed to work, right?
"Anyways, the markets evolved, and my giant investment bank could no longer earn massive commissions just by pushing a single button. Even though I knew my giant hedge fund clients were going to dump vast numbers of shares on the market, depressing prices for all the other investors, the markets no longer allowed me to trivially profit from that information! My 100,000 share orders get broken into small numbers of lots just like everyone else's. No fair!"
"So I started a new exchange to rewind the markets back to the glorious, equitable, fair, transparent days of 2006."
If the direction you're heading is "modern trading is scalping profits from pension funds and mom-and-pop mutual fund retirement plans", you'd probably want to be prepared to refute Vanguard's chief investment officer, who says HFT has in general improved outcomes for one of the world's largest and most trustworthy fund managers.
For the average passive indexer who buys some ETFs or mutual funds every quarter it's meaningless at worst, and probably a net benefit since these products have tighter prices now.
If you have your retirement fund parked in an actively-managed fund where you pay an annual fee to have someone who makes $500,000/yr base continuously push "buy" and "sell" buttons on their computer screen all day, you have bigger problems than HFT.
Going for more than market growth is a strategy only brokers, fund managers and banks make profit of.
This is the heart of the problem with IEX's explanation of the markets, and why I think most people view HFT as unfair. But its not how cross exchange market making actually works. The HFTs are not rushing to buy up existing inventory from others who are selling it, in an attempt to screw others. They are the ones who are originally offering to sell it in the first place. They aren't rushing to buy, they are rushing to change the price of their inventory.
The metaphor I like to use is that of a chain of gas stations running down a highway, owned by the same people. If a tanker truck pulls up at the first two and buys up all their gasoline, it would be perfectly reasonable for those first 2 to call ahead to the rest down the highway and tell them to raise their prices, as they clearly were being used to supply someone else's industry at lower prices than they should.
That phone call is what IEX prevents, so that their customers (guys with big tanker trucks) can buy gas without impacting the price of gasoline.
So a market maker that was super fast but didn't rest orders, would be competing with one that is as fast as them and willing to rest orders. So, the resting order ones win and drive out the non-resting order ones.
This proximity is actually cheaper, more standardized and easier to access now than it's ever been.
Do you really mean it's not free?
https://en.wikipedia.org/wiki/Flash_Boys
Kovacs' _Flash Boys: Not So Fast_ which in the best possible way reads like a long-form ELI5 Reddit post about modern trading and all the WTFWAT moments in Lewis' book: http://www.amazon.com/Flash-Boys-Insiders-Perspective-High-F...
Patterson's _Dark Pools_ which tells the story of Island and the ECNs and the advent of automated trading. Patterson is more ambivalent about HFT than Kovacs, and does a good job of explaining the Core Wars phenomenon of modern trading from '98 to the mid-'00s.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2580002