If you just want to trade on a bunch of exchanges so no information flows between them, you can easily (TM) write a program that either a) lines up the orders at each exchange to execute at a specific time or b) delays the orders from a central server by the line delay.
So say NYC is 13ms from Chicago. You want to hit both at once. As long as you're not 13 ms late, nobody can see your order in one place and react at the other. You don't need an atomic clock for that, NTP will do just fine.
They're doing this because they have a reputation as a technologically advanced firm, and they know it will impress institutional investors, most of whom are still living in a time warp where spreadsheets are an advanced means of getting an edge over the market. They meet these guys, who are basically from another dimension of investing, and they suddenly need an explanation to their bosses of why RT can generate the most impressive returns of any strategy ever. The answer is "we have loads of PhD math geniuses building the strategies and amazing execution technology".
They don't need to market; their returns speak for themselves.
As obvious as the idea seems in hindsight, no one on the sell-side has a product like this right now. The closest thing is the Thor router, which is a crude attempt to accomplish the same feat because it doesn't address variation in latency. An algorithmic execution product like this would effectively end latency arbitrage, which is a source of RenTech's livelihood. To hedge against that, they have secured the IP rights to the technology.
"As obvious as the idea seems in hindsight" And the "obvious" should not be patentable. And of note, this technique has been done before in the past in different market and theirs is only an adaptation to the current equity markets.
I'm conflicted on whether or not a patent should have been granted. I've worked in equity execution for about a decade and do feel that this is novel.
On the other hand, this is a clear case of using the patent system to secure and persist an inefficiency in the market. It's hard to measure the cost of such an inefficiency to institutional investors (mutual funds, pensions, endowments, etc) but I suspect that it is well into the billions annually.
What's described in the doc seems to be an algo for snapping up everything that's available across the various markets at one instant in time (rarely that much).
If you try to grab $100M when there isn't that much around, you'll have to wait until someone puts in more orders so you can trade with them. Any market maker (including human ones) will not let you trade again at the same price if someone's just taken out the whole market, so your next tranche will be executed at a worse price.
Dripping the orders into the market is very common, but of course you leak the information by doing it. For the HFT however it isn't as obvious as seeing that an order must trigger an order on another venue (NBBO requirements) and just rushing to pull your orders from there or trading ahead of that order.
Execution is atomic: if there is a $1m 'sell' posted and you issue a $1m 'buy' against it, then either that completes or is rejected (e.g. if someone else has matched it in the meantime).
The point of this system is to send N orders to N exchanges in a simultaneous enough way that an HFT trader can't spot an order executing on exchange A and then issue their own order against exchange B.
Usually the orders are split much more fine-grained than that. This type of patent is around executing the very small slices specifically designed to scrape off what's available at the top of the market across all the various venues at once. Once that liquidity is scraped, there's a delay to let liquidity replenish and then more is scraped.
Funds can be split up for any number of reasons including investor specific requirements including leverage limits or volatility limits, etc. Also, more likely the case, their medallion fund strategies probably have some kind of capacity limit. I.e. Can't put too much capital into it before they start to suffer.
Yes, the DD team needs to say something that sounds like they've done their work properly.
So they'll come in with a list of checkboxes, which as a fund manager you learn to tick. Even if the questions have no bearing on how you're actually making money. So you might get asked what qualifications people have (very few people actually have a qualification in building strategies), or you'll get asked something quite superficial about what technology you're using (what's it written in? C or C#? Those are the same, right? That's good...)
Unfortunately, most of the DD teams I've seen do not ask the questions they need to ask. They have a long list of irrelevant questions that make sense to people who are in the CYA (cover your ass) business, not investing or coding. I never met anyone who asked me whether we used version control, and only a small sample bothered to ask whether we had our own money in the fund (one smart guy avoided a 50% blowup by doing this and discovering the big boss had barely any skin in the fund).
Got it! Thanks for the response. I wasn't sure whether that was a comment on RT's strategy (i.e. implying something else going on) or on the industry as a whole.
It'd be an interesting short white paper/post to see what a technical take on DD would be in a fund of funds/institutional investing scenario. I guess that would take away some of the value of the rise in the "consultants" we're getting calls from every day.
Its invention, developed by the firm’s co-chief executive officers, Robert Mercer and Peter Brown, first sends an order to a central server, which breaks it up into multiple smaller orders. Those are then routed to venues that offer the best prices and most liquidity, much the same as brokers do now.
But before that happens, the smaller orders are sent to servers located as close to the exchanges as possible, along with instructions on the precise times they should be executed. The co-located servers sync their transactions so HFT firms won’t have enough time to identify an order on one exchange and then race to another to trade against it.
A crucial part of the system is the optical, atomic or GPS clocks that will be used synchronize those orders. Renaissance says in its application that GPS clocks are accurate to within nanoseconds and any time differences between them are “too small to be perceived” by HFT firms.
Maybe I'm missing something but sending orders ahead and releasing at a specific time is obvious is it not? If you add a really accurate clock suddenly it's patentable?
I don't know why you need a great clock either, if you have stable, symmetric network paths from a central location to all your servers colocated at exchanges, you can predict the delay between sending from the server and getting to the exchange, you can split your order and send it to the various exchanges with appropriate delays and know that everything will arrive at the same time. If you're wrong, it's going to still be close enough that nobody will be able to see it on one exchange and react to it on another before your order gets there.
This was the first thing I thought of when hearing the flash boys story on the radio: The banker was complaining he couldn't capture the whole book across exchanges because resting orders were cancelled before his order got there -- he just needs to get his orders to arrive close enough in time (although expect a bigger tick, probably)
I've evolved to keep this rule in mind when I hear almost any story/portrayal/claim/news anymore. Especially anything that feels sensational or too-good-to-be-true-at-first.
For example, people often lie about sex or money. National governments often lie about, well, pretty much anything that suits their best interests. When I say "lie" I don't mean completely wrong or totally false, merely, spun a certain way, sometimes a careful omission of critical modifiers, the use of weasel words, etc.
Network delay variance over distances is much greater than the timescales at which HFTs function. That's why HFT equipment is always so close to the exchange itself.
But it's not significant in the domain this thread is focused on. In order to lose signal to an rival in this scenario, the delay between your Order A hitting Exchange A and your Order B hitting Exchange B would need to be sufficient to give the rival time to learn the signal, and then to act on it. At a minimum, that is half the cost of the round-trip time between the Exchange A colo and the Exchange B colo. So long as you're not introducing such substantial delays in your order dispatch, you're fine.
I havent read the patent but I can say this is an exceedingly common (I'd probably say standard) strategy. I can only assume the atomic clock bit is what's novel.
Using ntp or whatever the new variant is is also standard, which as I recall can hit sub-microsecond consistency on a wide area network with good hardware. So yeah, not new.
It doesn't require hardware, though it really improves the performance. I've implemented 1588 a few times and was able to achieve ~ <30ns accuracies when using hardware timestamping. Also note that there are more and more MACs and PHYs these days that offer HW timestamping.
With software, it really depends on how deterministic your packet handling and timestamping routines are (or how deterministic the OS scheduler is). I was able to achieve accuracies of less than a microsecond on a Linux system, but it was "touchy".
For reference, there's an open source implementation called "ptpd" and "ptpd2".
Little know fact, you can use the GPS constellation to get atomic level precision time nearly anywhere on earth. Using an Atomic clock is purely to show off to investors/a red herring.
Exactly. GPS is by far an easier way of synchronizing clocks to an atomic reference compared to any master-slave networked approach, especially in situations where we're talking about a small number of stationary, expensive servers which are far apart.
PTP is mainly useful for situations where you want to synchronize many cheaper slave devices to a common master and it's not practical for each device to have its own GPS receiver, or for situations where the use of GPS isn't practical (or is prohibited) and you're more concerned about coherency between devices rather than traceability to a primary time reference (e.g. a telemetry network on an aircraft). Although, generally, the grand master of a PTP network is synchronized to GPS anyway.
Of course, you could achieve actual phase locked synchronization down to the clock cycle with something like SyncE + PTP, but with GPS, you need not worry about issues with asymmetry regarding messages transmitted over the internet (PTP needs to be routed through PTP capable switches which compensate for the residence time and was really meant for LANs).
I guess it really depends on your constraints, but if you’re able to use GPS, that would definitely be my first pick when it comes to synchronizing multiple devices.
The latest UBlox timing receiver (LEA-M8F) provides a PPS which is accurate to less than 20 nanoseconds (to the UTC second) and its built in oscillator has a typical holdover spec of 0.025 PPM (25 nanoseconds per second). If you want to get fancy, you can use the PPS to discipline an OCXO and get an even better holdover spec to handle the situations where your receiver may lose lock (which is unlikely if you’re able to have an antenna).
Basically, the accuracy of the UBlox GPS receivers (just an example since they're pretty cheap; I found a board for ~$150), is equal to or better than that of a usual PTP link (without SyncE), so you might as well just use GPS on each device if you can. It is simpler, IMO.
However, note that comparing GPS to PTP isn't necessarily valid since PTP is purely a method of conveying timing information between devices, and is not a time source itself, where as GPS is both a method of conveying timing information as well as a time source. In other words, a PTP network still needs a master device which itself is synchronized to (or is) an atomic clock.
The errors you see in your mobile phone's positioning are due to signal problems (reflections, etc) and the relatively limited capabilities of the cheap GPS radio in your phone. A decent GPS receiver with a well-positioned antenna will get a highly accurate clock.
Speed of light is 299,792,458 m/s. So if GPS is off by more than 1/10,000,000 you can't get accurate within 30 meters. Having used a GPS they are better than that, thus the clock must also be at least that accurate. Of note, stationary stations can get into centimeter precision which imply's vastly higher accuracy.
Except that the state of art in HFT is sub-microseconds
"London-based trading technology company Fixnetix said Tuesday it has the world’s fastest trading application, a microchip that prepares a trade in 740 billionths of a second, or nanoseconds." (WSJ, 2011)
That's one of the factors. Generally the optimisations that happen first are on network path length / network equipment induced delays, as there's relatively cheap and quick gains to be made. The bigger delays are invariably in your applications that are processing or generating data, which are more costly to optimise.
That said, the Fixnetix stuff is only talking about one aspect of what's involved, and about as representative of reality as Cisco's published WARP speed figures in their Nexus 3500 range.
The atomic clock bit isn't novel at all. I've worked for HFT firms the past 9ish years and using hardware timesources is 101 level intro to electronic trading.
The actual patent talks a lot about NIST GPS clocks, and not so much about atomic clocks. Never trust a headline. Gell Mann Amnesia Effect in full play here.
Sure. You can't get roof access (for a gps antenna) or a vendor ptp feed in every exchange. In those places, you get a rubidium decay stratum 0 timesource.
Not all businesses can afford this, but it is only 4 or 5x the price of a normal GPS timesource, which is affordable for the right people.
The idea doesn't help the hft'ers. It takes an order, secretly transmits it to computers each as near to the major markets as possible, with instructions so that the computers submit the trade offer at precisely the same time.
The hft'ers can't make money since they can't outrun trade offers that are synchronous across all markets.
As someone with zero domain knowledge, why aren't the exchanges already doing precision timed order processing? That just seems like it's should be a standard feature across the board. The broker sends buy/sell orders with planned execution times to all the required exchanges and the exchanges sit on the orders until the designated time.
Exchanges do time ordered processing on their own exchange (with different levels of precision). I don't know of any exchanges that offer execution time as a constraint, but new order types can be created if they were deemed valuable (it takes SEC approval).
That said, it wouldn't alleviate the issue necessarily. If firms detect problems in the clock sync between exchanges you are right back to the same problem, and now you've added a complex bit of tech that requires a bunch of competitors to agree on.
This seems, to me at least, to be one of those problems that it is better to let the problem surface than to try to alleviate with an abstraction layer that is leaky and error prone.
But this technology is patent is just implementing the exact same thing at one layer removed from the exchange. You still have to time the orders and you still have to keep the timed orders confidential. To me, using a 3rd party to do this instead of having it as part of the base system is... silly I guess.
What? It generally only costs a few thousand dollars per month to colocate servers next to exchanges.
Arguably, it is more fair now than it ever was in legacy "open outcry" markets where the size of the floor was fixed and if you didn't get a spot on it you weren't able to compete.
Disclaimer: I've worked in HFT 9ish years (10 soon)
All they're saying is that with absolute synchronicity among all of the clocks at all of their co-located servers, all pieces of the order are executed at multiple exchanges at precisely the same time. Even very small differences among the clocks at each one can create opportunity for others to step in front of the trade, and this helps them avoid that. While this may sound obvious, they wouldn't be doing it if it hadn't been a problem for them in the past.
Nanosecond differences are too small to take advantage of, that's only 30cm. My guess is that you'd be at least in (or close to) the microsecond range before you'd worry about HFT stepping in front of your orders.
If you're hitting multiple exchanges, then you can take milliseconds and still be fine.
One of the things HFT firms take advantage of is increasing a nanosecond lead into a microsecond (or more) lead by route optimization. If they can get information from one exchange to another faster than the original order, they can effectively trade by looking into the future.
Many firms invest heavily in direct microwave links for paths normally served by fiber because of the speed advantage.
Speaking from ignorance here but 10s of milliseconds sounds way too high.
This Wikipedia article [1] shows the standard Chicago to NJ connection at 14.5ms roundtrip and the dark fiber line Michael Lewis talks about in Flash Boys at 13ms.
Microseconds are not enough, because the arbitrage works by the HFT setting the order in one place, and then rushing the arbitrage to the other place. As long as you're more coordinated than speed of light between the two locations, you're fine.
Also, GPS clocks are cheap and precise, atomic clocks seem gold plating. (Which sometimes is actually necessary in electronics, BTW...)
I think the main innovation in Spanner is the notion of Truetime not the atomic/gps clocks though. In fact MSR has come up with something called ClockSI that does global snapshots with a modified NTP I believe.
Unfortunately for the traders, two of the clocks were at altitude and so time passed more slowly for them, resulting in several femptoseconds of misalignment :-)
And, boy, do the off-the-shelf commercial offerings suck. The vendors are not really used to dealing with the stress that Google puts these things under.
I agree. Atomic clocks are an implementation detail, and also overkill (at least during this decade). I've worked in HFT (albeit in forex, not stocks), and there are always internal delays within the exchange that you cannot predict with anywhere near the accuracy of an atomic clock. (Well, if you could, then that would be a true innovation.)
Hedge Funds don't patent strategies. They keep them as trade secrets. It would make more sense to me if this filing was part of an attempt to build a patent portfolio for defensive purposes, as tech companies do.
The way many patents are structured begins with "a method of..."
So, in that sense, "a method of coordinating orders across exchanges to minimize analysis time available to other traders" seems perfectly in line.
(I agree that from various computing-centric backgrounds, this might be trivial, but not all problem domains are well-saturated with computing expertise, and this might well be sufficiently novel to warrant a patent in the domain.)
> The co-located servers sync their transactions so HFT firms won’t have enough time to identify an order on one exchange and then race to another to trade against it.
That sort of sounds like DDOS to me. They patented a DDOS botnet.
I assume the traders can't place orders with time of execution (to be executed at the specified time, kept secret until then).
If the market accepted the time of execution from traders and kept the trades secret until they were executed, then there would be no need for these patents.
There may be some exchanges that accept some sort of specialized order type that allows for time of execution as a constraint. That said, this issue is about synchronizing between exchanges and there certainly aren't any exchanges that collaborate to do that.
This reminds me of how Google uses atomic clock and GPS for Spanner [1]
Google:
"“We can commit data at two different locations — say the West Coast [of the United States] and Europe — and still have some agreed upon ordering between them,” Fikes says, “So, if the West Coast write happens first and then the one in Europe happens, the whole system knows that — and there’s no possibility of them being viewed in a different order.”"
Renaissance Technology:
"Replete with schematic drawings, the filing describes a novel way for “executing synchronized trades in multiple exchanges.” The invention consists of not only sophisticated algorithms and a host of computer servers, but atomic clocks -- precisely calibrated to vibrations of irradiated cesium atoms -- to sync orders to within a few billionths of a second."
To translate what Renaissance is doing in technical parallel, they are trying to do a synchronous commit at multiple locations/exchanges at the same time. Submitting a trade to an exchange can be viewed similarly to committing data to a data center. By using atomic clock, synchronize these writes across multiple locations in effect eliminating HFT from jumping in.
If anyone is looking into prior art on this, Spanner is probably the closest I can think of. (I am not a patent attorney and don't want to turn this into a patent debate).
Reminds me how Einstein derived special relativity. He was thinking about synchronizing clocks at railroad stations across the continent. The idea of simultaneity depends on the location of the observer. The traveltime of light crossing the globe in 1/14 a second matters little at train speeds. But thats an eternity for HFT and noticeable in a google search.
These issue crop up on integrated circuits where clock cycles are shrinking and die sizes are increasing. Defining simultaneity where the microinstruction cycle time is much shorter than the chip or board propagation time is tricky. Each design has their methods to deal with this.
Generally speaking, as I'm not a patent lawyer either, the same technique applied to two different problem domains can generate two valid patents. The historical example is ship and automobile windscreen wiper.
Not to argue with you, but I don't see the problem any different from trying to synchronize two commits in a database. The same technique is not limited to Spanner or algorithmic trading, but other fields as well. It's not so much different from DHT or other algorithms, which have applications in multiple domains.
In a sane world wiping liquids off of smooth, hard surfaces would be the same domain whether the material under whatever the wiper is attached to is in the liquid, solid or gaseous phase.
In this case both observers are continuously accelerating towards the center of the Earth (because the Earth is spinning). So the theory actually doesn't work!
Aren't we accelerating _away_ from the Earth to an even larger extent as we are held up by its surface, against the natural motion of falling downwards?
It's also the method by which signals from radio telescopes thousands of miles apart have been correlated after the fact at a central location. This has been the case for a couple of decades now.
What does that have to do with transactions? Analyzing radio signals is simple compared to running transactions, since it is only a one-directional signal flow.
A few weird things stand out to me: (1) Renaissance is super secretive. If they want to use this strategy to make money, a patent reveals to competitors what they're doing and creates more issues than it seems to resolve. (2) Renaissance is an HFT firm. Why are they interested in thwarting HFT? (3) This really isn't that fancy an idea. It's fairly general: send orders ahead to co-located servers to be executed at specific times.
I wonder if what they're really trying to do is prevent banks or others from creating anti-HFT infrastructure, and then providing it as a service to market participants that want to place large orders. The patent would perhaps provide some protection in that case.
Think about it this way. They are publicly advertising to their investors that they have a weapon to defeat HFT. Because of this, they will get better returns.
When an investor decides to put their money in a fund, are they going to choose the one that has a patented defense against HFT or will that investor put their money into a fund that has a known vulnerability?
>It's fairly general: send orders ahead to co-located servers to be executed at specific times.
This is harder than it sounds. Coordinating a number of servers in different locations to send a trade at a specific time is not that straightforward when you're dealing with microsecond or nanosecond transactions.
Forget the trade, just trying to synchronize the time on all of the different servers in different locations is a large challenge. NTP has an accuracy of 10ms on the open internet [1]. The state of art in HFT is sub-microseconds [2]. This is at least 2 orders of magnitude faster than the NTP margin of error.
A hedge fund with track record of Reinessance Technologies does not have to advertise to investors in any way. They just choose not to raise more capital because in some ways they are unable to push it to such returns (when you are too big, you can't move as quickly without moving the market against you).
You have to keep in mind that the mentioned sub-microseconds are the _reaction time_.
HFT can't beat the speed of light either.
Therefore if your orders are synchronized to better than the speed of light distance to the next exchange, HFT traders won't be able to profit off them (-> you have milliseconds).
With GPS receivers or very good network connectivity to your ntp server (ie. not home dsl) you can easily get down to below 1ms.
I disagree that the problem the patent purports to solve is particularly difficult, given the technology that is already readily available (PTP, an OS with a high resolution clock, machines located in exchange data centers).
Renaissance is not that type of hft firm. Renaissance uses algorithms to predict price movements before they happen. The type of hft this system is designed to prevent is front-running. Which is me seeing your order on exchange A and buying ahead of you on exchange B before your order arrives.
Those types of hft firms are surely eating into Renaissance's profits in a big way.
Well, it is not illegal, that is correct. Whether or not it is 'front running' is a matter of debate.
It certainly seems to me that they are running out in front of the order. I'm not sure how exactly you might be contorting the meaning of 'front running' for this not to apply.
Front running is trading in front of an unexecuted order, which you have because of your fiduciary responsibility to a client.
You're discussing trading in response to an executed order, which you have as a member of the general public.
If that's front running, then what isn't? Is my purchasing wheat futures in response to news of an predicted drought "front running" the orders that bakeries will be making?
'Front running' has specific legal definition (which has existed for a long time) which definitely doesn't apply to these cases. The contortion is in fact on your side.
It's worth nothing that eating into Renaissance's profits in this way is good for everyone else because it means that accurate prices are reaching the market faster.
If a firm is able to consistently get to all the other exchanges first, they can shave pennies off of a large percent of orders. That's a pretty big downside for everyone that trades.
They get accurate prices to the market faster, but only by a millisecond or so. That's a miniscule upside.
If market makers (i.e. most HTF firms) weren't able react to large market moving trades very quickly, they would have to keep spreads wider to manage the risk of adverse selection. This increases the cost for everybody for the benefit of a few large investors.
You could say the same thing about insider trading. So, the argument that accurate prices reach the market faster is not, in and of itself, an argument that something is good.
I am on the fence myself, not having thought about the matter too much. Instinctively, I'd say insider trading should be handled as a breach of contract (ie if I trade on something that I signed an NDA for), not as a criminal matter.
What you are describing is an acausal (i.e. physically impossible, since cause effect happens before cause) version of demand anticipation - changing prices in response to market demand.
It's impossible because the HFT will only know your order has reached exchange A after exchange A has told him about it. Obviously A can't tell him about it until after your order has arrived.
Front running is a strategy where your broker sees your order, trades ahead of it, and then routes your order to the market. It's highly illegal.
Perhaps the parent comment was edited after you posted, but while what is described is not "front running", it's not acausal. The observation is happening at Exchange A, the reaction at Exchange B.
If he sees an order on exchange A (which is completely anonymized and disconnected from anything on exchange B), he cannot know if you are making an order on exchange B.
"Buying ahead of you" requires some kind of advance knowledge, and in the scenario described you simply don't have it.
If there's a pattern of sloppy traders trying to fill large orders by splitting them across exchanges and firing them off at roughly the same time, then having seen a trade at one exchange and not yet at another, it might be a good guess that a trade was incoming, if I'm faster enough to make those decisions.
I don't really know if that is the case (or ever was).
Suppose there is a pattern of 10% of traders being sloppy, and 90% being not sloppy. If you run this strategy, you lose 90% of the time.
Additionally, whenever a small trader comes along, you are again overreacting. I.e., I make quite a few (automated) trades. I never cross exchanges or blow up more than 1 level. Whenever I trade you are again buying all the shares and probably losing money.
What I described as acausal is responding to an order at BATS before it gets there. Guessing that maybe an order might go to BATS because you saw one at ARCA isn't acausal.
Did you not read my example? We are talking here about multiple, physically separated exchanges.
Example:
Vanguard sends a buy order of 1000 shares of AAPL to exchange A. Some HFT firm sees that order on exchange A and knows that it will also be placed on exchange B, and was surely broadcast from Vanguard's trading floor at the exact same moment as the order to exchange A.
However, the HFT firm also knows that Vanguard is physically closer to exchange A then B, and the HFT firm has invested a lot of money to ensure that they have the physically shortest possible route from themselves to exchange B.
This allows them to broadcast their order to exchange B and have it arrive there before the order broadcast by Vanguard, even though they broadcast their order later.
It is somewhat difficult to wrap your head around, but it is a physical reality, and people are making millions of dollars from it as I write this.
A) 'yummyfajitas has worked in hft and written several blog posts on the subject.
B) he is reacting specifically to the term front-running which has a known technical definition in trading and cannot happen the way you describe (I've said elsewhere I believe the term has been appropriated and redefined and am not willing to fight that anymore, maybe he is).
C) it's funny that you mention Vanguard as they have been pretty adamant that HFT market makers of the form that use latency are save them money.
He does not see "Vanguard" - the trade confirmation is anonymous. It could even be him! (Yes, you are subscribed to a multicast trade confirmation feed and even your own trade confirmations are anonymized.) He does not see Vanguard's physical location - for all he knows Vanguard is closer than he is.
So actually, all the HFT knows is that someone bought some AAPL.
Now what? Whenever someone buys AAPL he goes out and buys a bunch more? That doesn't sound like a moneymaking strategy to me.
What evidence do you have that people are making millions of dollars from this as you write this? Michael Lewis?
Actual front running requires your broker to violate the fiduciary responsibility they have to you, and violating fiduciary responsibilities is both immoral and illegal.
The behaviour your discussing does not involve any fiduciary responsibilities.
> This behavior, however, is identical to it in spirit and moral character.
That seems self-evidently false, but okay, I'll bite: What moral precept is being violated here, and why is it "identical" to the fairly serious sin of someone with a fiduciary responsibility to you violating your trust?
Let's play fill in the blanks: "Goldman Sachs wants to buy $400m stock in Apple, but after they buy $20m, a member of the public sees the strange pattern of executed orders, guesses that someone is buying a lot of Apple stock, and starts buying up stock too. This is highly immoral, because members of the public have a ______ duty to ______, and not allowing Goldman Sachs to manipulate the market in peace violates it."
What phrases can we put in the blanks that makes that not nonsense? I'm pretty sure it's not a fiduciary duty, and it seems quite clear it's not to Goldman. What duty is it, and to whom is it owed?
With that definition of "front running" people are doing the same interception as 'true' front runners, but they happen to be third parties. People are using the best available term, and making it slightly more general. What would you rather have them call it? They can't call it nothing.
Just like people will call things "insider trading" when there is trading based on insider information, whether or not it happened to be legal.
That argument is a bit like calling all sex rape, because someone having sex is doing the same thing as a rapist does, they just happen to have consent.
The reason front running is illegal is because it is a violation of a fiduciary duty. Third parties do not have a fiduciary duty, therefore it's not illegal. "It's front running, but without the violation of the fiduciary duty" is like "murder, but without the killing someone", or "fraud, but without the deception".
> What would you rather have them call it? They can't call it nothing.
Not everything needs a name. Since what you seem to be describing is "reacting to the public actions of other market participants", does it need a name?
Alternatively, if you think something serious is going on, why don't you define it, and then we can name it?
> The reason front running is illegal is because it is a violation of a fiduciary duty.
And nobody is calling this kind of "front running" illegal.
> like "murder, but without the killing someone", or "fraud, but without the deception".
In a murder analogy, you'd keep the killing but change something else, maybe it's properly manslaughter but people will still call you a murderer.
For "fraud without deception", let's look at what that entails. You lie. The victim knows you're lying, no deception involved, but they rely on your word. The victim is harmed. I think that's close enough to allow people to call it fraud, even if legally it's slightly different.
> "reacting to the public actions of other market participants"
The reason people dislike it is because it's not just 'reacting'. They're getting in their own action before the thing they're reacting to has finished.
> why don't you define it
Darawk defined this version for us. "front-running. Which is me seeing your order on exchange A and buying ahead of you on exchange B before your order arrives."
> The reason people dislike it is because it's not just 'reacting'. [...] Darawk defined this version for us. "front-running. Which is me seeing your order on exchange A and buying ahead of you on exchange B before your order arrives."
Seeing a completed order on exchange A, and speculatively purchasing stock on B in the hopes that the buyer might later buy stock on B sure sound reactive to me.
> also interrupting. People dislike the interrupting.
So your argument is...what?
Hypothetically: Let's say I have an inventory of stock A, which I think is worth X, and which try and sell whenever the market price climbs above X. Now there's some new public knowledge that materially impacts my estimation of the value of that stock (eg, a large hedge fund has started buying large blocks of this stock): I no longer think it's worth X, but actually Y, and I'd like to stop selling it whenever the price climbs above X, and instead wait until the price climbs above Y.
So you're saying that's okay, and I can price my inventory however I want, but only if I give the hedge fund a chance to buy a bunch of underpriced stock first? This raises some questions such as:
1) Why on earth is that a good rule?
2) How much time do I need to give the hedge fund? Do they only need a few seconds? Should they get a day? A week? At what point am I allowed to change the price I'm selling stock A for without it "interrupting" the hedge fund? Do they need to announce that they're done, or is their a timeout period after which I can just assume? Can I change the price I charge other people if I still let the hedge fund buy at the old price, or does everyone get the discount?
3) Or is it not about time, but about amount? Does the hedge fund have some divine right to buy as much stock as they want without it driving the price up? Why? And how come nobody else has that right? Do you have to be a hedge fund to get the right to name your own price, or do normal people get to do that too?
4) Or is it somehow okay if I'm selling, but not buying? Is it's okay for me to raise the price I'm willing to sell stock A for if I find out a hedge fund is investing, but not okay for me to raise the price I'm willing to buy stock A for? What happens if I find out a hedge fund is liquidating their position instead? Do I get to lower the price I'm willing to buy and sell it for, or just one of them? There's no laws or SEC regs about this; is there a list of rules somewhere? Is it in the bible?
5) Does this only count market actions? If the hedge fund gives a Bill Ackman style press release about how some company is terrible and should be prosecuted, can I change my prices immediately, or do I need to wait in case I'm interrupting some hedge fund strategy? I mean, maybe they were planning on giving the press conference and then buying a bunch of stock; if I think their arguments are bollocks and I buy a bunch first, is that interrupting them?
And so on. The entire argument seem awfully focused on why large hedge funds and investment banks should be able to ignore basic market rules. I'm sure they'd like to; I'm still waiting to hear why they should. (One of my favourite scenes for Lewis's Flash Boys was when a trader expresses outrage that his very large order in a thinly traded stock caused the price to move against him. How terrible; if only there was a law that required people to trade with him at the price he chose...)
For this particular case I can make the answer very simple. How about waiting half a second.
Can you not see how it's bad in the specific case where they already issued the order to all exchanges but your order gets processed first because you used a different cable? Ignore the more ambiguous cases for the moment.
Why is protecting the interests of hedge funds and investment banks important enough it needs a special rule? This isn't a rule that will benefit the little guy; it strictly benefits the biggest fish.
> Can you not see how it's bad in the specific case where they already issued the order to all exchanges but your order gets processed first because you used a different cable?
To be clear, your concern is strictly that if a hedge fund is buying a very large amount of stock, this will cause the price to move against them as people react to it, and you think they should get a full half second (an eternity at the speed of the modern market) to buy as much as they want before people are legally allowed to react?
That sounds like a terrible idea, and in the specific case you list: No, I don't see why that's bad. I don't see why anyone except a large hedge fund or investment bank would.
Further: Once the order hits the market, you're saying there should be a half second window during which no one can do anything except the hedge fund. But as soon as that half second window closes, there will be a race to (finally) react. Which will be won by...the HFT firms, right? Won't they be able to, hypothetically, get their rSo even if we accept your premise, isn't this just shifting the victims around without fixing anything?
"Half a second" was not meant to mean "enforce a delay of precisely 500000 microseconds". The point is that the problem would be solved if traders would chill out for the blink of an eye. The fact that it's hard to force people to chill out is a completely separate issue.
> To be clear, your concern is strictly that if a hedge fund is buying a very large amount of stock, this will cause the price to move against them as people react to it,
> in the specific case you list: No, I don't see why that's bad.
The original scenario has nothing to do with order size. The scenario is that X sends a simultaneous order to multiple exchanges, but while it's still in transit to most of the exchanges Y reacts to the order and submits their own on a faster cable, getting there before the order they're reacting to. I think such an outcome is clearly bad. I won't suggest a fix to avoid distraction. Do you disagree with it being bad? Picture it happening to an old lady if you have no sympathy for hedge funds.
Are you saying front-running is not an HFT strategy? Because that is incorrect.
In equities markets, for trades of a sufficient size it's not possible to place the order all at once. You'll hammer the order book, and take very suboptimal pricing for the latest marginal shares you buy/sell. Therefore you have to split up the order into chunks.
HFT firms will try to detect when equities traders are doing this. e.g. If they see a pattern indicating that chunks of shares are being bought, they will try to buy, too. When the rest of the order comes through, the price will rise and that firm will make money. That's why "front-running" is just a special-case of an algorithm to predict price movements.
I seriously doubt that a HFT fund as large as Rennaissance uses a single strategy, be it satellite images, or front-running or what-have-you.
HFT firms will try to detect when equities traders are doing this. e.g. If they see a pattern indicating that chunks of shares are being bought, they will try to buy, too.
Price discrimination against large traders - grandma gets a better price than Bill Ackman - is a valid and real HFT strategy. They do it in the manner you describe.
But it's not front running. It's only front running when your agent (usually your broker) does it based on private information.
Front running is just a term Michael Lewis misused in his misleading advertisement for IEX.
>Are you saying front-running is not an HFT strategy? Because that is incorrect.
I think dsl is quite familiar with the sort of HFT strategy you describe. I think s/he is just trying to correct your use of the term front-running, which was redefined by Michael Lewis et al. from its previous definition as the objectively illegal activity by a broker-dealer. Based on comments like yours, it's a losing battle.
Renaissance also reportedly charges a 5% / 44% fee instead of the standard 2% / 20%.
Quick math (which is wrong since they've changed their fee structure) -- If you had invested $1M with Renaissance in 1994, using a 5/44, you'd end up with something like $411M in 2014. Ren. would've made about $430M.
A basic summary is that market-makers add liquidity to the market and profit from their bid-ask spread, increasing the execution-speed and depth of the market. Many of the 'predatory' pricing strategies attributed to them by Lewis and others appear to be impossible when you try and write down the pseudo-code + order book that corresponds to the allegation.
RenTec's Medallion Fund has obviously done amazingly well - almost to the point where one wonders how it is even remotely possible. But their other funds have had more average performance. Maybe this move isn't about Medallion at all - maybe it about getting to better execution for their other funds in order to improve returns.
This assumes you're willing to trust their software and hardware.
I suspect they're right that this is a far more effective and comprehensive approach than IEX.
I think IEX is going to nail them to the wall because their target market understands and trusts a giant ball of cables having a particular length, but can achieve neither when faced with a giant ball of computer science.
Other than the first sentence, this comment was about potential customer response; I don't have an educated opinion about how trustworthy and/or effective any given exchange software is.
They don't care about "target markets" or what anyone else thinks. This technology is to protect their own trades.
The goal is to be able to execute buys across multiple exchanges (because the orders are so large) without other high frequency firms being able to see a trade on one exchange, then buy and resell stock to them at a higher price on another exchange.
> without other high frequency firms being able to see a trade on one exchange, then buy and resell stock to them at a higher price on another exchange
This is a pretty common misconception of how latency arbitrage works. In reality the other HFT are not buying/selling new orders. Instead what they are doing is cancelling or modifying their existing orders so that they don't get hit by incoming orders.
HFT firms can have orders that have been resting for a very long time (days/weeks depending on the exchanges/risk rules) and you will never be able to get an order now in front of an order from last week, no matter how fast you are.
That 100% depends on the exchange actually. Not all exchanges have priority rules like that. It certainly is the case for a lot of futures markets, but definitely not for many places for Equities, Options, or even FX.
Can't really say sadly, but it is true. But it is normally the case for Limits, except not all orders are limit orders. You're right mostly, for a certain type of strategy, but not all of them. Ironically I just looked up your profile and on the link you mention Flash Boys: Not So Fast.
I used to work with Peter Kovac, that book's author, at Madison Tyler Technology / EWT LLC (which is now Virtu Financial). He's one of the smartest guys I've ever met and that was a pretty decent book.
Its been awhile since I perused order type documentation so anything is possible, but I don't ever remember seeing an order type that can jump a resting limit order from the past on any any exchange I've looked at.
I have seen odd priority rules around how orders get around sweep or self match protections and the like, but not jumping a resting, visible order.
Given the order type doco is public for SEC regulated exchanges I doubt your IP agreement restricts you from slipping me a link right? ;)
ISO limits that create a new price are usually given better priority than Hide-not-slide orders entered with the equivalent working price, even if they were accepted before the ISO limit. Depends on the exchange obviously.
Sure and the non-visible parts of icebergs end up getting later priority than new limits at the same level. I'm not arguing that there aren't order types you can do that will put you at the back of the line priority wise, I'm suggesting the alternative.
If I put a vanilla limit at a level and don't cancel it (and its GTC) today, I've never seen an order type that would allow someone to jump me in the priority queue tomorrow.
See big sell order at getting executed at $10.30, $10.20, $10.10 at Exchange A. Short sell at $10.30 on Exchange B. When the big order comes, rebuy at $10.20.
"In reality the other HFT are not buying/selling new orders. Instead what they are doing is cancelling or modifying their existing orders so that they don't get hit by incoming orders."
Right. One of the strategies is:
- Put in standing order to sell a small amount of security slightly below market and leave it active.
- Wait until a buy order triggers it.
- Buy same security faster than rest of buy order can be processed.
- Sell security just bought at higher price.
- Profit.
There are lots of variations on this, but that's the basic idea. It has the profitability of front-running, but is legal.
This strategy could work, but only on exchanges where your own private trade confirmations are faster than the public quote broadcasts, and only if you can use the information to aggressively trade a different, but highly correlated security.
> Buy same security faster than rest of buy order can be processed.
How? You can't because your competition is already resting orders there (from potentially weeks ago) and any order you put in there will be behind your competitors no matter how fast you are.
What does happen is that you all are resting orders up and down the order book. When a big order takes out several levels of your orders, you race as fast as possible to cancel those same levels at other exchanges. If you are faster than the big order you only get hit for a single exchange.
Its a risk mitigation technique, to keep your profits from your regular business of making the spread, not risk free profit.
So one place firms are racing is on the cancel side. The other is on filling back in orders after the big order has cleared levels. The faster you are there the better priority your resting orders will have (potentially weeks from now).
Would eliminating the sub-penny rule remove the need to use this strategy? If there were basically an infinite number of points orders could be resting, you could just get in an epsilon above or below some other fund's order to get ahead in the queue?
Exactly. This is a defensive technology to defend against other HFT players.
Think of it this way:
- In the past (prior to HFT), when you put in an order, you know where the market is going to be. How you win is by building better financial models to identify which security to buy/sell.
- With HFT, traders, without sophisticated tools, can no longer trust the market because once they submit, HFT algorithms can shift the market. The focus now is to build a better HFT product to win in the marketplace.
- What Renaissance is doing here is neutralizing the effect of the HFT on their trades. Once they neutralize the "enemy", they can go back to doing what they are great at: financial modeling.
Throughout history, power has shifted with changes in weapon technology. This is just a shift of power back to traditional fund managers from HFTs.
When I said "where the market is going to be" I meant that from one second to the next, you knew it wouldn't move that far.
In the past, when you called up your broker to buy a stock, he could give you a quote (say 12.35) and you tell him to buy, you know that the price will be the quoted amount. (Market in finance lingo in this case means current price of the stock)
If you need to sell a large batch of stock, you know the broker can sell it to multiple people and get you the best price.
With HFT, trades are moving so much faster and you can get a quote from the exchange (someone want to buy 500 shares @ 12.35) and by the time your trade gets sent from your computer to the exchange, that offer is now 400 shares at $12.50. The market has moved in milliseconds. In other words, you cannot trust the market (current price) because you don't know where it will be in the milliseconds from viewing the quote to entering the trade.
In Flash Boys, they talked about ghosts in the machines. This is what they were referring to. When they entered their trades, it's as if the market knew beforehand and moved against you.
(For finance savvy people, apologies, I've simplified it in lieu of understanding)
The reason that the market wouldn't move that far is that the market makers were quoting a much bigger spread. That is the price was always way worse (a quarter at the minimum).
You can still get that pricing level almost everywhere by quoting through the market at that price difference.
An exchange could work just as well and provide just as much liquidity if it accepted sealed bids into a queue for one minute, then settled and showed the full queue, while accepting sealed bids for the next minute.
HFT would no longer be a thing, and everyone would trade on more equal footing.
("One minute" is a guess. Could be right interval is 20 seconds or ten minutes or whatever... But needs to be slow enough to allow full dissemination and reasonable time for sealed bids.)
There are exchanges that operate with similar rules, like POSIT. http://www.itg.com/product/posit-3/ But the slowness is a lot more, executing a few times a day. And you can't change your order AFAIK (and there's large random timespans). There's also complications to avoid revealing how big the buy or sell side are.
As far as your suggestion, that's what I was thinking for a while when I first read about them. I don't think it works. Simply cutting over every minute doesn't help, since there'd be a race near the edge. Like eBay, there's zero point in submitting a bid until the very last (milli|micro|nano)second.
It also doesn't work because other exchanges exist. For instance, suppose the interval is an hour. You submit at 0:01, then by 0:30 the price has moved on other exchanges. Obviously you'll want to change your order, and you still have time. But the price is changing on other marketplaces every nanosecond, so you're going to be changing your bids right up until 0:59.999999. Once you allow changing the order, you're back to racing.
If this was the only exchange type, and they all were synchronized, it might work a bit better. But you'd run into another issue: no one would submit an order until the very last microsecond, to use all available information. There's probably more complications too; I don't really know anything about this stuff.
HFT isn't a bad thing, or at least, I've not heard why it's actually bad. Other than runaway programs screwing things up. Which, I think is actually great. That big flash crash was fucking hilarious. People had sell orders in at 0.01 or market or whatever, and they executed, then they got all upset that things worked like they should!
HFT is only a bad thing, because it's removes bright minds from working on other problems; not because it hurts other market participants (which it doesn't---apart from being competition to other market makers).
I've read some interesting proposal that removing the subpenny rule would make speed less of a concern, since it would be possible to undercut on price instead.
Everybody would see the arbitrage within the next trading cycle. So, rather than try to eliminate the arbitrage opportunity, put reaction times into a human time scale and let every actor share.
Tie breaks can be fair split with cryptographically random split for the last share, for example.
Yes, "all or nothing" would complicate, of it's still a necessity, but all those things are solvable, if we believe in fairness of markets and avoiding the next flash crash.
Random split exchanges have been tried and are equally gamed. If your random split is based on order count companies stuff orders to gain an advantage. If it is based on qty its a pro rata market which are common.
Having windowed auctions is also common. They are still latency sensitive as the entry/exit information is still subject to races.
Budesh has a good paper outlining batch auctions and their advantages but it still doesn't account for the fact that cross exchange arb is still an issue.
Atomic clocks? You can beat high-speed traders by only settling transactions, say, every five minutes. Whenever the time is nn:n5:00, look at the roster of orders and match bids with asks. Make it a complete black-box; no information is available about the current orders, only those from the previous five-minute period.
Atomic clocks have been in use in capital markets for a while. I was at National Physical Laboratory recently and they were demonstrating how they pipe their atomic clock output to the traders / exchanges.
could you explain why you think it should be illegal? what exactly do you mean by computer based? can i use a computer to help me decide what to trade? if the computer tells me what to trade and i press a button to agree with it and do the trade is that OK? sincerely trying to understand your perspective and how one might possibly implement it.
What I mean is that now computers algorithms are making the trades, it is so fucked up now that people pay hundreds of millions of dollars to be physically near the stock exchange so that they have less lag.
I should have been more clear, computer based autonomous or semi-autonomous algorithms should not be allowed to trade.
What I meant is that computers that do trades all by themselves should not be legal. Even frigging signing for a gmail account forces you to pass a captcha.
HFT systems scalp. They make their millions .001 at a time on front running and volume.
A way to discourage this is by adding a very small fee to each trade. This eats/takes away their profits.
The problem is their are too many folks making money that are connected to the right people in Government.There will always be talk about doing something about it but nothing will ever happen.
The only positive outcome from all this is that it created an arms race in the industry when it comes to the technologies used to facilitate HFT or fight it.
Now whether that technology trickles down to the rest of us through new and interesting things remains to be seen.
If you wanted to kill HFT, it's actually pretty easy. Just make the minimum increment incredibly small. In other words, let people trade at $.0000000001 increments. No fees needed.
The idea you suggest already exits for stocks in the US with very high prices because the minimum increment doesn't scale with stock price. For example, Google is $700 a share, making the minimum $0.01 price tick equivalent to 0.0014% vs. 0.08% for a $15 stock like Bank of America. HFTs generally don't even bother with stocks like Google because they aren't profitable enough. The rebate and value of making a tick goes down as price increases. This causes spreads to widen and these stocks are generally harder to trade for the buy-side. Not a good situation.
Says who? Insider trading is illegal, so are you saying it doesn't happen? I think your belief that the system works like it is defined in someones definition ignorant.
Just like to add Google HFT Firms Front Running and let me know if that is enough info to convince you.
The first result is an article explaining how HFT can't front run [1].
How about instead of making vague arguments about how HFT does front running, provide a specific example of how HFT actually does front running as you claim.
And no, one anecdote is not enough. You have to prove that there is a systemic fault in the way the HFT trades in the market that allow them to operate illegally at scale without any repercussions. To find one HFT firm guilty of front running and then claim that they all do would be tantamount to saying all hedge fund managers cook their books just because Bernie Madoff did. I'm sure you can see the absurdity of that accusation.
Front running requires that you trade in front of your customers. So end customer calls up his broker, asks for 5000 of XYZ. The broker then quickly goes and buys ahead of him, to get in on the action.
HFT firms don't have customers generally. So they can't front-run even if they wanted to. They'd have to make a deal with someone that does have customers. That's a pretty big conspiracy theory. And it isn't needed because HFT can make money just by seeing what orders hit the exchanges; no need to know about the orders before that.
When HFT was new, amateur forums that discussed it (like this one) were full of apologists saying that it didn't matter, it somehow didn't really effect the market. Some even claimed it made the market better.
The tone of the conversation, the framing assumptions, seem different now.
It still doesn't matter much (it helps with price discovery, as HFT participants provide smaller spreads than traditional market makers, which helps with liquidity, and the volatility they add in case they withdraw from the market doesn't seem problematic, because when they withdraw they do so in order to prevent trading in very non-understood regimes, so they don't trade in chaos, wow, blame them, and/or ban them, but you lose the much greater benefit too).
And yes, it's different, because when HN was new its comments were a lot more precise and fact-based, now it's full of dogmatic luddites.
Highly recommend reading Flash Boys [1] and it explains why time is so important and HFT firms. Great book for filling in the picture of what HFT is and I found it pretty entertaining too.
At risk of me-tooing: This book is great. It's a fun read watching him destroy Lewis's book.
The Flash Boys book is so obviously incorrect. Lewis writes stuff that, if taken seriously, would imply things like major traders not understanding basic price impact. Or major traders having trojans placed on their computers. One anecdote he mentions is something like "I entered an order but didn't press execute. The price changed!" ... As if there's some sort of conspiracy.
Lewis's main point seems to be that it's no fair you can't sell a million shares of something without moving the price.
It is disheartening to see someone get things so wrong, have glaring inconsistencies, and yet get wide acclaim. Just another instance of Gell-Mann amnesia effect. Probably many popular books are just flat-out wrong and we collectively just go on thinking incorrectly.
I'm about 90% sure that Lewis mentions it in his "acknowledgements" in Flash Boys, as a great resource to go more in depth (which is why I ended up reading it). You might have made that association.
This is marketing spiel.
If you just want to trade on a bunch of exchanges so no information flows between them, you can easily (TM) write a program that either a) lines up the orders at each exchange to execute at a specific time or b) delays the orders from a central server by the line delay.
So say NYC is 13ms from Chicago. You want to hit both at once. As long as you're not 13 ms late, nobody can see your order in one place and react at the other. You don't need an atomic clock for that, NTP will do just fine.
They're doing this because they have a reputation as a technologically advanced firm, and they know it will impress institutional investors, most of whom are still living in a time warp where spreadsheets are an advanced means of getting an edge over the market. They meet these guys, who are basically from another dimension of investing, and they suddenly need an explanation to their bosses of why RT can generate the most impressive returns of any strategy ever. The answer is "we have loads of PhD math geniuses building the strategies and amazing execution technology".