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?
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)
If you're hitting multiple exchanges, then you can take milliseconds and still be fine.
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.
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.)
That sort of sounds like DDOS to me. They patented a DDOS botnet.
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".
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.
Say you send out $100mil orders broken down into 100 $1mil orders, they are going to take some time to execute completely.
Won't the HFT's detect them long before they are executed completely ?
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.
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.
the answer is 'we have loads of PhD math geniuses building the strategies and amazing execution technology'
that the institutional DD team is saying this to the investment committee or there is actually another entirely different reason?
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).
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.
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.
[1] https://en.wikipedia.org/wiki/Network_Time_Protocol
[2] http://stackoverflow.com/questions/17256040/how-fast-is-stat...
With GPS receivers or very good network connectivity to your ntp server (ie. not home dsl) you can easily get down to below 1ms.
"On a local area network, [PTP] achieves clock accuracy in the sub-microsecond range"
https://en.wikipedia.org/wiki/Precision_Time_Protocol
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).
I thought RenTech didn't have investors anymore.
Those types of hft firms are surely eating into Renaissance's profits in a big way.
What you are describing isn't front running (or even illegal).
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.
Admittedly I have zero special knowledge of Renaissance and what particular strategies they use.
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).
[1] http://www.theverge.com/2012/11/26/3692392/google-spanner-at...
http://research.microsoft.com/en-us/um/people/lamport/pubs/t...
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.
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.
Funnily, that works only if the two observers are in rest relative to each other, as relativity theory tells us. :-)
For instance, at the VLBA: https://en.wikipedia.org/wiki/Very_Long_Baseline_Array
I should have been more clear, computer based autonomous or semi-autonomous algorithms should not be allowed to trade.
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.
As another person commented elsewhere, front running is a specific term with a clear definition and is illegal. HFT firms do not do front running.
[1] https://www.amazon.ca/Flash-Boys-Michael-Lewis/dp/0393244660
[1] https://www.amazon.com/Flash-Boys-Insiders-Perspective-High-...
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.
("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.)
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!
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.
See https://www.chrisstucchio.com/blog/2012/hft_whats_broken.htm...
You'll probably be interested in reading Eric Budish.
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.
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.
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.
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.
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.
Despite the hilarity of such a thing, is this a sign that HFT is not nearly as profitable as it once was?
That said I have no idea why they would patent this. I can't see how they can make money from it.
Jeezus. That is about the same OOM as Moore's Law.
for comparison, according to one source, over 50 years, Berkshire Hathaway grew at about 21 - 22%
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.
The tone of the conversation, the framing assumptions, seem different now.
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.
https://www.chrisstucchio.com/blog/2012/hft_apology.html
https://www.chrisstucchio.com/blog/2012/hft_apology2.html
https://www.chrisstucchio.com/blog/2014/how_to_not_get_rippe...
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.
http://www.npl.co.uk/commercial-services/products-and-servic...
Case in point: gang switch in the era of voice brokers. Nothing says 'synchronized' execution better which is all this application is.
