This page has been archived and commenting is disabled.
Latency Arbitrage: The Real Power Behind Predatory High Frequency Trading
A new white paper out of Themis Trading analyses the impact of predatory algos comprising various HFT strategies. In Themis' view: predatory HFT generates $1.5-$3 billion in profit. Themis concludes with the following three market integrity questions:
1. Instead of belittling the impact of latency arbitrage, and representing it as a gloriously naive $0.01-$0.02, does the regulators' thinking change if that impact is as high as $3 billion a year?
2. In a quid-pro-quo worlds, are market centers merely charging HFTs a higher fee in exchange for an advance look at the NBBO? Market centers should be protecting all participants equally.
3. The most critical question: "When a market center provides an HFT the ability to out-maneuver institutional orders, is not the exchange putting institutions and their brokers in breach of their fiduciary responsibilities, especially those institutions managing pension funds governed by ERISA?"
- 8115 reads
- Printer-friendly version
- Send to friend
- advertisements -


"It's that a differential exists at all."
"1. Instead of belittling the impact of latency arbitrage, and representing it as a gloriously naive $0.01-$0.02, does the regulators' thinking change if that impact is as high as $3 billion a year?"
Pure marketing techniques. Make the item appear smaller than it is by breaking it into it smallest components. Who wants to quibble over a few pennies we wouldn't even stop to pick up. Nothing here to see, move along.
"2. In a quid-pro-quo worlds, are market centers merely charging HFTs a higher fee in exchange for an advance look at the NBBO? Market centers should be protecting all participants equally."
In the mafia, it's called the VIG and the customer always pays it. Youse got ta pay if youse want ta play.
"3. The most critical question: "When a market center provides an HFT the ability to out-maneuver institutional orders, is not the exchange putting institutions and their brokers in breach of their fiduciary responsibilities, especially those institutions managing pension funds governed by ERISA?"
Ethical considerations? You've got to be kidding me, right? Ethics are for losers.
I though truncating fractions of a cent in banking... instead of rounding was illegal.
Would a one penny per share tax help reduce these shenanegans?
"Yes," to all the above.
"The vast majority of buyside firms surveyed by research firm TABB Group--84 percent--said they wanted regulators to "take no action" to constrain about high-frequency trading, according to Laurie Berke, a principal at TABB and the author of a new study about institutional trading. The study examined a variety of trading practices in the context of a regulatory environment that is changing."
http://www.tradersmagazine.com/news/buyside-technology-high-frequency-ta...
Question number three is HUGE, and is a great new angle.
You see, now Fiduciaries on pension/401K committees as well as the investment managers could be open to participant CLASS ACTION LAWSUITS.
This will be huge if skilled investment class action attorneys start mass lawsuits.
How about everybody who has traded stocks in the last three years joining that class action?
Good luck proving breach of fiduciary duties here.
That tabb study said that 55% of survey respondents were not concerned with HFT, and that study was cited as proof positive that there is no issue with HFT.
I never thought that studies could be misconstrued so badly. So if 45% are concerned it is no biggie LOL.
Ahh, question ... can we please call HFT what it rally is: front-running. And prosecute it. Now and to the full extent of the law. And take all these fraudulent profits away again. How about it?
HFT is not front running. There are many high frequency strategies that are as legitimate as the low frequency strategies and are just mimicking what a trader would do given the information amassed and what that information has yielded. Examples of this are rolling correlations which could change as often as every 5 minutes or less.
What is front running is a select few strategies, mostly in the latency arbitrage genre that get's to look ahead at the NBBO before anyone else, or see an order triggered and get in front of it before anyone else.
Not all HFT is front running but HFT that is latency dependent is. If your only edge is to act on information faster than others than you are front running. It doesn't matter who you get in front of, any dupe will do.
hear, hear, quant-this.
good luck being heard though. public wants blood and they'll get it; even if crossing that rubicon brings the whole deck of cards down, right upon their own heads (201 f's ~~ 1/2 of a 401 k).
If they claim the 1 or 2 cents per order is not consequential for HFT then a 1 or 2 cents per order Tobin Tax should not be consequential.
Funny how it is ok for HFT but not for a Tobin Tax.
thank you....the markets "surviving" HFT front-running skimming, albeit most participants are poorer for it, this is proof positive that a Tobin Tax, at least, would not be anymore harmful than HFT. However, a tax would be transparent, equally applied to all, and at least some of the tax collected would trickle down to do good things for regular people, where as HFT just concentrates skimming from much regular people money like 401k, pension funds into the hands of a few people that contribute almost nil to economy.
What bothers me is Timmy G saying traders would get around a Tobin Tax easily....so what does the imply? If you have to trade on regulated markets, how do you hide your transactions? Either he is lying, or if he is truthful, markets are more corrupt, corruptable than I imagined.
moneymutt, the traders wouldn't hide their transactions. The transactions would simply go overseas. e.g. Dubai has a futures market for oil, but nobody trades on it now. If you tax the NYMEX .25% per transactions, they'll simply trade on the Dubai exchange.
Think you can legally tie US companies to a US exchange? Maybe so. But there is no reason why someone couldn't trade single-stock futures based on US companies on a tax-free foreign exchange.
Plain and simple, Tobin tax does not work unless all countries implement it. And you really think Barbados cares about the G8 nations' desire for a Tobin tax?
So I guess Joe and Sal never paid for a Bloomberg terminal?
They must never have used Level-2 quotes.
Come to think of it, if they believed everyone had the same access to all quotes at the same time, they must be using the free 20-minute delayed SIAC quotes on Yahoo Finance.
There has always been a technological advantage by some market participants. When the railroad was invented, information was moved to some in Manchester from the London exchange before others got it.
When the telegraph was invented, the same thing happened (and thus began Reuters).
The ticker tape and the telephone were no different, nor were earlier direct connections to the exchanges... and so too is the situation today.
Sal and Joe surely miss the volume that they used to move as more and more trades are electronically placed, and I do imagine that they have quite a bit of angst towards the computers that are taking away their business... The next generation will view human traders with as much amusement as an 8-track tape.
It was a hoot reading that some institutions using VWAP strategies "must" make trades, as if the institutions didn't have bounds set on price. Quite amusing.
FYI, a $25K account will get one access to all major ATS feeds about 2ms behind what you'd get with a co-location feed - and ahead of the SIAC feed.
First, thanks to Joe & Sal of Themis for such a quality White Paper that is not only a great synthesis of the topic du jour but also written lay such that joe the plumber can 'understand' the basic frame of HFT itself.
absolutely love reading your comments here on ZH, peter(2) ... and KidDynamite is busting up crappy traders like the TICK's own dynamite triangle today during the rolling 14:20ish "predatory" algo starting line before the finish line that is the moc. dunno why Themis hates HFT so much? oh yeah, bc it is more efficient and quantifiable with a much greater R multiple / performance summary / etc. etc. than damn near any human ... which threatens the very existence of their very line of work. i mean, maybe, it could have something to do with it. having said that (great f'ing Curb...): they make several great points and do so from a point of authority through great lay writing, so ya can't really fault 'em for essentially talkin their book when they speak so very well about the idiosyncratic machinations of Mr. Market and its big boyz.
thanks for sharing your thoughts peter(2) and KidDynamite; especially since you each do, actually, have a clue as to what you're speaking of ... distinctly unlike so many pension pushers who think: 'damn HFT, damn brokers, damn goldman, blah blah blah'; without any clue as to what the TICK is let alone the differentiation between a Pips-Hunter and a VWAP Predator w/ staggered bi-directional fill-or-kill child legs attached.
Retail thinks that getting their way will help them or their dead mufu money ... lol ... it is just heeelarious to think what a "multiple" (haha) would be for any market NOT girded by an HFT structure. does a 4 handle sound good as a conversation starter for the black-eyed S&P 500 ??
does HFT suck for retail, sure, kinda, but so does sunburn; f'ing deal with it and move on you whiny children who act like you're mad at one dude who just called you an idiot after some other guy stole your wallet, while yet another used your own car to run over your dog.
psssttt, retail:
stop trying to be jim simons; stop thinking that you can replicate f'ing PrizmaL in a fortnight; and for the love of whatever stop getting all uppity about an issue that, while certainly, certainly is a very valid one ... is just, welp, not in the remote vicinity of what actually: 1) matters to your investments; 2) matters to your daily life; and, 3) which, in the grand scheme of all things distinctly financial, just might turn out to be one of the most ass-backwards Pyrrhic victories for the "common man." you wanna see a real tragedy of the commons ? just keep pushing to impede if not outright outlaw HFT as a whole and see what happens when Asia opens at 20:00 that Sunday night; you thought last September - October was fun, hahaha ... either way, you ain't seen nothing yet. keep fussin about HFT like you did evil "short-sellers" and see who's left to buy your pos individual issues from ya once GETCO is run into rocky shores. almost as funny as those Mo Rons calling for a return to some pre-Basel 'gold standard' in the voice of Williams Jennings Bryan w/o having any clue as to what he was actually calling for. such, such misguided angst. just amazing; even though it really shouldn't be since nothing, ever, changes but technology and aesthetics. human nature is; it just is. period.
imho: you retard HFT and you permanently brain-damage modern equity markets.
it does not matter how you feel or what you think about the issue du jour of topicality ... the bottom line is: yeah, HFT does suck a wee bit for retail; frickin deal with it since it is just about the only thing holding this absolute JOKE of an equity market structure in place. puhlease.
i am sure that i'm far from the only semi-educated trader / technician / quant who has been just so disgusted by / with the absolutely shameful display of ES n SPY market maker grand-daddy algos during key interval periods of time such as those surrounding the rolling european close, the 14:18 - 14:28 rolling predatory algo starting line and the finish line of the cash moc, especially that of the 3:38 old NYSE Specialist 'free-look' which no longer exists but, again, the idiosyncratic machinations of Mr. Market remain. but i digress.
"end HFT."
'and where do you work sir ? oh ok, that's nice. thanks for sharing and good luck.'
the public does not want the truth or real analysis; they want realmoney and a mainstream financial news media filled with dennis kneale's and charlie gasparino's who intone like howling monkeys and spend at least 3 SD's of their time worrying about how they are going to shriek something instead of what they are actually going to shriek about.
Anyway, thanks for helping to be a voice of actual reason / knowledge, peter(2) and KidDynamite; especially amidst the phantom feed / print that is retail's opinion .... a phantom feed / print may initially appear to be correct, even if it does smell a bit specious, but later on is proven to be wholly incorrect / entirely specious and, ultimately, is erased from the data series itself ~~ collective memory.
and if ya wanna pick on me for calling out retail, irrespective of blatant honesty (although this is all just my personal opinion) and intent of design ... or ya just wanna toss a basic pot-shot at someone who actually does trade, which pisses you off, bc you don't / can't ... then please do it within the link below:
Four Basic Qualities of Great Technical Indicators & The "Stochastics Default Club"
And on a related petulant note: please note how many comments were left on that rather long-winded rant-comment-turned-article; then contrast it with how vociferous all the gold bugs / technical analysis haters (read: can't read/ understand TA so it must be bogus) have been on just about every single one of my & our articles here on ZH.
i'm trying (admittedly, not well) to be a little more mature about laughing; at least publicly. honestly: just hate to see so many folks work all of their lives to "save" money to "invest", then "invest" like total fools in the explicit hope of short-term greed combined with the asinine regurgitation of / belief in lazy anal-ysts / crappy advisors who have but only a series 7 / 66 / 86 / 87 license and a few phone scripts in front of them.
as my godfather and hugh 'the silver fox' johnson used to tell me when i was their drycleaning bitch-boy:
look: all i'm saying is that if you're trying to create prosperity for your progeny then wake up and smell the tulips. and if none of this makes any sense and just sounds like a bunch of "Stamford-ese" ... then just chill in cash n cash equivalents; as Marla wrote earlier today and Prechter has written, rather presciently, since well before the 2002 original release of CTC, cash IS king and it will only be growing a lot more powerful over the coming years (IMHO). in any case, have a good night and a great session tonight / manana. arrogant lil shit out.
So, you need to trade on information not available to other market participants in order to save the market, keep 401k's afloat and maintain liquidity. hahaha
The markets have been around much longer that the HFT boys and girls and can function perfectly well without you but I do thank you for your concern. :)
No.
Mr. Market NEEDS HFT to buttress its irreparably fractured internal structure across innumerable machinations / functions. lay ~ there are both good and bad aspects of HFT.
bad: annoying to VWAP running prime brokers et al. and harms predatory institutions in and of themselves who are simply not as bright, creative or studious as their senior brethren (loosely termed the Hedgehog landscape).
good: the S&P hasn't crashed as much, yet, as the inflation adjusted REAL DOW (DJIA), where the central us equity indices are down c. 80% since the collective top in 2000. data-hound latency / prop info algo clowns, which you refer to JK, are not only a small portion of the aggregate HFT pie (outside of SIGMA X and similar SLPs) but are also on the lower end of the respective totem pole without much trading talent or technical expertise (they're pure programmers). you kill HFT on the whole and ya ain't gonna like what happens to your crapple, gurgle n goldilocks let alone momo driven nuclear footballs like dryshits or chiagra hot potatoes (bidu, solar spec). have a good day and a great session all.
Yeah, I believe it's the exploitation of latency advantages that are being addressed in the post. Anyway, I do agree that you don't need much talent to front-run, whether you have info a half hour or a half second ahead of everyone else, it's just theft, no talent required.
I really think it's hard for people to comment on HFT when they don't understand market structures (not directed at the two of you, just adding my .02). HFT is actually makes short term moves more efficient. If you run market efficiency tests, after 10 minutes the market becomes much more efficient than the market at 1 or 5 minute intervals so these quick 1 minute strategies are adding to a more efficient market.
This is not to be confused with MOMO algos which will pounce on a move and make it more dramatic, yet this is still a market being efficient because in theory efficiency happens instantly, so for example if the market was 100% efficient (strong - which I do not believe in because I'm a huge behavioral finance fan) then a move would be severe and instantaneous as it often now is.
What is not fair is that certain players get to look at order flow that no one else does at no cost. The specialist is able to do this but is made to make markets and truly provide liquidity when no one else will. There is a huge amount of predictive power in looking at the order books and I think those privy to that information should have to pay a price. If not then the information should be made available to everyone much like tick data is now. Add those with access to dark pools and all of a sudden you have a clear picture of where the market is going in the next few minutes. Now, information asymmetry has always plagued Wall Street and always will but at some point regulators need to make the asymmetry less skewed so more players can get in the game and efficiency can continue to evolve.
No problem with that, speed is not the issue, nor is momo, if a trader wants to go faster and harder that is up to him/her, the issue is trading ahead of market data that no one else has. I'm not sure why the HFT defenders don't do a better job of delineating the black hats and white hats. It seems to me that if you don't deal with the "sub-set" of traders who are dirty you run the risk of being taxed or regulated out of business altogether, you all go down with the ship. And all will cheer (maybe not rightfully so).
Smart people sometimes make the deadly error of thinking everyone else is stupid. Truth be told most of these algos are not that complex, if you can trouble shoot an air conditioning failure you probably have enough smarts to understand and write succesful trading algos. This whole air of superiority and the notion that HFT players are innovators and the smartest guys in the room is a crock of shit.
Yes, I am always amazed by how simple the algos are. VWAP for example has been a low frequency for ever (I love btw how the books take something simple like ARMA or MA and just make it seem much more complicated than it is). The MOMO's are the simplest. I remember back in the day when my old moss had a print out of up day probabilities on the SPUZ and a few other futures contracts. MOMO's now are not so different, if you have been up 2 days in a row with a close higher than yesterday with increasing volume and a 30% increase in articles in the news feed and the open is indicated up, what is the probability that you will get at least a .10 pop.
Some of the predatory flash (now banned) algos though were pretty sick. From what I understand so are what the SLP's are using. Not so much about complexity though, more about processing power. Imagine knowing what all 500 books on the SP500 looked like along with the options and index futures with rolling correlations (non linear or chaotic) giving you up to the minute equilibrium relationships. People are doing this and making a ton of money doing it. At some point more and more players enter the market and it becomes more efficient.
huh? Nothing has been banned, the SEC hasn't even gotten to the comment phase.
I thought some ECN's cancelled the flash orders.
> What is not fair is that certain players get to look at order flow that no one else does at no cost
Everyone who knows how to process the full data streams likely already has access to it either directly or via an intermediary.
The costs are quite high if you want to co-locate, but if you are willing to deal with aggregated feeds (which still have every bid/ask from all of the trading venues that matter) that are about 2ms behind the fastest co-lo feeds, you can get that cheaply.
But... you then have to be able to write software to process upwards of ~500K messages per second (just for US equities at peak - not even touching options) - as well as the end to end bandwidth to get it all (this isn't going to work on FIOS or a cable modem).
There is no technical way to deliver that much data on a broad scale without the use of leased lines and having your server sitting in either an exchange's co-lo space, or the co-lo space of a third party (broker or just well connected data center) that in turn is directly connected to the exchanges.
Most people get SIAC aggregated NBBO because that is all that they can technically process.
I was mostly referring to SLP's which have a look at order flow that others aren't privy to. Well aware of what is out there for HFT strategies, companies like StreamBase are making it more accessible every day. I am not against HFT, I'm just against companies being treated like market makers and liquidity providers without the obligation - that is what I meant by no cost.
this is written much better than previous tripe from Themis, but there are a few issues:
- not the core issue, but "predatory trading" is redundant. Trading is predatory.
- in Themis's diagrammed point by point example on page 2, they write: "2. Due to Latency Arbitrage, an HFT computer knows that there is an order that in a moment will move the
NBBO quote higher, to $25.54 bid /offered at $25.56."
It's tragic, manipulative, and incorrect that Themis still attempts to paint this as frontrunning. The LatencyArbHFT computer does not see your buy order and trade ahead of it. What MAY happen is that the HFT algo is offering stock, and when they get lifted, they go and lift offers on other market centers. Again, this is nothing new, and is a basic, old, trading strategy - they just do it faster. Similarly, HFT Latency Arb algo's may see publicly available trade data before JoeSixPack, but there is nothing immoral or illegal about that
most importantly: Themis writes "6. Because it is following a volume driven formula, the institutional algo is forced to buy available shares from the HFT at $25.55 or $25.56."
no no no no no.. NO ONE is forced to buy shares from anyone else. If the institutional algo thinks the price is to high, it can sell stock instead of buying it. that's the beauty of markets. NO ONE forces you to trade at a price you don't want to. If the institutional algo is leaving a footprint that lets everyone see what it's doing, that's it's own fault. Again, this is especially ironic because Themis's Saluzzi has previously admitted that he's a "tape reader" which tries to do EXACTLY that: forecast stock movements based on trading volume patterns.
haha..so HFT is forecasting? 90% of HFT transactions simply cancel, it's an always on, always win game. You get in front of somebody, it doesn't matter who, you make money. That isn't forecasting.
Yet another worthless white paper from the geniuses over at Themis. These guys are a total joke. Let's just look at their example on page 2:
"The Computer knows that there is an order" No it doesn't. "The institutional algo is forced to buy available shares from the HFT at," no one is ever forced to buy anything and if its an intelligent algo it will wait.
The fundamental problem is that NBBO is too slow. The solution is not to ban HFT, the solution is to fix NBBO so that it is not god awful slow. In this case, even the most ardent attacker of HFT has to admit that even if you disagree with the algorithms -- the cause of the problem is the crappy NBBO/trade tape.
hi JohnKing, i didn't use the term "forceasting," but yes - in a sense, these algos, per Saluzzi's admission, are forecasting demand from institutional VWAP algos because the VWAP algos trade dumb, on a certain pattern. This (forecasting) is exactly what all professional traders try to do: forecast supply and demand. HFT guys just do it faster.
My point was that these latency arb algo's are NOT seeing JohnKing's buy order and frontrunning it. No matter how much people want to believe that, it's simply not happening. What they ARE doing, thought, is seeing JohnKing lift an offer on ARCA, and, reacting to that data, lifting offers on NYSE or BATS. There's nothing illegal or immoral about that.
You don't know what these algos are doing.
I've seen some strategies that could be considered front running by the mere fact that once an order is triggered it get's there before you. Regardless, I tend to agree with you that 95% of the HFT strategies are just replicating strategies used by humans, they just do it faster and without emotion. I'm a low frequency guy but many of the strategies I use could easily be used on tick data and I would not consider any of them predatory.
There is a subset of HFT shops that are doing some sketchy things with orders and execution and I know this as a fact as I have been privy to some of their strategies.
The real problem with HFT is it exagerates moves to the upside and the downside. Everyone assumes it is a trader trying to buy or sell 100 shares. When you have thousands of traders trying to buy shares you exasperate the move up or down not by pennies but Dollars. So, every time one of thoes 1,000 traders places a trade the computer front runs that trade, do that times 1000 and the price substancially moves up or down. Enough Traders are trying to Buy and the price can move Dollars not pennies.
Traders have liked the HFT because Stocks have gone up on the FED stimulis and intervention in the Market and it has caused stocks to skyrocket because of HFT front running.
What Traders fail to see is that the HFT front running also works on the downside. When 1000 traders want to Sell it front runs thoes Sell orders. Which excellerates the downside in Stocks.
HFT offers nothing for the market. It does not provide liquidity because when the market is going down it front runs the Sellers. It is not there to buy into weakness like the Floor Trades used to. So, you have excellerated moves to the upside and to the downside.
Not only is it not fair it makes the Market more Volitle. HFT does not all allow for the Market to determine the actual fair market value of a Stock, just the price based on how many Traders want to Buy or Sell that particular day.
Everyone who posted above used super-fast posting techniques and proprietary algos to gain an (unfair?) advantage when commenting on this blog post.
Your quickness, diligence, technology, location, and experience allowed all of you to capitalize on my inability quickly to respond to market conditions.
I consider it predatory.
You had no idea what I was going to buy... er... say, but this phenomenon somehow makes my comment less valuable.
Simplest way to correct U.S. market problems is to set a buy-sell spread at 5% of any equities value. This will mark a return to the true purpose of markets: to connect long-term investors with companies seeking capital. If a person is unable to take a longterm position, they can invest via mutual funds where they would not take a 5% hit if they must pull money out the next day.
American industry does not exist to support the Wall Street moneychangers. The moneychangers exist to support American industry.
For those who will say, "Oh my God, the investment firms will leave!" I respond: Yeah? Where to? London wants more regulation than the U.S. Dubai gambled as much as the U.S. and is seeking bailout. Wall Street, listen up: gambling with dad's (Uncle Sam's, really) credit card adds nothing to our society. No one wants to hear your BS any longer.
Its not the algos, its the unequal access that is the problem. Spevialists on the NYSE at least had the fiction of providing liquidity when they front ran ... u can't say that flash orders and 200msec executions are providing liquidity more than baiting liquidity.
"Fiduciary responsibilities"
That's funny. On Wall Street, right? Just too funny.
A year and a half ago. Page 6, first paragraph.
Some light reading from 2008...
http://www.densitydynamics.com/uploads/File/White%20Papers/V06-007_Value...
Themis trading - these guys are a total joke. They are just trying to get some business by acting like experts on the big bad HFT, therefore, institutions will HAVE to execute through them, because only they navigate the dangerous waters.
I've read all of their white papers, and it's painfully clear that they have no idea what they are talking about. That example is ridiculous. If the institution wanted to get some size done at $25.54, it could just lift the offer at $25.54 - there would be no way to respond to that order before the order hits. Instituions usually won't however, because they will on average get a better execution price by placing a bid at $25.53. There are more things that are completely silly about this (and all the other Themis white papers), but I've already wasted too much time on this POS.
Here are some relevant facts on the impact of HFT. According to Elkins McSherry, which specializes in trade cost analysis for institutions and publishes an annual survey in Institutional Investor found that transaction costs (commissions, fees, and market impact) have plummeted in the HFT era - over the last 8-10 years. Overall, they have dropped 35% and dropped the most in markets that have the highest HFT prevalence (50-60% in us equities) and the least in markets that have the least HFT prevalence (~10% in asian equities, where HFT only represents 5-10% of volume).
Hi Chris :)
Strike one.