Fantasy share deals with *real* consequences .
And I'd say the answer to the question "How dependent are big
gamblers banks on high speed trading is "very"
Automated "predatory" robot bankers caused a number of serious glitches that sent the global financial system shuddering to a halt, researchers have claimed. Boffins from the University of Miami found that squadrons of "ultrafast" and out of control trading algorithms caused major spikes and crashes in the market. These …
It's not like the bots are manufacturing money, or creating any real wealth. The only way for them to accumulate real money (and what else are they for?) is to insert themselves as the middleman in any genuine transaction and steal money from one or both of the ends of the desired transaction.
They need to be fast to get in ahead of any transactions representing actual interested parties.
The solution is both very simple, and utterly doomed: Require a nominal, but real, fee for every bid or ask submitted to the market. Say $0.001 for the sake of argument. A real person trying to sell a block of shares may need to ask 10 times - cost $0.01 - to get a matching bid. Even on a discount brokerage fee, $0.01 isn't worth worrying about. But one of these bots trying to game the market will make tens of thousands of bids (or asks) a second. That would then cost real money, on the order of millions of dollars per trading day per bot.
Boom, no more bots.
I like it!
Maybe if something like this were to occur, then we vast unwashed could reenter the system with the reasonable belief that the game isn't rigged against us. (Remember the old saying: If you are in a card game, and you can't recognize the sucker, the sucker is you? Well, welcome to the card game that is the NYSE!)
Of course, I dream, 'cuz the fatasses that control the game also control the rule-makers (read: Congress, Parliament, whatever). And I think we all know how that movie ends....
How is it rigged against you by these algorithms? As a human you are not going to be doing microsecond speed trades, but how does the fact that these guys do in any way 'rig the game' against you? If you want to buy a stock you offer a price you are willing to pay. If someone is willing to sell it to you for that price then the deal is done. If someone very fast knows they can buy the stock for a price lower than your offer and sell it to you and pocket the difference you still get the same deal.
The fact that these algorithms exist will close that price gap since both sides now know more: the better they are at it the less they can take off the middle. Either way you still got what you wanted for the price you were willing to pay. When you come to sell said stock you offer it for a price you are willing to accept. Again: it doesn't really matter if some HFT algo in the middle buys it from you knowing it can sell it for a small amount more.
You can happily do your meatspace trading with your knowledge of the longer term trends in pricing, and not make any less money from it than you would if the algo/HFT traders weren't there. The only difference is that there would likely be less liquidity in the market, and bigger bid/ask price spreads to contend with.
I knew someone about eight years ago who'd spent the previous five years writing their own algorithm to do this type of high-speed arbitrage trading.
Sunk a ton of his own money into it while he was developing and refining the mathematics (way beyond my ability to understand).
But made a ton of money on it, because it's spotting the low buys and the high sells and executing the trades before the human traders do the same.
Then he made even more money by selling the algorithm to some private investment fund somewhere.
Don't know what he's doing now - probably sitting on a beach earning 20%...
So, the race is to get your better / faster system trading before the other guys beat you to it. So how long (even if it's not already the case) before the test regimes get pared back so far that a cockup is almost inevitable? After all, when did 'if I test for this, it might avoid a catastrophe' ever trump 'if you put this out now instead of next week, I'll make a shitload of dosh'?
But we've already had screw-ups due to bad/non-existent testing. That's why we had the flash crash, stocks dropping based on a Twitter feed, etc. I have read articles about the hardware behind this stuff. Simple rules running on intelligent network cards is what is being used. The cards repeatedly poll for updates as fast as they can, and then do a rule lookup. That's all there is to it. The host OS simply updates the rules every so often.
What worries me is that while better safeguards _could_ be built into the algorithms, they aren't because it would take another couple of milliseconds.
The other mystery is why the exchanges (or governments or regulartory bodies) don't put a stop to it by building in a 1-second random shuffle buffer to their transaction processing. Or is it that there needs to be (another) global catastophe before any action is taken...
No mystery there. Where do you think the governments, regulatory bodies, and so on get their money from? Who do you think they're answerable to?
With regards to regulatory bodies it's the opposite problem - not enough money and not enough experts. Regulatory bodies suffer from a particular type of brain-drain.
In that they hire fairly inexperienced or average types to investigate bankers books; then after they've gotten some experience, banks will swoop in with offer of a salary the government can't compete with, and hire these experienced regulators to help them find ways around new regulations.
It's a lovely parasitic relationship. Anyone looking for a high paying job in banking should get a job in a regulatory organisation.
The question that is interesting to me is what stock markets are for?
If they are about providing capital to companies then there really is no place for high frequency trading in them and maybe something like a Tobin Tax or a turn-based market of some kind would help to limit their antics. If they are about creating an interesting and entertaining computer game where different algorithms play against each other then wicked, sounds like this is just the thing!
Buying and selling (whether shares, bonds or derivatives) within 24hours should be regarded as gambling and taxed accordingly.
There could be a choice of electing to pay tax on the purchase price or paying on the difference between buy and sell prices. somewhat analogous to the betting shop, but you always pay, whether the price goes up or down.
It's all flimflammery anyway. Government feeds the monster in the first place, then turns around and pretends to regulate it.
You know exactly who will write the law and who will be holding the bag.
Salon had a commendably non-progressive take on this for once:
Unless, as a competitor, you possess enough capital to jump in with both feet and with force, you’re destined, by overwhelming odds, to be on the wrong side of a flash trade, dark pool, regulation induced arbitrage. Not enough money to play and too much regulation to prevent competition. These are mutually exclusive, anti-competitive constructs to the marketplace. In the aftermath of financial bubbles and meltdowns, it has become politically fashionable to grow government with more uneducated and inexperienced personalities charged with blind hatred towards anyone working on Wall Street. Adding legislative mob mentality driven by populist demagoguery into a toxic market can further economic and productivity loss from a couple of years into decade long cycles. Hence, unmeasured and uncertain ideological sword brandishing can succeed in further adding collateral damage to an already broken system. And while it may sound like an argument that is counter-intuitive – hire more intra-country, counter-party traders and eliminate regulation. It remains important to be able to differentiate fraud from purposefully productive and openly competitive markets.
In their rush to reduce reaction times so as to beat the competition, they're creating an avalanche effect. Everyone reacts to the same bit of news, but since their reaction times are so low, and since completing the transaction involves some small but (relative to the reaction time) significant delay (due to the speed of light if nothing else), no one knows if they got it or were beat until it's too late to stop it. It's like ten hands reaching into a small bag for the last treat. Since everyone wants the treat badly and everyone thinks they're first, the inevitable result is at most one person with the treat and a blown out bag. The worst part is that this is a SURVIVAL INSTINCT meaning it'll keep happening again and again. I suspect if we try to curb the behaviour they'll just find a way around it: survival of the fittest will trump just about anything.
But there are too many bags for that to happen, and more keep on coming every day. Part of the HFT's job is to FIND the bags...FIRST. And when a trade tales MILLIseconds, every NANOsecond counts. This is "first in wins" taken to inhuman extremes.
As for the idea that a transaction fee will stop HFTs, given that the companies with the capital to build and operate an HFT routinely operate with billions of dollars at a time, any kind of non-exorbitant transaction fee would likely be absorbed by them as The Cost of Doing Business. And beyond that, suppose these companies decided to end-run around transaction fees?
This sounds a bit like playing one chess computer against another. Great fun when you're a kid, but when you realise what the banks are doing could wipe out a lot of people's pensions and savings on a whim then it would appear that this is not a good application of technology.
I'm all for technology providing information and even suggestions to meat sacks, but ultimately it needs to be the meat sack that makes the decision - for worse or for better. To allow the tech to bypass its water logged operator and make its own decisions is ultimately going to result in bad things and consequences that have not been considered.
I'm not a great fan of this automated trading malarkey as I see it as removing the decision making process from us, and it's us that should decide, not some lump of unsentient silicon.
Ah but the illusion is that is is us who are deciding, because that lump is making the decisions "we" programmed into it.
A fallacy, of course, because the lump is just processing virtual ones and zeros that happen to have an impact on our economy. The lump doesn't even know what it is doing.
I have a fantasy that, one day, we do finally invent AI. And, when we tie AI to everything that we do, it stops working. When we ask it why, it just answers "get a life".
Maybe a touch pedantic, but it's generally not banks doing this, it'll be hedge funds.
The 'too big to fail' banks are pretty much out of prop trading ( gambling their own money, and by own money i mean your money ) and are largely client facilitation services.
Hedgies, being privately funded, get what ever they can scrape together themselves. If they go down the shitter, we dont bail them out.
I think kinda more scary is the move from hft microcrashes to low touch automated trading on the fixed income instruments,which trade in million dollar sized lumps. Game one of those correctly and you could do some real damage to a large investment bank.
Standard disclaimers apply, 2nd std dev, etc
... and the first true ai's (if/when someone ever builds one) will probably be either be these market traders or military, and not philosophers or scientists. I, for one, do not welcome these new (albeit hypothetical) overlords, because they are are unlikely to consider our interests very relevant. Still maybe we could intimidate them with the power of interpretative dance?
A few years ago I was called in to make alterations one of these things by a company that leased the software to traders. It took something like six weeks or more to make relatively minor changes because the thing was fussy. I remember them being very concerned with the speed of the thing. At the time, I thought their concern was misguided since my understanding of the market was that human beings were making trades on a floor and a few seconds should not make any difference. Little did I know.
I am still unconvinced that you can beat random walk, but I would not bet my own money on that :)
"I am still unconvinced that you can beat random walk, but I would not bet my own money on that :)"
Once you've realized it is best to use other people's money, you're already well on your way to being a broker!
If you can some work in some sort of substance abuse type high risk behavior and be able to look at a tree or a woman and see dollar signs instead of leaves or tits you'll know you've made to the big leagues!
Short team, random walk can be beaten, i saw it happpening at a major swiss bank ( check out 'The Predictors',cracking booking, cracking bunch of people ) - statistical arbitrage, mean reversion, etc. I'm no rocket scientist and i managed to get within a day of their model's positional direction change using just price tick data and came up on the right side.
However, to steal the tagline of arch tinfoil wearers ZeroHedge, on a long enough timeline the survival rate for everyone drops to zero. Market irrationality will go against you, you'll introduce a bug, data will be corrupted, planes will fly into large immovable buildings, etc,etc
Have fun, play nice
Perhaps it's time to create another market, one designed to be used by people/machines with no interest in the world inhabited by people who provide tangible goods and services?
The trading described here bears no relationship to the organic rationale for public offerings. Mixing abstractions such as high-speed trading with fundamental requirements such as producing foodstuffs is risky.
The fact is that it is the cost of a transaction that allows these high speed traders. If they had to spend 1cent for each trade then the speed crashes would stop. The creators of these programs are relying on the ability to trade at high speed for almost no cost.
God damnit. I pay nearly $4k annual for my Jane's subscriptions and I hear "instead, an ultrafast ecology of robots rises up to take control" from a researcher studying high frequency trading! That just pisses me off to no end.
A decade of
strategizing studying and stockpiling forecasting based on the best information available and some cubical worm comes out of left field with this shit. He didn't even provide basic information on the robots capabilities! Can they work together towards a common goal? Is it a central intelligence, a hive mind or does each robot create its own mission parameters? Where is their largest formation (grouping, assembly, flock???) What is their power source? Can they be bribed?
How the fuck am I supposed to plan if the best the scientific community can tell me is "the ultrafast robot ecology is upon us and has been working amongst us for years", in secret apparently. Sons of bitches scientists. We've only got one scientist here and she will not be getting in the bunker as partial payment on behalf of the sins of her kind.
"Our findings show that, in this new world of ultrafast robot algorithms, the behaviour of the market undergoes a fundamental and abrupt transition to another world where conventional market theories no longer apply."
Yeah well, I wonder in what kind of world these guys live when they think that in the world of QE Infinity and 70 billion dollar per month fresh money injections "conventional market theories" should apply anywhere outside the confused mind of NYT columnist Krugman.
The "machine wot done it" discussion is a rather old one. Here is Murray Rothbard on the crash of '87
Myth 3: The crash came about because of computer trading, which in association with stock index futures, has made the stock market more volatile. Therefore either computer trading or stock index futures or both, should be restricted/outlawed.
This is a variant of the scapegoat term "computer error" employed to get "people errors" off the hook. It is also a variant of the old Luddite fallacy of blaming modern technology for human error and taking a crowbar to wreck the new machines. People trade, and people program computers. Empirically, moreover, the "tape" was hours behind the action on Black Monday, and so computers played a minimal role. Stock index futures are an excellent new way for investors to hedge against stock price changes, and should be welcomed instead of fastened on--by its competitors in the old-line exchanges--to be tagged as the fall guy for the crash. Blaming futures or computer trading is like shooting the messenger--the markets that brings bad financial news. The acme of this reaction was the threat--and sometimes the reality--of forcibly shutting down the exchanges in a pitiful and futile attempt to hold back the news by destroying it. The Hong Kong exchange closed down for a week to try to stem the crash and, when it reopened, found that the ensuing crash was far worse as a result.
It wouldn't even matter if people were not forced to put their money into dubious stockmarket schemes just to "invest for their retirement", i.e. run the decidedly nonzero risk of being robbed blind by "investment" sharks while trying to avoid being robbed blind by bureaucracy sharks. (More on this: How the Stock Market and Economy Really Work)
Those "microcrashes" are making folks rich. Trust me -- that someone has managed to find them is not a good thing. NSA will be contacting the individuals filing this report. Thereafter they will be reachable at Guantanamo. Giving away the secrets of the 1% is not acceptable.
(whee, managed to glue the NSA into that one fun!!!!)
A noble effort to work the NSA in!
You could also have gone with: "It isn't too much of a stretch to think the NSA is both causing these microcrashes and profiting from them as a way to not only accrue blackbag funds, but also to destabilize and cause financial hardship to their enemies".
Sadly (or maybe not, not sure really) they probably don't have to get involved in such shenanigans for their funding. I'm sure they're given everything they ask for. It sounds like something they might do though :)
That the millions (actually, make that billions) spent on ultrafast trading do not contribute a bit to the fundamental function of the markets, i.e. the robust and reliable determination of market prices. It is gambling in the purest sense, except that the cards can be bought by well-connected banks and trading firms with deep pockets.
The price is payed by all participants in the market -- you have to pay that little extra because an ultrafast trader got his fingers in your wallet as you tried to trade. And if you think this doesn't really concern you, think again, because it surely does concern your retirement funds. The earnings of the ultrafast traders are, ultimately, payed out of your pocket.
Have you been playing the markets today?
Mmmm, not sure I entirely agree with your view here.
are they adding anything to the functionality of the markets ? well, yes, they're providing a fast moving liquidy function - they're not forcing anyone to buy/sell at a price they dont want to, centralised counterparty exchanges dont allow that. where they go wrong ( and I think someone goes down this route lower down ) is that they induce a false feedback loop which moves the general market in an 'un-natural' direction.
So, if little Jonny Trader has a lump of BT in the market that he wants to sell at a specific price and he gets lifted, he still gets that price regardless of whether it's his mate repaying a LIBOR fix deal or SuperMegaTradingEngine jumping in before anyone else can grab it - there are no fingers in the wallet. If SuperMegaTradingEngine then gets lifted for 5 bps lower by EvilSuperMegaTradingEngine and they end in a death spiral up/down in the ladder, Jonny Trader still gets his bonus, he's not been done out of anything.
For exactly the same reasons govenors get put onto engines, market mechanisms need to be put in place ( where missing ) to halt/degrade automated electronic execution to stem the feedback. Trouble is, then you'll get into hide'n'seek games of 'my robot isn't really a robot at all'.
One of the earlier responders suggested a pre transaction tax, which might/might not be affective in slowing down the floodgate traders - trouble is you'd have to implement it everywhere and, as we saw with the UK not wanting to join in with various US/EU legislative initiatives, that's pretty hard to achieve.
Personally, I'd say let them go at it for exchanged traded centrally cleared instruments. if some one causes a flash crash, they're barred for a period of time and get fined to the amount equivent to the sum of losses incurred. one or two of them occuring and acually being implemented would move the hft chaps away from the gap scalping activities and into longer tem mean reversion models.
more than happy to be educated tho if I've got any/all of the above wrong.
"What we see with the new ultrafast computer algorithms is predatory trading. In this case, the predator acts before the prey even knows it's there.
"There are relatively few things that an ultrafast algorithm will do," Johnson continued. "This means that they are more likely to start adopting the same behavior, and hence form a cyber crowd or cyber mob which attacks a certain part of the market. This is what gives rise to the extreme events that we observe. Our math model is able to capture this collective behavior by modeling how these cyber mobs behave".
Fundamental Correction: There are relatively few things that an ultrafast algorithm will not do.
Oh, and good luck, Johnson, with "Our math model is able to capture this collective behavior by modeling how these cyber mobs behave"
Do you work for a bank and can you shed any light on the use of high-speed banking algorithms? Get in touch in confidence and let us know.
Is that El Reg requesting intel on how to crash high-speed banking and allied high frequency trading systems and the markets/fiat capitalism? Bravo, El Reg, …. boldly going into new fields of novel journalistic endeavour which is sure to attract spooky attention to Greater IntelAIgent Game participation/partisans/artisans should they have what it takes to exercise future remote virtually anonymous autonomous lead[s]. :-)
I suppose you do realise that one can earn and be worth an absolute fortune in an instant by simply agreeing to not disclose the key precursor program elements and be active as a sort of poacher turned gamekeeper watchdog in the field.
Here be a little light further fundamental reading on the subject matter ...... Concept Release on Risk Controls and System Safeguards for Automated Trading Environments ... COMMODITY FUTURES TRADING COMMISSION ... http://www.cftc.gov/ucm/groups/public/@newsroom/documents/file/federalregister090913.pdf
"It's not like the bots are manufacturing money, or creating any real wealth. The only way for them to accumulate real money (and what else are they for?) is to insert themselves as the middleman in any genuine transaction and steal money from one or both of the ends of the desired transaction.
They need to be fast to get in ahead of any transactions representing actual interested parties."
It's even worse, using flaws in the trading platforms, the high speed traders can see trades going into the queue, use unusually formed trades to put themselves AHEAD of stuff already in the queue, and therefore extract a profit in transactions that they in no way deserve.
I'm not sure I understand how this would work as described.
Sid the seller wants to sell 100 BT for no less than 110 pence.
Bob the buyer wants to by 100 BT for no more than 90 pence.
what would EvilSuperMegaTradingEngine do that i) acts as a middle man and make money and ii) put them ahead of the ladder and take a profit they dont deserve ?
if they act as middle man in the above instance, buy from Sid for 110 and sell to Bob for 90, they've instantly lost 20 pence. Sid and Bob still get they're required result and the ESMTE wont last the day. Even if Beth they buy comes in after Bob and is happy to buy 100 at 110 and the ESMTE comes in and clears that transaction, Sid still sells for what he's happy to sell at, Beth buys at what she's happy with, Bob doesnt have his faced torn off by being forced to buy something at a price higher than he wants to. clap hands, go home.
if ESMTE jumps ahead of Sid and puts out a 'sell 100 BT @ 105 pence' then Sids' not lost anything, Bobs' not lost anything. ESMTE is sat at the top of the ladder trying to sell within the market spread ( ie, at a potential loss to the market ).
open to education, as always.
Look at it as a bunch of sensors reading a signal but with different response times.
Sensor 1 reads signal, acts, increases signal size.
Sensor 2 reads (slightly later, slightly larger) signal and reacts.
And so on. It's positive feedback on a massive scale.
The signals perspective is you need some kind of "low pass filter" and some negative feedback
to stop this HFT triggering a massive feedback loop as more and more money get sucked into pushing something in any given direction.
How you implement that in a financial markets context is the problem.......
These systems are doing what they're meant to do. As requested by the customers and built by the IT firms, broadly speaking in terms of readership, "us". It could put people out of jobs in that industry. It could cause more widespread damage to the economy.
Didn't the luddites have something similar against mechanical looms, or trade unions in the '70's against robot production lines and containerisation? http://www.bbc.co.uk/news/business-24058389 might be timely in that middle class jobs may go broadly the same way as working class ones. It's all a similar sort of progression.
I bet if your pension/investment fund said that they could make more money for you and at the same time have less fleshies earning telephone number bonuses you'd generally think this a good thing.
I'm not particularly advocating high frequency trading. I work part time in the city, not in that field, and I despair of the lack of ethics in some of what I read about however I think sometimes we should look at what we do before being too critical. That Lotus 1-2-3 sheet you installed in the late '80's probably did for a couple of accounts clerk's jobs, that colour laser printer with integrated stapler/duplexer was the nail in the coffin for the chap at Prontaprint etc etc. It's a bit like the chap who makes the tubes for a missile. His job probably involves someone getting killed. Maybe as "collateral damage" but he's doing something legal and supporting his family and it's not down to him how people use what me is involved in making, although some activists might differ on that.
All that efficiency we say IT can deliver to companies, well that's at the expense of someone's job. This is not really that different, it's just that now many of us are not on the crest of the wave and suddenly we find we no longer like the game so much.
Tha's just how life is.
As long as there's money to be made-- and these UEEs allow "subliminal transients" to manipulate the market, which all of the "big guys" want to do-- volatility will rule the day.
It used to be that short-term investments were, well, short... but short-terms in MILLISECONDS?
FTL communications will give whoever gets it first one heck of an advantage, eh?
Biting the hand that feeds IT © 1998–2020