SEC Blames Computer Algorithm For 'Flash Crash' 218
Lucas123 writes "The US Securities and Exchange Commission and the Commodity Futures Trading Commission today issued an 87-page report (PDF) on the results of a months-long investigation into the May 6 'flash crash' that sent the Dow tumbling almost 1,000 points in a half hour. The Commissions are holding a single trading firm's automated trade execution platform responsible for the crash, saying it dumped 75,000 sell orders into the Chicago Mercantile Exchange over a period of minutes causing an already volatile market to come crashing down. The SEC has already enacted some quick rules to pause trading if a stock price should rise or fall by 10% in a five minute period, but the regulators said they expect the results of the investigation to prompt additional rules limiting the functions of automated computer trading systems."
Re:A time out is the right solution. (Score:3, Interesting)
One of the thing that was made clear to me over the last few years was that the price of stock is
whatever the last person bid for it.
It isn't based on the book value of the company.
If 99.9999% of the stockholders are not selling or buying- then the .00001% of remaining traders can walk the price wherever they want to walk it.
Re:What? (Score:2, Interesting)
This was one of those things that happened outside of the rules.
Re:Regulatory Agencies Don't... (Score:5, Interesting)
I agree that "prevent" could be considered wrong.
What I detect is that the smartest & most motivated people do NOT inhabit the regulatory agencies and indeed probably do not inhabit the exchanges themselves.
The various brokers hire people at much higher salaries &/or bonuses and pay them VERY well to find the tactics, some would say loopholes, to allow quick profits each day. That in itself is not what the original intent of share ownership markets were about.
I wonder when the word "Day Trader" was invented, but it certainly was quite awhile back but it didn't include the ability to do tens of thousands of trades out of one broker in a matter of seconds, and it certainly wasn't considered why we needed a share exchange in the first place. Exchanges were to allow companies to raise funds and to promote their value based on earnings and assets over time and allow a company to achieve an immortal status that an individual person could not achieve.
I think governments as the regulatory overseer are flawed, but then recognize the brokers are also very self-interested, so the whole mess needs more transparency.
That sort of transparency has been achieved with the likes of Linux.
I wonder if open sourcing the rules of the share markets could achieve the results where everyone knows the rules of the game & small individual investors have the same info that the large brokers do?
The worst thing in the world for a share market is to eliminate the small investor leaving only the whales to thrash about.
It is a big problem to solve and the self-interest of the big brokers cause all sorts of broken arms in WDC, if I guess right (meaning $s passed behind between arms).
Transparency is the only solution I see.
Re:Regulatory Agencies Don't... (Score:3, Interesting)
The worst thing in the world for a share market is to eliminate the small investor leaving only the whales to thrash about.
Frankly, 99.9% of all people who "invest" in the markets do not have sufficient training in the ways they can be screwed by people who know what they're doing, and are therefore not the sort of reasonable actors that would tend to create rational markets, but are instead cattle to be slaughtered by manipulation. The prices are bogus, nothing more than bait to lure them into the pen where their trading accounts are drained and the bolt stamps a hunk of their skull into their brain.
The best thing for the markets would be to require investors to be certified to put their money there.
But the people running the markets don't want the best thing for the markets, they want the best thing for themselves, and they can afford to buy enough votes in Congress to make sure it stays that way, at least until they make a mistake and show a little of what's behind the curtain, as they did here.
The Infinium Case Study (Score:3, Interesting)
Remind me, why do we have such a fragile system at the very core of modern civilisation?
Define 'core of civilization.' I don't view stock markets as that kind of thing. Regardless, I believe the reasoning they allow it is that -- like everything else in that crazy place of Wall Street -- it can help you make or lose money. This wasn't the only investigation where an algorithm screwed up. I submitted a story that wasn't accepted [slashdot.org] about an algorithm that lost one company a million dollars in five seconds.
So, you know, before you sign up to let a high frequency trader manage your trades, take note of the risks you are accepting [businessinsider.com]. In the story I reported, the company that lost the money just fired the guy who wrote the algorithm and keeps doing it.
If it's like margin trading where people were taking loans and lost it all and everything died because everyone was doing it, then it's bad. The question is whether or not these micro translations are going to suddenly force everyone all at once to realize their losses. I don't think that's the case but the 'flash crash' might be proof otherwise.
In defense of high frequency trading, I don't see it as anymore of a gamble than regular trading. You are shifting money around to make more money. So you shift tinier amounts faster and for shorter periods of time to get better returns. I'm not doing it so if it turns out to be bad for the people doing it then I'm going to benefit. If it turns out to be good for the people doing it then I bitch because I don't have that same benefit. If the HFTs are putting everyone at risk, I'd like to hear precisely how that logic follows because right now it's looking like it sporadically injects chance volatility that we've dealt with before.
Re:A time out is the right solution. (Score:5, Interesting)
But if a large organization wanted to sell stock to itself at increasingly higher or lower prices there isn't anything you can do to stop it. It's illegal as hell but hard to prove.
the only thing that makes prices rational is a fluid market.
A low volume market produces irrational prices and makes it easy to move prices around inside the limits of rational prices.
Put it this way...
Millions of baby boomers are locked in on a large chunk of their retirement money at 14,000 dow.
As they get older that price they are willing to accept to cash out is degrading (a lot of boomers would cash out immediately if the market got above 13,000 now).
As long as the price doesn't get too high or too low, the boomers are paralyzed and the market is not fluid.
In 2012 to 2016, that price will degrade more. I think we have a decade of overhead pressure from boomers cashing out. At some point, the price won't matter- they'll *have* to cash out to pay bills or go back to work (oh yea, you can't really find work if you are in your 60's these days- I mean 50's.)
Re:Because it works? (Score:5, Interesting)
I'm not sure that the proposed solutions will fix the problem. I'd much rather a degradation in response times as a function of orders (so the more orders there are, the slower the system gets) rather than a temporary hold on that stock. Temporary holds assume that software won't do what it has always done in the past - try again until it gets through. If you flood the system with retries from enough computers, the results won't change. It will merely have short gaps in it. If you have gradual degradation, then flooding will slow things way down until the flood stops. The negative feedback loop will guarantee that a crash becomes impossible.
In fact, that is something the market could do with more of - negative feedback loops. It should be possible to prevent market bubbles as well as market bursts, as a bubble is just a positive feedback loop in the opposite direction.
algorithms synchronized (Score:3, Interesting)
High Frequency Trading algorithms are most likely written by a very small number of people, who probably even know each other. The approaches to creating these algorithms should be very similar.
So it should not come as a surprise that given the same set of market data (news/some stocks going up/some going down/interest rates velocity/housing data/confidence/M1/Mx/etc.) the algorithms used by different HFT houses would respond in a similar fashion.
Imagine HFT House 1, 2, 3.
Now if one of them (1) noticed the market data at the same time *(and they saw that Japan was doing something funky with currency at that time) it started calculating probability of stocks going up or down and decided to play it safe (which at that time meant moving out of equities and commodities into cash), so it started to sell.
Now the other one (2) noticed the same thing about the market and noticed that (1) is selling, so it (2) upped the probability that stocks will go down and also decided to 'play safe' and started moving into dollars.
Same with the last one (3).
Now everybody is trying to sell at increased velocities. First they do their normal 5000 transactions/second post and cancel routine, but eventually they would actually stop canceling, prompting the rest of the market to start selling, triggering the automatic retail stops etc.
The HFT algorithms are really synchronized when it comes to overall data and they magnify the resulting movement by each others' actions.
BTW., you'll soon notice that bad news will no longer cause stock markets to go down, but instead they'll go up and so will commodities, that's because it is now recognized that bad news = more quantitative easing = more inflation = weaker dollar. So who wants to buy dollar on bad news, if bad news really means that the Fed will print more dollars? Same with bonds, buying bonds is stupid, they'll eventually figure it out. Buying bonds is like buying dollars to be received a number of years into the future. BUT if you don't want dollars that are inflating NOW, why would you want the inflated dollars in the FUTURE? Makes no sense, so bonds will also go down upon bad news eventually.
You can see these mini flash events caused by HFT in different market segments through the day, if one bank goes down in a very short time frame, then you can be almost certain that most of them went down by the same amount, and then they'll all come back to almost the original levels minus the retail auto stops that'll be eaten. Don't be a sucker, move your money to commodities or foreign equities.
Re:Valuation is an art (Score:3, Interesting)
a) no one had any idea who the winners would be
b) the game was sort of a tournment for firms so there were going to be many losers and only a few winners, and
c) most of the dot com companies issued very small portions of the company.
Re:A time out is the right solution. (Score:3, Interesting)
You didn't point out that the OP gave a terrible summary. 75,000 sell orders were not issued; a single 75,000 e-mini contract sell order was issued by a single trader to the firm's execution platform (that slices and piecewise sells, rather than issue a single large sell to reduce market impact). The firm's execution platform ultimately sold 35,000 e-mini contracts in 7 minutes before the firm shut it down. The program did what it was instructed to do -- the trader made a fat-finger error of selling $4.1 bn worth of e-mini futures (75,000 contracts).
If a driver presses the accelerator pedal while there's a brick wall in front of the car, you can't blame the engine or wheels for crashing through the wall. Can technology prevent this from happening? Sure, a car can have proximity sensors to prevent stupid users from crashing into a wall. Just because technology can fix the problem doesn't mean technology was the cause of the problem. It usually always is human error, and indeed in this case of fat-fingering in a sell order of 75,000 e-mini contracts.
Re:A time out is the right solution. (Score:3, Interesting)
You use a market order if you have no time to sit and wait for your limit to trigger and you're very motivated to buy or sell to get rid of a stock or to pick a stock up off the floor.
Market orders rule when it's 3:59:50 and you WANT to get that trade done so you don't get exposed to overnight trading and the morning futures market.