Thursday, August 02, 2012

$440M in 45 minutes

Does this qualify as the most expensive software bug of all time? Raises concerns about our future as passengers in driverless vehicles ;-)

I suppose it's a positive that Knight had to recognize the losses immediately, instead of sweeping them under the rug by adjusting a parameter in a risk model (see, e.g., JP Morgan whale + a million other recent examples). Would Knight have lost even more money if the exchange hadn't shut down trading in the affected names?
NYTimes: $10 million a minute. 
That’s about how much the trading problem that set off turmoil on the stock market on Wednesday morning is already costing the trading firm. 
The Knight Capital Group announced on Thursday that it lost $440 million when it sold all the stocks it accidentally bought Wednesday morning because a computer glitch. ... 
The problem on Wednesday led the firm’s computers to rapidly buy and sell millions of shares in over a hundred stocks for about 45 minutes after the markets opened. Those trades pushed the value of many stocks up, and the company’s losses appear to have occurred when it had to sell the overvalued shares back into the market at a lower price.   
The company said the problems happened because of new trading software that had been installed. The event was the latest to draw attention to the potentially destabilizing affect of the computerized trading that has increasingly dominated the nation’s stock markets.
This says it all. Previous posts on high frequency trading.

Update: My representative is on the job!
NYTimes: ... Some critics of the current market structure have said that much bolder reform is needed. One change that has been contemplated is a financial transaction tax, which would force firms to pay a small levy on each trade. At the right level, this could pare back high-frequency trading without undermining other types, supporters say. 
“It would benefit investors because there would be less volatility in the market,” said Representative Peter DeFazio, a Democrat of Oregon. He introduced a bill containing a financial transaction tax last year. 
Opponents of such a levy say that it could hurt the markets and even make it more expensive for companies to raise capital. 
“I would be very concerned about unintended consequences,” said Mr. Sauter. 
But Representative DeFazio, who favors a levy of three-hundredths of a percentage point on each trade, says he thinks the benefits of high-frequency trading are overstated. “Some people say it’s necessary for liquidity, but somehow we built the strongest industrial nation on earth without algorithmic trading,” he said.


  1. Ken Condon12:55 PM

    It’s been said that the “little guy” has fled the stock market and is unlikely to return. I don’t know how that can be quantified but it is obvious that the market is indeed dominated by powerful hard and software. Or as one article I read which stated “the five robots that run the entire market system”.

    When it goes nuts this is what happens. Intrinsic value of a corporation seems to be lost in this process so no wonder the little guy has fled. Headline and computer driven stock values make for a volitle ride.

  2. Ken Condon1:48 PM

    More on this topic:

    While they may mitigate the likelihood of an individual stock flash crash, they do little to prevent another market wide Flash Crash. They treat the symptoms of a Flash Crash, rather than the underlying causes. Despite the Committee’s testimony from numerous voices with academic and industry experience cautioning on the dangers of unfettered speed in our markets, there is virtually nothing in the 14 recommendations to address that issue.

    Nothing much has changed since this report was issued from what I can tell. Perhaps if LY is still lurking he might have some insights he can lend.

  3. I wondered the other day why you need real-live-traders nowadays. A competent analyst and a computer should smoke a 'discretionary' trader, I think. This example aside. This result says to me that KCG doesn't understand their business model. Some report said that it was 70% of the value of the company. If you believe the job ads that want people to write software with ~1ms response times, you are talking about more than 2.5M trade opportunities in this loss window. Who makes 2.5M decisions without having some state-of-health metaware?

  4. Ken Condon5:46 PM

    That quote was from a satirical article. Perhaps I should have inserted a ;) Still it was not that far from the truth IMO.

  5. Richard Seiter5:46 PM

    I think the 1996 Ariane 5 explosion was a more expensive software bug. gives $500M as direct costs for rocket and cargo ($7B including its development, but most of that has probably been amortized over the successful launches). Some other candidates:

    Driverless cars present an interesting risk/reward tradeoff. I think the software bugs would be less frequent (at least eventually) than the existing "human error" bugs. That said, I think driverless cars will be severely impeded by the American legal climate. It will be too easy to sue the manufacturer and people will not feel any sympathy as they would towards a "human error" accident. Also, I think it likely that 1% machine error/99% human error in an accident will end up translating to "machine manufacturer responsible."

    Thanks for the post, Steve. Does anyone here know if companies are able to (and do) buy insurance for incidents like this? It was interesting to take a look at KCG activity for this week. I wonder who all the smart money getting out yesterday was. Pretty amazing that a 440M mistake could knock 700M off the market cap (Tues. close to current bid/ask) in two days.

  6. Reed Burkhart3:19 AM

    It would be interesting to know who profited on the other side, and whether there was any uncharacteristic concentration of who it was that so profited. Mistakes, historically, have sometimes been architected -- the classic being favored boxers or favored sports teams throwing a game for profit. What protections exist against this with computers?

  7. Of course, the last-level metaware might be inadvertently bypassed, or have its own bugs/design flaws.

    I'm quite sure that social "software" bugs are still costing us way more than computer software bugs, but it'll be interesting to see how long this remains true.

  8. Ken Condon11:23 AM

    Whoa--even the screwed pooch in your Facebook is on the uptick today! Time to load up my friends and beat the bots. GRPN is also sailing away. Although this was micro seconds ago. Could be ancient history by now.

  9. "Headline and computer driven stock values make for a volatile ride." Nothing satirical here. Predicting headlines may be the next market play.

  10. LondonYoung9:14 AM

    I haven't seen any meaningful legal/regulatory changes yet as a result of the flash crash. The regulators are buried blowing deadline after deadline trying to implement a flawed Dodd-Frank - they don't have any spare time. To be fair, the flash crash was a one day kinda thing, so not that worrying. However, if Knight had built up 10 or 100 times the position it did, maybe these things would take on a new urgency.

  11. LondonYoung9:28 AM

    The little guy ought to use to screen for stocks that are cash flow positive after a decent dividend, put in a standing order to buy deep below intrinsic value, and hope for the next flash crash, right? ;-)

  12. Ken Condon11:34 AM

    Assuming they decided to fill the trade if it plummets more than 60%. I own some stocks but I have no sell limit orders standing. Some say that’s crazy but is there is another, say 50% flash crash, I could be cleaned out below the limit and then see my (ex) stocks recover in an hour. So either way it is somewhat disconcerting sitting, watching and hoping as a “naked” long.

    Any advise?

  13. LondonYoung3:04 PM

    Stop orders are crazy, IMHO. In fact, ever since the flash crash, I won't even use market orders. I only punch in buy orders limited to just above market, and sell orders limited to just below market. Note that if you have placed a buy order well below market and the market did in fact trade down there, the major respectable brokerages have always filled you.

  14. Ken Condon8:02 PM

    I don’t mean to post too much here Steve but this topic really does interest me. And LY seems to have some good knowledge on this and I am grateful for his responses. So London- here is an example of what I was trying to convey about an individual (little guy) being burned on a stop loss order. And I also wonder who scooped his shares for cheap, but don’t know how this was ultimately resolved.

  15. I forget where I saw it, but I saw a number like 1000 cancel-replace orders per [minute|second|?] in this little adventure. Seems like they are trying to induce a little market drift to create their opportunities.

  16. "Of course, the last-level metaware might be inadvertently bypassed, or have its own bugs/design flaws."

    I would call that exactly not-understanding-your-business-model. Especially if your business model is built around automated stock trading.

    I have a lot of latitude to set my own direction at work. My biggest fear, and maybe misgiving, about leaving is that I would end up somewhere with people who make gross mistakes like this.

  17. Just by-the-by, shouldn't you be hedging your downside risk [if you think it is relatively immediate] by buying puts, rather than using stop-loss orders?

  18. LondonYoung10:12 PM

    They have played around with the rules for cancelling orders but, IMHO, cancelling trades causes more harm than good. IMHO, the exchanges need to NOT ACCEPT trades unless they are not going to cancel them. This little guy got burned (by the way, I am shocked MS didn't make him whole, normally white shoe brokerages protect the small accounts) but stop orders are notorious for this problem (particularly back in the internet bubble nasdaq days). It is well within the capabilities of exchange software to not accept as "filled" orders that the exchange intends to later reject.

    But, let us turn for a minute to what this guy who got burned by a PM stop loss really wanted. Why leave a stop instead of putting in a limit sell? Presumably he imagines the market "trends" so he want to sell when a downtrend gets going even if he isn't watching. But stocks are more stochastic than trendy, so stops are a nightmare of false positives. So what he really wants to say is "sell for me only if the stock goes down and stays down" - but if he does that, you have to wait to execute in which case the stock might fall further. So, there is really no winning. Now, we can step in and regulate in any number of ways to stop what happened to him - but show me any regulation and I will show you its downside. Speaking for myself, I would only allow accredited investors to leave stop orders ...

  19. LondonYoung10:18 PM

    Lots of cancel/replaces irritates a lot of people. A commonly suggested remedy is to mandate a minimum time that an order must be good for. But this is a problem if you want to show prices on more than one exchange and cancel everywhere else if filled in any one of them. Such rules tend to favor there being only a single exchange - and then you have to deal with the resulting monopoly. Then, the next remedy if that if you put in multiple orders you may only cancel quickly if you really are filled somewhere - which is perhaps do-able as a rule. Another rule might be to allow people to populate a central order book and execute the book only once per day - say at 4 p.m. This would have the advantage of giving CNBC very little to talk about all day ;-)

  20. Ken Condon11:24 PM

    I have never used stop loss orders but have investigated puts and calls. Can’t say as I understand them as far as guessing short term trends-but have done quite a bit of study on them- so have stayed away from them. I’m old school-trying to invest in companies that I think I have some clue about regarding their business model. And prospects. For stocks I do as well on DD as I think I can-even if I do own a few shares of GOOG ;)

    And thanks BTW LY. Your responses are always appreciated.