Slashdot Mirror
Knight Trading Losses Attributed To Old, Dormant Software
New submitter alexander_686 points out a Bloomberg article about the cause of Knight Capital Group's $440 million algorithmic trading disaster from a couple weeks ago. The report says a dormant software system was accidentally activated on August 1, which immediately began increasing stock trade volumes by a factor of 1,000. The Wall Street Journal has further details:
"Knight Capital Group Inc.'s accidental trades earlier this month were triggered by a flawed upgrade of trading software that caused an older trading system connected to the computer code to inadvertently go 'live' on the market, according to people familiar with the matter. The errors at Knight on Aug. 1 involved new code the Jersey City, N.J.-based brokerage designed to take advantage of the launch of a New York Stock Exchange trading program, which was introduced that day to attract more retail-trading business to the Big Board, the people say. ... When NYSE Euronext trading floor officials called Knight at about 9:35 a.m. to try to pinpoint the cause of unusual swings in dozens of stocks, just after the Big Board opened for trading, Knight traders and their supervisors had a difficult time detecting where in its systems the problem was located, say people familiar with the morning's events. The NYSE had to call Knight several times before deciding to shut the firm off, the people say."
They really need to stop giving these high frequency traders these parachutes. You screw up your algo, its your own damn fault. Lost your butt on the market - oh well.
This is why mission critical systems should have a "No Dead Code" requirement.
General Relativity: Space-time tells matter where to go; Matter tells space-time what shape to be.
Comment removed based on user account deletion
http://www.fool.com/investing/general/2012/08/10/the-terrifying-graphic-that-shows-stock-trading-r.aspx
This is what happens when the pre-production environment is not identical to the production environment. Got egg on my face (though no direct financial cost incurred) when the production environment had that 0.01 JRE increment that addressed the new-fangled daylight saving time, and the pre-production environment did not. It caused some very strange bugs due to the change in date handling, even though it wasn't anywhere close to spring forward time. (We developers had no access to the machine, so it took a while to figure out, too.)
Yeah, with capitalism the birth lottery is often more important than merit.
Liquidity isn't about time, it's about spread. Stock markets are double-auction systems, where you are free to bid and offer at any price you want. Trades only occur when someone offers a stock for less than someone else's bid price. The stock has a 'price,' and to buy stock you have to bid a little higher, and to sell you have to bid a little lower. The difference between the bid and offer price is the spread, and the spread represents an inherent transaction cost to most investors.
Now liquidity is just how easy it is to convert your stock into cash. There is always some liquidity, as long as you're willing to accept a bad deal. You offer to sell your stock cheaply enough, or buy high enough, and somebody will buy or sell. Of course, on blue chip stocks, the spread has always been pretty small, so it has never cost a lot to trade in those stocks. But in medium-cap and small-cap stocks, where HFTs have had the biggest impact, they've reduced spreads enormously.
Twenty years ago before the rise of HFT traders, you might had paid a market maker $0.50 / stock on the spread for a trade in a medium-cap stock. If you want to rebalance your investment portfolio annually, those kinds of transaction costs could wipe out your gains. It effectively priced individual investors out of the market, and if you wanted to save money you were forced into the hands of large institutional investors, who will happily charge you a 2% management fee for the pleasure of handling your money.
Today we take it for granted that most stocks have very small spreads, and you can make regular trades without seeing all your gains lost to transaction costs. HFTs have put the Serious Men in Suits market makers out of business, and have pushed the cost of trading down to the point where the individual can manage their own savings, without having to fork over most of their profit to other Serious Men in Suits.
So yeah, you may not like high frequency traders, but they're better than the old-boy networks of "specialists" and stockjobbers that they replaced.
Knight lost the money, there was no parachute.
You're right.. but how about some details?
http://www.businessweek.com/news/2012-08-09/knight-says-it-may-face-more-burdensome-costs-from-trade-error
Knight was saved from collapse on Aug. 6, when it received a $400 million cash infusion through the sale of convertible securities to a consortium of investors.
Getco LLC, Blackstone Group LP, brokerages Stifel Nicolaus & Co. and TD Ameritrade Holding Corp., as well as Stephens Inc. and Jefferies Group Inc. invested in the rescue funding for knight, according to the Jersey City, New Jersey-based company. The investment will give the firms a 73 percent stake in Knight once the shares are converted into common stock.
So there you go... they were forced to give away control of their company to a number of outside investors.
Twenty years ago it was also much harder to match up buyers and sellers, and actual trades took a lot longer. It's hard to say how much of the decrease in spread is due to high frequency traders and how much is due to improved technology providing a more efficient, easier to access market.
Not that there seems to be anything particularly bad about encouraging people to buy long term investments.