Algorithmic Trading Glitch Costs Firm $440 Million
alstor writes "Yesterday an update to Knight Capital Group's algorithmic trading software caused massive volume buys and sells, resulting in large price swings on the New York Stock Exchange. As a result, the NYSE canceled some of the trades, but today the loss to Knight has been calculated at $440 million. Ignoring adjustments for inflation, this makes the cost of this glitch almost as much as the $475 million charge Intel took for the Pentium FDIV Bug, which might warrant adding this bug to the list of worst bugs. In light of this loss and the May 6, 2010 Flash Crash, perhaps investors will demand changes from firms using algorithmic trading, since the SEC is apparently too antiquated to do anything about it (PDF)."
Here http://www.youtube.com/watch?v=dOO9XxH5Nyo&list=UU6NBj2q25QL4kN8tqwU9c-A&index=2&feature=plcp
This space for rent.
For those not interested in going through all of the links just to find the one that links to the relevant article:
http://www.forbes.com/sites/steveschaefer/2012/08/02/knight-capital-trading-disaster-carries-440-million-price-tag/
A common defense of flash-trading is that it provides market liquidity in that it provides counterparties to the desired transactions of the rest of the market.
But I've yet to see someone discuss how the added-value of millisecond liquidity is substantially superior to having exchanges post transactions in 1-sec. intervals to discourage millisecond arbitrage during which no new events have occured and no new market analysis has taken place, only speculation and playing the system against proper investors? Can someone illuminate me on this point?
I'd tell the firm "too bad". It shouldn't be up to the NYSE to make sure companies don't do something stupid. Back in time a ways, when someone tried to game the system and then failed hard they would be ignored and forgotten. Now, with bailouts and do-overs and participation trophies, we ignore hard working americans who don't expect handouts and reward those who don't want to take responsibility for their actions.
-SaNo
Yesterday an update to Knight Capital Group's algorithmic trading software caused massive volume buys and sells, resulting in large price swings on the New York Stock Exchange. As a result, the NYSE canceled some of the trades...
So if I were to write an auto-trading script using the eTrade API, and as a result of a bug it made bizarre trades and I lost a lot of money, would the NYSE agree to cancel those trades? Didn't think so. Why should the big boys get a second bite at the apple? If you write an algorithm to do trading, then from the POV of the stock markets, that algorithm is you. (Just like the way user permissions work in Unix/Windows.)
Allowing mulligans and do-overs when well-connected firms make mistakes is only going to reinforce the perception that Wall Street is a casino rigged in favor of the rich.
Today, after the stock dropped 50%, analysts are beginning to downgrade the stock from buy to hold. Excellent analysis there!!!
http://finance.yahoo.com/news/knight-capital-downgraded-hold-buy-155956204.html
Some programmer's going to lose their job over this error that resulted in a $440 million loss. If the programmer had done the job properly, Knight would have lost $1 billion and been eligible for a government bailout.
Don't cancel the trades. If some idiotic "investment" firm lets a computer program spend hundreds of millions of dollars in seconds then good for them. They get to keep the profits and the losses.
If one of your human trader makes a typo or a computer program has a bug then bad luck, they should have had checks and limits to make sure it doesn't do too much damage to them.
The rest of us don't get do-overs.
Heck just last month I when trying to limp in $2 poker game I picked up two $100 chips and threw them forward by mistake - I didn't get do-over even though everyone at the table new I made a mistake, my $198 raise into a $5 pot plays.
I'm pretty sure if I accidentally typed 100 instead of 10 when making a trade on schwab.com I'm not getting a do-over if the trade completes.
If they didn't sufficiently analyze the code they were going to turn loose in real time trading, and it did something they didn't expect it to do, then that's their screwup, and theirs alone, and they need to own it. Period.
I can see NYSE cancelling some trades because the volume of trading was getting people confused about what the pricing should be, but I can't see it as fair that they'd cancel trades as a favor to the company. If a day trader screws up and takes a bath on a stock due to poorly-thought-out trade orders, they don't get a do-over, those trades are placed and cleared and they're done, no going back. I don't see any reason wild program trades should be held to any lesser standard, and I see plenty of reasons why they shouldn't be. What the company needs to do is get some competent programmers in to code their algorithms properly, and get some competent analysts in to double check the coders' work and validate the algorithms, and be prepared to own their own s**t if the code does something like this. Sorry, no sympathy, these guys should d**n well know better.
Why make a comparison with an event 15 years ago and ignore the different in value of the dollar?
Intels FDIV bug costs of $475M in 1994 is equivalent to $735M in today's dollars. I guess it's just not as impressive as saying "The cost of this glitch was a bit over half of the $475 million charge Intel took for the Pentium FDIV Bug."
If you want to make it sound more impressive, go back further in time "This loss was greater than the entire GDP of the united states in 1955 (ignoring adjustments for inflation)"
No way any of these trades should be unwound. You want to give an algorithm your wallet and let it make lightning trades on your behalf? Fine, but learn to live with the consequences.
Can someone illuminate me on this point?
I'll give it a try. High Frequency Traders (HFTs) are not investors, they are market makers. They find a willing buyer and a willing seller, arrange the transaction, and execute the trade. They make a profit on the spread between the buy price and the sell price. The problem is that once they locate the buyer and seller, they need to buy the stock from the seller first, then turn around and sell it to the buyer, but the buyer may have cancelled they transaction, or they may have already bought the stock from someone else, in which case the HFT is stuck with the stock and may have to sell it to someone else at a loss. If transactions are granulated to one second intervals, instead of say, millisecond intervals, then the risk of this happening is a thousand times higher, and the HFTs will insist on higher spreads, resulting in lower liquidity and higher transaction costs for both buyer and seller.
Since the introduction of high frequency trading, transaction costs have fallen considerably, saving plenty of people a lot of money. The only losers are the old market makers that used to have lucrative sweetheart deals with the exchangs Many of those old market makers are now bankrupt. Good riddance.
Now, let me turn the question around. What is wrong with high frequency trading? Other than people ranting about something they have made no effort whatsoever to understand, I haven't seen a single good argument against it. HFT was originally blamed for the 2010 "flash crash" but the full investigation found that HFTing actually made is less severe. Some HFTs have lost money because they screwed up their algorithms or fat-fingered a trade, but that is their own fault, they lost their own money, and for every penny they lost, someone else gained.
I have no personal interest in HFT, but I find desire of so many willfully ignorant people to control the behavior of others to be pretty disgusting. The advantages of HFT are pretty obvious to me.
Why not just a single trade resolution per day ?
Because traders would then just trade directly with each other or set up their own exchanges. If Emron was bad, think what would happen if the huge brokers simply decided to just trade directly with each other, or worse they set up "third party" exchanges to trade securities? The exchanges would then loose the fees they charge.
You can trade stocks and bonds on the street corner, at the farmer's market, in you living room. We just don't do it because it is hard and expensive to trade stock certificates in small numbers. Limiting trades to one per day would just encourage transactions to take place off the exchanges.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
The new HAL 9500D will make you rich, and poor again, and rich again...and all in less than 4.2 milliseconds.
What political party do you join when you don't like Bible-thumpers *or* hippies?
Contrary to TFS, Knight was not running algorithmic trading. They are a "market maker" for retail brokerages, like Fidelity, Vanguard, E-Trade and, in particular, Scottrade. (About 40% of Scottrade's traffic was going through Knight). The NYSE had just brought a new retail trading interface on-line, and Knight's software did not conform correctly to the protocol. As a result, it kept re-entering the same orders, over and over. These were small retail orders, just a few hundred shares each, but they were submitted to the exchange thousands of times.
The two outstanding questions are: Why was their interface not tested properly and why did it take them over 30 minutes to pull the plug?