The 3 Billion Dollar Typo 398
Rand310 writes "Mizuho, the world's second largest bank based in Japan, with total assets of nearly the GDP of France (around 1.2 trillion USD) accidentally sold 610,000 shares, valued at $3.1 billion... for 1 yen each. A 27 billion yen loss would almost match Mizuho Securities' group net profit of 28.1 billion yen for the financial year ended in March, though... the incident would not threaten the brokerage's financial stability. FYI 1 yen is about .83 cents. Yesterday one share was selling at $5,065, today you could theoretically have bought 610,000 shares for $.0083 each. An expensive switch of variables."
Re:come on now (Score:3, Interesting)
The real gist of the article is this was an IPO, that there was a typo, nothing actually sold that low, but the confusion caused by the low prices probably deflated the opening day IPO pricing--therefore the firm is trying to buy back what did sell so they can fix things. The buyback costs could be around $224m.
Wild. (Score:2, Interesting)
Im curious how this is going to effect the business itself.
You've got to admire the Mizuho execs... (Score:5, Interesting)
The company made a horrendous mistake and yet, there you see two executives bowing apologetically and taking responsibility on the day it happened .
I have to wonder how a U.S. bank would have handled such a mistake?
Scary forms (Score:4, Interesting)
This guy's problem was presumably different; he knew what the forms meant but entered the wrong numbers. Still, it's kind of scary to be looking at a computer screen and thinking, "I hope this is right, or it's REALLY gonna suck."
Trading typos are hardly that rare... (Score:4, Interesting)
And a Bear Stearns employee typed in $4bn instead of $4m in 2002, again moving the index (thsi time the Dow Jones) down.
Mostly though these positions are unwound by agreement between the parties. I don't understand why that didn't apply here.
.
"Was ignored" -- human nature, SPOF (Score:3, Interesting)
I'd be really curious to know how something so dramatic could possibly be written with a "check" that could be ignored with trivial effort or due to plain inattention. Yes, it's human nature to ignore "Confirm" dialogs, and efforts to explain things to the user within the standard Windows API so often end up like the "Do you really want to save this as a CSV" dialogs in Excel. But c'mon -- no single point of failure should result in something like this happening.
There has to be an escalation process in place to bounce serious problems up a review tree for others to scope out. You'd think bankers, of all people, would demand that from their software. I can't even post a news item on our intranet without legal reviewing it, for goodness' sake.
Re:Data Validation (Score:2, Interesting)
I work in the software department of an investment management company. Trade errors are bad, but are certainly not unheard of. Sometimes it's a bug in the software generating the orders (yes, in spite of strict QA and user acceptance testing, weird "one in a million" bugs make it through to production). Frequently it's user error. Sometimes it's bad data in the database, or some simple misunderstanding that creates bad data.
When we have a trade error, we have to follow a process not only to fix the problem in the short term, but implement a process to make sure that same error doesn't happen again. It works pretty well.
But to say they just clicked "Yes, I'm sure" is oversimplifying it a little.
Re:You've got to admire the Mizuho execs... (Score:5, Interesting)
I guess it worked out - my boss retired a year or so later and I got his job. And it's a lesson that I've never forgotten (and, fortunately, only had to use once).
-h-
Re:Would be nice, but not really... (Score:3, Interesting)
Re:1 BILLION Shares! Muuuahhahaahahahaha (Score:3, Interesting)
While the end-user software could in theory figure out whther the sale was a short sale or not based on how many shares you currently hold, this would not be a feature! There are circumstances in which you want to sell shares short while holding those same shares (this used to be a tax dodge, but the IRS closed the loophole a while back). The user should always know whether he's selling short or not.
The software in this case was just poorly designed.
Reading between the lines (Score:1, Interesting)
Re:You've got to admire the Mizuho execs... (Score:3, Interesting)
something is wrong with the Japanese market (Score:3, Interesting)
To operate a computer-based market without such a protection in place is pretty reckless.
Re:Would be nice, but not really... (Score:4, Interesting)
What's worse is that the default button on this dialog box is the "Reboot now" button.
So I'll be typing in an email or a posting to something, and after 3 minutes are up, it will suddenly pop up the nagbox into the foreground, just in time for me to hit spacebar after finishing a word. In a blink of an eye, the computer forces every program to end without prompting to save and reboots...
I can understand and can live with crashes. but that... That behavior was PLANNED.
Re:You've got to admire the Mizuho execs... (Score:4, Interesting)
Here's an article that focuses on these very kinds of problems in the music industry: http://www.indie-music.com/modules.php?name=News&
You might find this particular passage of interest:
A few years ago at a party, I asked a CEO of a major label why this practice seemed so prevalent at the top executive levels of the music & film industries and the response was astounding. He said, "What you have to understand about the decisions to hire executives at that level, is that very often the boards of the company hiring them are much more comfortable with someone who's already had the position and done the job regardless of their past track record than someone they don't know regardless of their ability!"
This problem is endemic not just in the music industry, but in almost every sector. With all the fat perks, it's almost set up so that bad performance is completely inconsequential.
Re:Computer that cryed wolf (Score:3, Interesting)
What I hate even more (yes, Microsoft I'm talking to YOU) is that "OK" and "Cancel" are not synonymous with "Yes" and "No".
It's just stupid. Or even worse is the 7 (or so) lines of text Paint Shop Pro gives you describing the effects of leaving lots of data on the clipboard before you exit the program, followed by "Yes" and "No" buttons. I have to read the damn thing every time because I can't recall if it asks you to leave the data on the clipboard or delete it...
How about naming the buttons "Leave on Clipboard" or "Delete" so I don't have to wade through several lines of text when I already know what you're frickin' asking, but can't remember how you worded the sentence.
I think programmers too lazy to write message boxes with buttons other than "OK" and "Cancel" are just another reason why I'm convinced most software developers (especially at Microsoft) simply hate users. Microsoft gives you an error message, "You can't do X unless you do Y." If they cared, you could press a button on the message box to do Y, or at least being up the Y dialog (which is buried in some arbitrary configuration area that randoomly changes every time there's a new version of Windows, but that would require 1.) The high school interns who implement Windows UI to get off the foozball table and 2.) Microsoft to have something other than complete indifference if not outright hostility to the people who use their software. It'll never happen.
That and small dialog boxes with long lists that can't be resized. There's a special circle of Hell for people that do that and there's already a whole section roped off just for MS.
Of course, everything started going to hell when "chrome" began to be equated with "usability" about 8-10 years ago... and now every hack who's barely literate with Photoshop considers himself (and worse, IS considered) a usability expert.
Most UI is no better than it was in 1990, and any good ideas we've seen since then were probably on the NeXT or even on the Xerox PARC software. Extrapolating Windows Media Player to the year 2010, the actual viewable size of the screen will be the size of postage stamp, and there will be 31 different skins to let you make it look like different parts of the body and full-screen view will have ad banners on the top and bottom. But that's just a guess...
Re:Would be nice, but not really... (Score:4, Interesting)
Was it asking, "Would you like to format your hard drive?" or "Press OK to transfer all your bank account funds to Joe Smith.".
This has been a problem for ages on Windows, and I can't believe they have yet to address it. A simple fix would be to create a brief quiet period that would allow the user to react and stop typing and respond.
--
Q
Re:I work in this industry... happens a lot (Score:2, Interesting)
Most traders make a few mistakes a week - but they catch them in seconds and back out the position. It's the ones that are made and not noticed for weeks that cost the real money.
It's part of the game - the risk a trader takes. When you find the error, you fix it by going the opposite direction immediately and then forget about it - Like Tiger woods at the next tee after he misses a 3ft put.
It can be *really* entertaining though! I once watched a lady melt down when she realized 20 minutes after the market closed that she had not sold 100 shares of a blue chip stock, she had types in the 12 digit account number and the firm was on the short side of a multi-hundred-million dollar trade.
Flip a coin... She lost the company only $80,000 as it was covered at 9:30 am the next morning and it could have just as easily been a brilliantly profitable trade.
THis is always there - you'd think $250 Million would buy a company a good set of electronic surveilance systems to identify outliers which just seem odd ('why is our guy selling this 610,000 yuan stock at 1yuan?!' says the comp in a split second). But there will *always* be human error.
Torontoman
Re:Data Validation (Score:5, Interesting)
Regarding the FYI part of the post. (Score:2, Interesting)
Re:Would be nice, but not really... (Score:3, Interesting)
It does not matter how large or explicit the warning box was. If every large transaction got the same verificaton box re-stating the terms of the trade (and I suspect this is the case) the warning was useless because it will be ignored 999 times out of 1,000. Only if warning or confirmation boxes appear infrequently, when certain parameters are exceeded, will they be paid any attention. A warning box that constantly gives false negatives is poor design.