Goldman Sachs => NYC jail

Goldman Sachs will invest almost $10 million in a New York City jail program that will allow the investment firm to profit if it can reduce recidivism rates, city officials announced Thursday.

http://www.msnbc.msn.com/id/48460327

LOL.    They want to upgrade the facilities in case they are handcuffed and loaded in the van soon?

"I like what you've done with the place"

It would be a nice retirement home for those enemies of main street...

Ah, more tragedy on Wall Street. Who's gonna give them cash to make their mistake go away? That's how it works, right?

This investment company just lost half a billion dollars due to a software bug? I call BS.

There is no way on earth I am going to believe that any company would deploy software that is so buggy it cost half a billion dollar mistake. That's a ridiculous story. Programmers test software. Financial trading software will be the most tested and robust system around. Or am I to believe that with so much money at stake, the big shots won't notice a bug here or there? Or maybe they employ Martians to write software. Ludicrous.

Same looting that 'mysteriously' cost Goldman Sachs 1.1 billion a week or two ago. Looters. The shill is being called before the governments can stop the laundering. It's putrid news. Let them enjoy their jail cell makeovers firsthand, I say.

Speaking on Bloomberg Television, Knight Capital Chief Executive Officer Tom Joyce said the firm had "excess capital right now." On Tuesday night, it had put in new software that had a bug, he said.

"This issue was related to Knight's installation of trading software and resulted in Knight sending numerous erroneous orders in NYSE-listed securities into the market," Knight said. "This software has been removed from the company's systems."

http://news.yahoo.com/knight-seeks-financing-440-million-trading-loss-124602408--finance.html?_esi=1

Oh, I thought you were announcing some head honchos are heading off to jail... what a disappointment

@Minh: Same thought here.

I don't know, in such unstable economic situation, those investers will proactively try to invest in new kind of things and see if they can make profit out if it.

19 hours ago, JohnAskew wrote

This investment company just lost half a billion dollars due to a software bug? I call BS.

Have absolutely no desire to defend the "crooks", but that statement is obviously said by someone that's never worked in the financial industry.

5 minutes ago, wkempf wrote

*snip*

Have absolutely no desire to defend the "crooks", but that statement is obviously said by someone that's never worked in the financial industry.

How is the financial industry different? How could such important software be so shoddy? I cannot put it into my brain. It is not even close to logical.

For me, it is too coincidental to happen just now, as the crooks are being spotlighted and their chance to loot is likely ending soon (until we forget why). Yes, I do think there is that much graft on Wall Street. And, no, I cannot believe any company would allow 440 million dollar software bug to go into production -- it is ridiculous! What can be said? "Whoops"? Criminal negligence at best, I'd be lawyering up if they'd poached my investments.

So who tests sofware?

A simple bug in a financial application can cost millions of dollars per second. Some of these subtle bugs can pass developer testing, QA testing and even be in producation for months before they are caught. While I've not seen a half billion dollar bug, I've seen enough multimillion dollar bugs to not question at least the possibility.

@wkempf: I'd want an autopsy report for the 440 million dollar software bug.

Yes, I'm hostile, but if the financial investment sector is so complex that software can't be reliably written against it, then I would say it is too complex.

I don't trust complexity. It too often is a smoke-screen for some other objective.

If writing investment software is so hard, then isn't it in the sector's interest to reduce the complexity? -- my point is that simplification would eliminate what I see as a strategy to allow 440 million dollars to fall off the table (and into a gym bag on the floor).

Perhaps it was a honest mistake. Either way, how can this happen, it is less rational than I am right now.

Speak of the devil, guess who bailed Knight out last night?

Goldman Sachs.

Under trade-settlement regulations, the Knight-Goldman block transaction must
settle three business days after the initial agreement, which means the seller
must come up with the $440 million in cash by late Wednesday. As of June 30,
Knight had about $365 million in cash, according to a securities filing –
raising important questions about whether the brokerage firm could generate the
additional money by early next week.

http://www.cnbc.com/id/48489000

Good Ol' Boys Network... keep the money in the club... move along, nothing to see here...

I smell fat rats.

@JohnAskew: You don't get it. It's not about complexity, but scale. A simple bug that rounds incorrectly, losing a cent on every 100th transaction can still cost millions due to the sheer scale. I have not a clue what caused the problems Knight-Goldman is claiming was due to a bug, but I can give you any number of scenarios that could lead to this despite best efforts to prevent it (i.e. testing). I've seen these occur... and it's not fun to be even indirectly involved when it happens, but it does happen. Always made me wonder about medical software, to be honest.

I understand that 440 million can rack up quickly with millions of transactions, I get scale.

This was no simple rounding error. From what I read, it was that orders to buy stock went askew.

The stock that the software ordered was not profitable, not at all. It was as if some random lookup was used. The evaluations of stocks was so buggy that it yielded weirdly random stocks. Now those who happened to own the stocks that were 'erroneously' ordered -- they got a huge bump. Who might that have been? I'm curious.

Let's pretend that the data used in testing was not reflective of production data, so that the algorithm was untested and the bug was missed that way. Now I can relax, or pretend to do so.

 EDIT: angst grows at the thought of test data that isn't reliable. 440 million is too much for this type of mistake -- I just cannot believe it is true. Programmers are to blame?  - gulp -

From what little I read it sounded like the same transactions were submitted multiple times. I'd guess that that would occur from bad error handling. Obviously I don't know a thing, but I'm not going to jump to any conclusions, because I've seen precisely what I just described happen. Regardless, I could care less that it happened, it's what happens afterwards that matters and is where all discussion should be focused.

@wkempf: I have more respect for my profession than to think this expensive of a bug would go into production. I just can't imagine it. How could they not capture production data and use it in the test environment? I don't believe it, really. Run it parallel to production with production data? How could it go into production untested? I'm shocked, honestly.

@JohnAskew: The trouble with that is that the "rules" that such systems work with change frequently, as people try to optimise the processes involved to maximise their profits. It's just as easy to "lose" a few hundred billion by delaying a change for more extensive testing as it is to rush out software without sufficient testing.

A good article of what went wrong:

http://www.theregister.co.uk/2012/08/03/bad_algorithm_lost_440_million_dollars/

"Knight Capital's software went out and bought at the "market", meaning it paid bid price and then sold at the ask price – instantly. Over and over and over again. One of the stocks the program was trading, electric utility Exelon, had a bid/ask spread of 15 cents. Knight Capital was trading blocks of Exelon common stock at a rate as high as 40 trades per second – and taking a 15 cent per share loss on each round-trip transaction. As one observer put it: "Do that 40 times a second, 2,400 times a minute, and you now have a system that's very efficient at burning money"."

2 days ago, Proton2 wrote

A good article of what went wrong:

http://www.theregister.co.uk/2012/08/03/bad_algorithm_lost_440_million_dollars/

 

"Knight Capital's software went out and bought at the "market", meaning it paid bid price and then sold at the ask price – instantly. Over and over and over again. One of the stocks the program was trading, electric utility Exelon, had a bid/ask spread of 15 cents. Knight Capital was trading blocks of Exelon common stock at a rate as high as 40 trades per second – and taking a 15 cent per share loss on each round-trip transaction. As one observer put it: "Do that 40 times a second, 2,400 times a minute, and you now have a system that's very efficient at burning money"."

Like I said before, I have a hard time accepting such an obvious bug went into production in such a expensive environment. It is beyond negligence. A bug like that would show up in testing immediately, am I wrong? Blame the programmers when, in fact, there may have been investors holding Exelon stock Sunday night who are laughing now. I guess programmers on Wall Street are hacks.

Why is this bug "obvious"? Have you read the source code and seen what bug caused this? As I've stated, I've seen similar bugs get into production, and I know full well the level of testing that occurs in these systems. But, if you want to cling to some conspiracy theory, no amount of logic is going to sway you, so I realize I'm wasting my time typing here.

Ok. It wasn't a conspiracy. I can let that go.

How could that bug go unnoticed? Wow. I would not be able to sleep at night if I let that go by.