CNBC Software Flaw Worth $1 Million?

Strudelkugel writes "BusinessWeek tells the story of one obsessive fan who unraveled a software glitch worth one million dollars. Jim Kraber was a regular CNBC viewer, and when the opportunity arose he took the 'Million Dollar Portfolio Challenge' very seriously. At one point, he was spending 12 hours a day on the contest, using three computers to trade 1,600 different portfolios in a theoretical stock game. His efforts got him into the top 20 finalists, but in the last round of trading he noticed some unusual patterns. 'One trader had a stream of near-perfect picks, consistently placing huge bets on shares that soared in after-hours trading. Kraber suspected the trader and perhaps others were getting help from someone who was changing their picks after the stocks' increases — and he quickly notified CNBC ... Kraber says CNBC rebuffed him at the time, but now it looks like he may have been right.'"

Stick to poker -- fewer variables...

[by Anonymous Coward]

Looks like Kraber wasted a lot of time, effort, and electricity and has nothing to show for it but 'TFA'.

Re:Stick to poker -- fewer variables...

[Chris Pimlott (16212)]

Not true - according to the article, he won $10,000 for winning one of the first round weekly games that got him into the final.

Re:

[GodaiYuhsaku (543082)]

The $10K was for making it to the final 20. The weekly prize was a bunch of stocks. So its more then just the $10k.

Re:

[Chris Pimlott (16212)]

It doesn't appear he is being accused of any wrongdoing himself, rather he is raising the issue about others.

Re:

[eric76 (679787)]

As I interpreted the article, players were explicitly permitted to have more than one portfolio at a time. From the article:

The key was that CNBC put no limit on the number of portfolios a player could manage, and only the best-performing one would count.

If there is no limit, than it would be inaccurate to claim that it was against the rules or an exploitation of the rules to run 1600 separate portfolios.

Re:Stick to poker -- fewer variables...

[PopeRatzo (965947)]

If stock trading was simply a matter of building hundreds of pseudo random portfolios but only the best ones counted we'd all be billionaires.

No, we wouldn't. I guarantee that I could manage hundreds of portfolios and not make money on any of them.

It's why I stay out of the stock market, except for my managed TIAA CREF account and the few Google shares I bought at the IPO.

Re:

[AdamWeeden (678591)]

Or, more likely, your net losses + net gains would likely track a similar pattern to the index of the stock markets you invested in.

"Glitchs" in the financial markets

[Plutonite (999141)]

Capitalizing on real software glitches is something that happens frequently on more volatile markets like the foreign exchange (fully automated since for retail brokers since 2000). Once the cat gets out of the bag however, the immense loss of the brokers (who usually automatically place orders opposite the sheeple, who are expected to lose on average) causes an alert. Nobody can fool the money markets on things like this for too long.

On the other side, automated trading means that brokers can engage in dirty practices like sending incorrect data to a particular client connection to trigger a trade (they call this stop hunting). Again, this is found out when clients compare data streams from more than 1 broker.

Dangerous stuff. If you are good, try to stay invisible.

Re:"Glitchs" in the financial markets

[Plutonite (999141)]

Those interested in the volatility (and thus profitability of "glitches") of the forex markets should check this excellent article on the structuring of the currency markets.

http://www.forexfactory.com/showthread.php?t=7484 [forexfactory.com]

Glitches aside, forex is a great place for smart geeks to make (and lose!) money on the side. All you need now is an internet connection and a mini bank account, whereas 10 years ago this was the exclusive arena of millionaire hedge fund managers. Warning: addiction very probable. Try at your own risk.

Re:

[nacturation (646836)]

A while back there was a discussion on stocks and someone pointed out a site where you could open an account, code up rules for stock trades, and then execute those trades based on your algorithm. You could do this either in a sandbox environment with play money, or pony up actual cash and make or lose big. A stock market API of sorts. I've been meaning to find that site again (likely there's a bunch) -- anyone have some URLs for sites like this?
 

Re:"Glitchs" in the financial markets

[Plutonite (999141)]

All retail brokers for the forex market offer these services now, it's not just one site. Stocks are a different story.

The most popular client platform is MetaTrader, and every broker has their own customized version which connects to their servers. There is an entire C-based language used to program your experts (they call them expert advisors) and you can indeed open multiple demo accounts at no cost. And the capabilities are so great you can easily download scripts that parse news events released real-time on websites and react accordingly with no input from you.

Isn't it amazing what Turing machines can do? Word of warning though: you will never get rich on an black-box like an algorithm. Use the technology, but don't get carried away.

Re:

[nacturation (646836)]

All retail brokers for the forex market offer these services now, it's not just one site. Stocks are a different story.

The most popular client platform is MetaTrader...

Thanks, that might have been the one I was thinking of. And for stocks? What's the different story?
 

Re:

[Thing 1 (178996)]

And for stocks? What's the different story?

I've been interested in trying out VectorVest [vectorvest.com] for a few years now. Anyone have any actual experience with it? Their marketing shows great returns, 90% a year.

Re:

[inviolet (797804)]

Thanks, that might have been the one I was thinking of. And for stocks? What's the different story?

Because any prediction algorithm running on publicly-available news will quickly be duplicated by the millions of other algorithm developers out there, cancelling out whatever legitimate edge it might have enjoyed for a few days or weeks.

Re:

[TopShelf (92521)]

I know TD Ameritrade offers a tool called StrategyDesk which performs such tasks...

how is stop hunting done ..

[rs232 (849320)]

"dirty practices like sending incorrect data to a particular client connection to trigger a trade"

This *is* interesting. Do you have anymore detailed information as to how this is done. What technology is the data connection run on? How is it possible to fool.

Re:

[Plutonite (999141)]

That's ok, we already know you.

Regards,
The FBI

I know. Central banks will often have taps on the activities of major players (investment hourses), so that they can have a measure of control. Very exciting currency fights happen at times. Extremely successful traders will be "shadowed" by their brokers and I don't know how far the intelligence thingie goes, but currencies ARE a matter of national security.

Say hi to Tom!

Regards,
-P

The glitch

[niceone (992278)]

Here's 'the glitch' if you don't want to wade through the article:

If you made the selection before the close of regular trading at 4 p.m. EST and left your Web browser open, you could execute those trades after hours and still receive the 4 p.m. closing price. ..... The allegation is that certain traders may have used the technique with companies that were reporting earnings and other important news after the market's close. They could select as many as 50 stocks and then execute trades for only the one or two best performers.

Re:

[Timesprout (579035)]

Its a surprising oversight for such a time sensitive activity and quite easy to stumble upon I would have though. It does point to a lack of end to end testing though or else the testers were just not paying attention as long as a trade was accepted.

Gaming the system?

[by Anonymous Coward]

Seems Kraber was certainly violating the spirit of the contest, if not the letter. Relying on 1600 portfolios and the law of averages to "win"? Seems like he's pissed that somebody else found a better, easier way to cheat.

Re:

[someone1234 (830754)]

This was exactly my idea, he's a sour loser who did not cheat good enough. But well, all those cheaters should be disqualified. That software should have extensive logs about when people placed a bet. It is impossible they don't have a log about the exact betting time, even if it is only some firewall log (sure they need some intrusion defense). If they are relying only on the client data, the whole contest is a fake.

Re:

[timeOday (582209)]

Considering that someone could easily have 1600 portfolios and make use of the law of averages in real life (if they've got the cash for it), I wouldn't think of it as cheating.

Uh, no. In real life you can't just run a bunch of high-risk portfolios and welch on all but the best one; when you lose you have to pay. Kraber's tactic does nothing to increase returns on average.

Fantasy stock markets.

[tygerstripes (832644)]

Remember playing Monopoly as a kid?

"Imagine, right... nononono, listen, listen. Just imagine... if all this was real money!!!

This has the same kind of feel to it.

And I have to ask, if the guy's prepared to spend 12 hours a day doing this with "Monopoly money", even sacrificing his professional accreditation studies in the belief that he might end up as the best market-player in over 300,000 and win $1m... why the hell isn't he just playing the stock-market??

Re:

[ZarfMouse (154055)]

According to the artile, the only reason he believed he could win was because the contest allowed him to maintain over a thousand simultaneous portfolios and only the top scoring one mattered. His strategy was to take high risk knowing that most would lose but some would win big time.

If he was using real money he'd lose out over all but in this contest he's guaranteed to have one great portfolio. He had a very high chance of winning $1m for his efforts if the game had been more secure but only because he un

other peoples real money ..

[rs232 (849320)]

Because he first has to get other people money to trade with. Else he is what is known as a day trader, a mug who bets his own money on the stock market. To see how the stock market is really run, take a look at Liars Poker [amazon.com].

was Re:Fantasy stock markets.

Letter to Jim Kraber.....

[IHC Navistar (967161)]

From: Mark Hoffman, CNBC, Inc.
          Steve Ballmer, CEO Microsoft Corp.
To: Jim Kraber
Re: Software Glitch

Dear Mr. Kraber,

          I regret to inform you that, after a thorough investigation of the alleged trading irregularities by independent Microsoft software engineers, we have determined that the perceived trading irregularities were not the result of a software 'glitch' or 'bug', but were in fact security features.

We sincerely apologize for any inconvenience this may have caused.

-Mark Hoffman
  Steve Ballmer

Anyone can win with 1600 portfolios

[bhmit1 (2270)]

This reminds me of an old email/fax scam trick. You start sending 1600 messages to people on a stock that's going to have big news the next day, either very good or very bad, no one knows. To 800 of those people, you say it will be good, and the other 800, you tell them it will be bad. The next day, you take the 800 who you predicted the right answer for, take another stock with big news coming out, and 400 of those people will end up with the right answer. Then 200, and on the 4th day, 100. Now for those 100 remaining people, you send a message saying that you've been giving valuable stock picks for the past 4 days and how much would they be willing to pay for your tips. The moral, everyone is a winner when your losers don't count. If you were hoping to find a good stock trader from this contest, this wasn't the way.

As for the bigger picture, I'm not a fan of "trading", though it does have it's place. I'd rather use the market for long term "investing" and doing something that provides value to the world with the rest of my time. But just like with power and politics, money corrupts, so we should expect that people will abuse the system and just do our best job ensuring we aren't the ones they are abusing.

what a horrible waste

[jollyreaper (513215)]

42-years old and spending 12 hours a day playing a stock trading game. Wasn't there an MMO he could be applying himself to instead?

The technique is used in Biblical text dating

[Budenny (888916)]

The Higher Criticism, which started out in Germany in the 19c with the aim of establishing a definitive chronology of Biblical events, laying out exactly what the historical evidence for them was, and to data all the various books, used this as a criterion. Scholars still do.

The rule is that if some publicly dateable event is clearly forecast in a text, the text was written after it. How long after is a question. Hume made a similar point. Miracles are by definition violations of natural law. To the extent that they are miraculous, it must be more probable that the natural law held and that either experimental conditions were not correctly reported or the story is false. So they end up either not having happened, or not being miracles.

Funny to see this stuff coming up in exactly the same reasoning about stock market predictions....

Keep business logic off of your clients

[ishmalius (153450)]

Keep all of your decision-making on the server where it belongs. Let the client be a view to the server, and validate input. But don't trust the client with any control of the process. It only results in problems like this.

Flaws in contest software

[elh_inny (557966)]

I find that Marketing departments are completely unfit to provide a secure platform for fun & just competition.
It usually is a very talented guy, who however has his focus on the looks, not the engine.
I once, for curiosity's sake took part in one contest. Scoring poorly, I began to analyze the inner workings of that FLash site.
I have quickly found that the answers to the trivia question were stored in plaintext in my browser cache!
I notified the organizers, but no actions were taken, I also soon began to notice how people bagan to score more than it was possible according to the game's rules.
Eventually, they didn't change a thing, except banning people beyond certain score, in the end all my friends got the prizes, CD players, cups etc.
One year later there was a new contest, almost identical glitches, this time however I decided not to get my friends in trouble, just in case.

Re:Flaws in contest software

[MadCow42 (243108)]

Behind every Marketing person with an idea, there is a programmer that has to implement it. I don't think you can blame the technical issues you quoted on bad marketing... just poor programming.

Yes, I'm in Marketing... and yes, I also program (semi-professionally). However, I realize that I am not the person best qualified for doing "mission-critical" projects like that (mission-critical to a Marketing program, in this case).

MadCow.

Re:

[Lockejaw (955650)]

I assure you that there can be plenty of stupidity in the requirements. If it really is bad, no amount of programming skill will overcome it.

Re:

[Duhavid (677874)]

Lemme guess...

The programmer probably needed about 3 months to put together the project,
but Marketing had already set the dates, and he/she/it only had about 2 weeks
to push it out.

But it's not marketing's fault that they set a deadline without any knowledge,
it is the programmer's.

Re:Flaws in contest software

[WombatDeath (681651)]

Behind every Marketing person with an idea, there is a programmer that has to implement it.

Behind every Marketing person with an idea there's a project manager gritting his teeth, pulling on his knuckledusters and preparing equal doses of coffee and sedatives for the programmers. And he's probably speed-reading the old classic: "No, We're Not Going to Rewrite our Corporate Website Entirely in Flash, Because it's a Stupid Fucking Idea and You're Retarded For Suggesting it. Again."

Re:Flaws in contest software

[Fred_A (10934)]

I once, for curiosity's sake took part in one contest. Scoring poorly, I began to analyze the inner workings of that FLash site.
I have quickly found that the answers to the trivia question were stored in plaintext in my browser cache!

Did the same once in a game that was supposedly random, based on a Java applet, before Flash was popular. The applet ran a "one armed bandit" thingie. Locally. And then sent the result to the server which tallied points based on that. It was of course trivial to send whatever result you liked to the server in the applet's place since the communication channel was completely unsecured.
I notified the game organisers but they didn't seem to mind much either. Oddly enough it didn't seem to have been abused at the time (it was a fairly high profile, although local, european site).

Re:

[mikael (484)]

The same thing happened in a cable TV set-top box game - a cash prize was offered for anyone who could solve a 'nethack' type maze using the least numbers of weapons and moves in the shortest time. Pressing 'cancel' rather than 'yes' or 'no' for the submit score, bumped your score up by at least 200 points. Needless to say, the top two players always had scores this high - trying different moves would only cost +/- 1 point, so there was no other explanation.

Re:Flaws in contest software

[montyzooooma (853414)]

Back in the days of Prestel ( http://en.wikipedia.org/wiki/Prestel [wikipedia.org] ) and Micronet I used to play the competitions they had. Prestel used a numbered pages system and competitions on it would charge, say, 20p per question with 10 questions so getting through to the last page with the prize claim would cost two quid or a bit more if you got a question or two wrong. I soon figured out that the prize claim page for this one company's games was almost always the same offset number from the initial question page number. Unfortunately my haul consisted of a lot of disposable cameras, sponges and money clips so the novelty soon wore off.

Re:

[foo fighter (151863)]

The de facto first programming lesson is creating output: "Hello, World!"

The second lesson needs to be accepting, cleaning, and validating input: ";drop table;".

Re:

[Rakishi (759894)]

Trading is nothing but a gamble most of the time.. He gambled his time and money, and like many lost out completely. And you never know, maybe the 'other trader' found an even more Efficient Market Hypothesis.

Now replace "trading" with "life" in your statement, amazing how well it fits.

Re:

[hhcv (1094593)]

The fact is, no matter how calculated a risk is... some people have more information at hand when doing the "calculations." Therein lies the profit making opportunity for the other trader, on the back of this first guy.

Re:

[BlueTrin (683373)]

Even though some people have access to rumours earlier and have a better circle of friends, you can get yourself a Bloomberg [bloomberg.com] and Reuters [reuters.com] terminal and here you go you will have the same real-time news than the other traders !

It often comes down to the fact that some people are better at spotting/evaluating opportunities ...

Re:

[Baddas (243852)]

Botnets running on compromised windows machines run > $10k / hr for the larger ones.

So in a way, they can.

Re:

[Eivind Eklund (5161)]

Speculation provide market depth (ie, always allowing trades to be made), which is of value to the "real users" of the market (the original producers and investors).

It is also considered that speculation helps give correct prices, though that is more arguable. It is fairly difficult to put a dividing line to where investment ends and speculation begins, anyway.

Eivind.

Re:

[Colin Smith (2679)]

taking money from people who produced it and randomly spreading it to a bunch of speculators.

So... Where does money come from then? No, really. You say produced, but do you really know how money comes to exist?

 

Re:

[Plutonite (999141)]

What he says is almost correct, with bad terminology. Speculators were allowed into the currency markets to give liquidity to the big boys who need to exchange goods and services across financial borders. GP is wrong to think that all fail, however. Please see links I posted below, very informative stuff.

Re:Expected

[SpinyNorman (33776)]

Margin is just taking a loan using your existing investments as collateral - nothing to get excited about there.

Personally I don't see what's wrong with shorting - it does provide some balance. Anyway, shorting has its place for regular investors also in the form of "shorting against the box" which means shorting a stock that you already own to lock in a profit - typically for tax purposes. e.g. Say it's December and you own a stock that has appreciated that you now want to sell but don't want to pay taxes for in the current tax year... so you short the stock in December (same amount as you own), then in January you close the short by handing over the shares you already own. What you've effectively done is get the December price, but not actually completed the trade until January, so you've got another year to pay the taxes on your gain.

Re:

[FiniteElementalist (1073824)]

Apparently you are unfamiliar with hedging [wikipedia.org]. Shorting can be used as a mechanism to eliminate certain types of risks depending upon the relationship between two assets. Of course there are other hedging mechanisms that can be used in a similar role, such as options and futures, but shorting may be the specific thing you want. Futures require a margin account as well, but you'd be sort of silly to complain about their existence. I'm sort of curious what your opposition to speculators is rooted in, as the "ta

And not the last time either...

[ArsenneLupin (766289)]

Wow. I'm surprised this could still happen. Back in the 90s, there was a mutual fund that got ripped off by the same practice. Basically they would let you buy into the fund after the market close, and still get that day's price. Well you can guess what happened. People would watch the stocks that were in the fund and if they went up, they'd buy into the fund at the old price. Of course that ended up costing someone real money since the fund would then have to buy those stocks the next day at the higher price.

Our company has an employee stock option plan, which allows you to sell your options for the previous day's closing price. So you can pull similar shenanigans, in reverse.

CNBC's reply

[Notquitecajun (1073646)]

QQ PVP server L2P noob.