from Hacker News

Oil crash busted broker’s computers and inflicted big losses

by wclax04 on 5/8/20, 4:59 PM with 176 comments

  • by haltingproblem on 5/8/20, 7:30 PM

    “Five days, including the weekend, with the coronavirus going on and a complex system where we have to make many changes, was not a sufficient amount of time,” he said. “The idea we could have bugs is not, in my mind, a surprise.” He also acknowledged the error in the margin model Interactive Brokers used that day.....We have called the CFTC and complained bitterly,” Peterffy said. “It appears the exchanges are going scot-free.”

    Thomas Peterffy must think we are idiots. Anyone who trades commodity contracts for any period of time knows that the real cost of the contract is the actual cost of the commodity - storage costs. When storage costs spike and the actually commodity spot costs go down, the future will become negative!

    One way to get a handle on storage costs is think of them being inversely proportional to the value density. The higher the value density, e.g. gold the less the storage costs. Oil is not so dense so storage costs matter. Financial instruments like the Treasury Bonds and the S&P futures contract have zero storage costs. Storage cost is of-course different than carry cost (the cost of funding your long position).

    On another aside, I have known folks who have worked at IB in the past, and their systems absolutely suck dead goats. Huge masses of legacy C++ code with poor testing. Most of these brokerage firms have legacy code base from the 90s that is poorly understood. They also have nonexistent organizational quotient around code validation, correctness and testing their risk models. A futures margin model is not something one can whip up over a weekend but a good CS undergraduate can program one over a couple months.

    Sorry for the IB customers but I have zero sympathy for IB or should I say negative ;)

  • by elliekelly on 5/8/20, 5:41 PM

    > Peterffy said there’s a problem with how exchanges design their contracts because the trading dries up as they near expiration. The May oil futures contract -- the one that went negative -- expired the day after the historic plunge, so most of the market had moved to trading the June contract, which expires May 19 and currently trades around $24 a barrel.

    > “That’s how it’s possible for these contracts to go absolutely crazy and close at a price that has no economic justification,” Peterffy said. “The issue is whose responsibility is this?”

    It’s pretty well known that commodity futures contracts are a game of hot potato for most investors as the expiration date approaches. But the Interactive Brokers CEO doesn’t offer an alternative solution. How would the contracts be structured instead that would avoid this? I don’t see how it would be possible.

  • by AndrewBissell on 5/8/20, 5:31 PM

    Incorrectly assuming values can never be negative is an all-too-common occurrence in trading and financial software. In 2012 Swedish stock futures trading was suspended for a time because their matching engine used an unsigned type for order quantities and someone submitted an order with a negative value which wrapped around to 4 billion: https://www.reuters.com/article/markets-sweden-bug/swedish-s...

    Interactive Brokers' software is usually very solid. I'm surprised they weren't ready to handle this, the possibility of oil going negative had been discussed for some time before it happened.

  • by mathgenius on 5/8/20, 7:23 PM

    If you look at the CME website using the wayback machine, you can clearly see a hi and low limit for the price. There is no high limit, but the low limit is set at "0.01". [1] This is kind of weird, because the snapshot is pulling current market prices, and still showing an incorrect hi/lo limit. It looks like they have since removed the hi/lo display: [2]. Someone should take a screenshot of this.

    [1] https://web.archive.org/web/20200117115242/https://www.cmegr...

    [2] https://www.cmegroup.com/trading/energy/crude-oil/light-swee...

  • by nromiun on 5/8/20, 7:48 PM

    > Its software couldn’t cope with that pesky minus sign, even though it was always technically possible -- though this was an outlandish idea before the pandemic -- for the crude market to go upside down.

    Wow, just wow. They are handling millions (billions?) of dollars every day and couldn't find the time to test that they can just DISPLAY a minus sign. That's insane.

    And it's not even that outlandish. People were saying it could go into the negative weeks before it happened. This just seems like pure laziness. Just pretend everything is business as usual.

  • by codenesium on 5/8/20, 6:04 PM

    Wasn't the original purpose of futures to let farmers and others lock in prices early so they can mitigate risk? Speculation on futures seems dumb if you have no intention of taking delivery.

  • by flint on 5/8/20, 6:15 PM

    I'm surprised IB let speculators trade in a contract going to delivery. I worked as a risk manager in a commodity trading firm and only hedgers qualified to take delivery were permitted to hold contracts going to delivery.

  • by socrates1998 on 5/8/20, 6:00 PM

    If you don't know what is going on, then why on earth would you risk so much money?

    IB fucked up, no doubt, but these idiots are trading shit they know nothing about.

    Don't trade on margin.

  • by 0x8BADF00D on 5/8/20, 5:49 PM

    It's really odd that this bug occured, as IB has no issue pricing credit spreads with negative values. Must be an issue specific to commodities futures contracts. I wonder what data types they were using.

  • by supernova87a on 5/8/20, 6:58 PM

    For futures on physical deliverable objects (well, I guess most futures are such, but anyway) -- would volatility and speculation be dampened/improved if the clearinghouse forced everyone (or the seller) participating in a trade to certify that they had rights to the specific thing being traded? Could actually produce the contract -- like the oil producer is certified to have <xyz> barrels allowed to be sold?

    My notion is that if much of the trading (and it can be shown by futures volumes) cannot possibly be on actual physically deliverable quantities, then most must be "speculation" by people who cannot actually produce the asset. Would this be a help to stabilize the market?

    I know it all has to get settled in the end by the expiration date, but just an idea.

  • by LatteLazy on 5/8/20, 7:27 PM

    Fuck IB and this “trader”, idiots should lose their shirts, that’s one of the intended outcomes of an efficient market.

    That said, there is an issue here with futures contracts: you can get very very large leverage when the price is near zero. This is the real issue with instruments that can negative price and just like their are “circuit breakers” in markets for big price swings, there should be breakers for entering the “near zero” range.

  • by danans on 5/8/20, 8:05 PM

    Hopefully, one of the things that we price in going forward is the volatility in the price of oil vs alternatives, especially those that can be produced domestically in a way decoupled from international events. That includes natural gas and especially renewables. Remember that just a 6 years ago, oil was north of $100/barrel [1], and recently it's close to zero. Wind on the other hand has had a steadily reducing LCOE[2][3]. Solar's LCOE is also steadily reducing [3].

    Volatility has a cost. With oil, it's one that the US and other countries hae historically tried to dampen with various industrial and political methods (the national strategic oil reserve, military/political "influence" on foreign oil producers, subsidies for domestic production), but seems like the current situation is beyond those methods' ability to control.

    1. https://www.macrotrends.net/1369/crude-oil-price-history-cha...

    2. https://www.energy.gov/sites/prod/files/2015/08/f25/LCOE.pdf

    3.https://en.wikipedia.org/wiki/Cost_of_electricity_by_source#...

    4. https://en.wikipedia.org/wiki/Cost_of_electricity_by_source#...

  • by kragen on 5/10/20, 1:01 AM

    I've written real-time trading systems trading on IB. Their client library is a real pile of shit, so it's not surprising to hear that their internal systems are too. Reassuring to hear that they're going to eat the US$100M loss themselves instead of letting their customers have it. (Not that they were going to get US$9M out of this Shah guy anyway.) They never did us any wrong when we were their customers.

    Does this pose a risk of IB going insolvent? If they do, is there a risk of their customers being just another creditor of a bankrupt corporation, with respect to the stocks and futures that IB holds on behalf of those customers? Or are those instruments held in bailment, or actually by some other company, rather than as IB assets?

  • by ralph84 on 5/8/20, 5:55 PM

    Would be nice to have more details on exactly what happened with the trades. Did the trades clear? Did IB liquidate the contracts before expiration? Was someone on the hook for taking physical delivery?

  • by Stierlitz on 5/9/20, 12:30 AM

    > Crude was actually around negative $3.70 a barrel when Shah’s screen had it at 1 cent

    A bit of a design flaw, I wonder did they spend as much money on the software as Equifax did?

    if (-3.70 != 1) { send an alert; }

  • by muyth on 5/8/20, 5:32 PM

    Would any of the traders mentioned that owe all that money to Interactive Broker's have to pay the total amount? Or is that part of Interactive Brokers' loss claim.

  • by raviolo on 5/8/20, 9:39 PM

    Comments about storage costs are nice theory but don’t tell me those storage costs went up 50 dollars a barrel to cause CLK0 to go from +10 to -40 within something like 30 minutes. It was all forced liquidation by brokers like IB and people absolutely bamboozled by negative prices puking their positions. Nothing to do with costs for actual storage.

  • by neonate on 5/8/20, 9:34 PM

  • by dilandau on 5/8/20, 6:20 PM

    More to the point is that a bunch of dumbasses were speculating on oil futures without understanding the nature of the market, and no ability to take delivery on the oil in any event. Nobody should be holding those when they're so close to expiry unless they know what they're doing.

  • by thanatropism on 5/8/20, 6:55 PM

    About two months ago I delivered to a private client a (monthly-basis) prediction model for ethanol prices that included Brent -- in logarithm -- as a predictor.

    I know the monthly spot average of that statistic (fetched from FRED, the Fed of St. Louis system) won't turn negative, but still...

  • by huy-nguyen on 5/9/20, 9:06 AM

    Another entry should be added to the “things programmers believe about commodity prices.”

  • by changoplatanero on 5/8/20, 5:33 PM

  • by tester89 on 5/9/20, 12:26 AM

    I don’t really understand why the traders would end up owing money. If they thought they were paying 0,01 $/bbl but we’re actually “paying” -37 $/bbl wouldn’t IB owe them?

  • by unixhero on 5/8/20, 9:47 PM

    I would recommend http://www.saxobank.com , it's a kick ass broker.

  • by SrslyJosh on 5/8/20, 5:45 PM

    I am playing my second tiniest violin for them right now.

  • by ninetyfurr on 5/8/20, 5:55 PM

    Oh a guy with $77,000 in his day trading account didn't know futures can go negative. I'm shocked.

    Ironically this is the same guy who will sell you a gym contract without a cancellation option and blame you for not doing your research when you owe him $1100.

  • by BubRoss on 5/8/20, 5:32 PM

    As usual, this nonsense headline is clickbait for the truth - some traders couldn't trade when oil went below zero.