You know what’s the same distance away from the B as the M? The T. I’m not entirely sure what Proctor & Gamble would do having sold a trillion shares.
Sounds like this computer system could use Stocky, the Microsoft Stock mascot. “It looks as if you’re trying to sell your entire company. Can I be of assistance?”
I’m trying to imagine an interface where this could happen. Were they trying to enter x million shares? x million dollars? “Billion” is actually an acceptable option in this interface? Order of magnitude is determined by typing a single letter? There’s no “Are you sure?” type confirmation screen for transactions of this size? There are no automatic alerts triggered anywhere for humongous transactions that probably should go through several levels of approval?
It’s been stated that this is an unconfirmed rumor, and I have to say that I am skeptical. I’m not saying that it’s impossible, but it just doesn’t make much sense to me.
**Important Caveat: At present this isn’t looking like it was caused by a fat finger. ** Precisely because, as you say, this sort of thing is pretty easy to stop.
Short answer: There will be risk management features to stop this kind of thing but very often they’re turned off or misconfigured.
All modern trading systems that I am familiar will have an “are you sure?” type question for a large order, but being human he’s going to acknowledge it unthinkingly unless he wasn’t expecting it to appear. And that’s assuming that he wasn’t in the habit of just clicking go ahead as it was set incorrectly.
It’s perfectly possible to introduce all kinds of systems that will prevent ridiculously large positions; however they’re only appropriate for relatively simple types of trading (“300 yours… 300 mine” kind of thing) - say a local in a chicago proprietary trading firm trading t-notes on globex may have a dv01 limit or even just a limit of x lots, which will be enforced by the software, making it impossible for him to put any more size on.
Of course it can be better than this; say an options market maker will have limits in terms of the greeks and the system can ensure that he stays within them. Others I’ve seen use Var/Span etc to adjust limits dynamically - all this kind of stuff is easy to automate so long as one with dealing with simple exchange traded insturments.
(And, talking of automation, most black boxes have far more sophisticated risk management than that; for exmaple they will shut themselves off in extreme volatilty; in fact that will turn out to be the main reason the dow printed 998 down - said boxes will have pulled all bids and possibly even gone to market to get out. They are the only possible explanation for one beer company trading 0 while another trades $100 000 btw; they will have been long the 0 beer company and short the 100k one. Even if they themselves didn’t drive the prices to the extremes (and I doubt they did) the crucial point is they drove prices enough to set off stops (a market distorter as old as the hills) and the same time as almost all liquidity was withdrawn in a “wtf” moment. It is almost certainly black boxes that caused things to get as bad as they were; happened over the course of about 2 minutes, and the effects would have been cascading as said boxes attempted to take risk off along with everyone else, the very boxes that so much of the time are providing the liquidity)
Anyway, back to risk… banks don’t actually do all that much trading like that any more, well, actually they do, but it is fairly bread and butter low margin stuff, you may be surprised to learn… for them the money is in more complex structured instruments, with varying degrees OTC … and in that case getting a computer to work out everything on the go is pretty much impossible - but it is essential that the trader have flexibility to keep his book balanced. They are extraordinarily clever folk, these guys, in the book smart sense. As opposed to the street smarts of a lairy essex guy at tullets. They have to be trusted to get things right. And because the computer can’t always keep track they will be trusted to get things right (of course they do have risk limits, regularly monitored, but not automatically enforced… you can think of the risk as being in kind of concentric circles - each trader will have a limit, each team will, each desk will, each depatment will and the bank as a whole will - and due to miraculous correlations between different assets the risk of the bank will be considerably smaller than the sum of all the traders - and all this is reviewed regularly)
That said… The notional value of the alleged trade here was utterly ridiculous. There is nothing that could possibly justify going to market with size that large*, because as demonstrated the market just won’t take it and, needless to say, there are no products where all that size would have to be put on in one go. Can’t remember where S&P was before the crash, but say around 1150. To hedge $16 billion worth, they would have to sell over 250 000 lots in the e-mini S&P (where allegedly the initial fat finger was, although no one seems to be sure yet).
There would almost certainly be safeguards against that, which is why the story doesn’t sound right.
*Save, say, a nuke attack on washington or whatever, but whatever he was trying to hedge should not have a delta of 1 with the SnP… otherwise he just has an SnP position, and he won’t be allowed one that large unless he’s head of prop at Goldman - it’s pure speculation…
And in fact there was never any evidence that this was the cause. It was a supposition made by somebody which was promptly picked up (and then repeated over and over) by the news media because it made for a great story. Now that everyone is actually looking for a manually-entered order of this magnitude nobody can find one.
Think of this as the news media equivalent of programmed-trading.
