Slashdot Mirror
Somebody Stole 7 Milliseconds From the Federal Reserve
An anonymous reader writes "Three to seven milliseconds before the fed moved interest rates, billions of dollars of trades were input that took advantage of the changed rates, reaping huge profits. According to a report at Mother Jones, 'Last Wednesday, the Fed announced that it would not be tapering its bond buying program. This news was released at precisely 2 pm in Washington 'as measured by the national atomic clock.' It takes 7 milliseconds for this information to get to Chicago. However, several huge orders that were based on the Fed's decision were placed on Chicago exchanges 2-3 milliseconds after 2 pm. How did this happen?'"
The Insider in this case would have the information well before it was announced in DC. He has the trades all setup and ready to execute, and then set the timer to have it happen at exactly 2PM. He forgot about the speed of light delay however and accidentally outed himself. After a decade or so the FTC might slap him with a couple of thousand dollar fine or something to make sure he never abuses insider information to make a billion dollars in a millisecond again.
I read the internet for the articles.
Of course there is a story here.
And of course it doesn't have to involve breaking the speed of light. It has to do with the Fed failing to properly enforce a supposedly very complex information lockdown and the information likely being either leaked or pre-loaded on remote servers, resulting in (according to the article) over $600M in trades via high-speed computer trading in the few milliseconds after the information was released - and now the Fed is investigating.
It's both related to technology and possibly involving criminal activity (or at the very lease a failure in information security). Sounds like a relevant story to me...
Another possibility is that it was a big player like Buffet or Sachs, counting on profiting one of two ways: a) precede the market b) bet wrong but cause the market to reverse, triggering massive losses to those who told their broker to trade according to the news, with a stop-loss protection.
To carry out B, you have to be able to place a second bid greater in magnitude but opposite sign to the first, about ten minutes after the news breaks. You also have the liquid assets to do it. since the bailouts doubled the national debt and gave it to banks in interest free loans with no set date of repayment, then yes, itmay be possible that investment banks did it.
Correct Horse Battery Staple: 72 bits of entropy. Enter "Correct H" into google. When it generates the phrase, that's
Correct. Now the question is: Can they be prosecuted for insider trading? This is an interesting situation where the speed of light may factor into the legality of their action.
Sure it's plausible. Atomic clocks. What the exploiter did was try to time the leak so that it happened exactly after the fed announced its decision before anyone else had a chance to react on it. They performed the trades at exactly 2:00 so that it all looked kosher. The problem was that they did it faster than what is physically possible and thus what was a plan to give plausible deniability backfired and thus exposed that they had insider knowledge.
Of course, this hack isn't even new. There's the value of delaying a radio broadcast was documented in the movie The Sting. And before that there was the Count of Monte Christo who did a man in the middle injection of fake tweets into France's Semaphore packet system to ruin a banker.
I would assert that High frequency trading is simply parasitic. Many people have suggested a transaction tax could fix this. It would damp the frivolous trades yes, but it would not fix these mega scale singleton trades that can happen like this.
my proposal: What one should do I think is fix this by injecting random delays into the trading system itself. That is you would queue up all trades for the last 100 milliseconds into a block. Then randomize their order. then execute the trades in that new order. This would erase any value of a trade that depended on beating another trade by a few millisconds. You'd still have some edge cases to worry about (i.e. racing to be in the block before the next block). But you could fix that too (dither the interval size between 80 and 120 milliseconds at random, so no one would know where the block boundaries were.
For this to be viable enough people would have to agree that HF trades have no actual value added. Additionally, one would have the potential problem of exchanges popping up that did not honor this. But those would not likely have enough liquidity to matter.
Some drink at the fountain of knowledge. Others just gargle.
The clocks weren't quite synchronized
Yes, it's so difficult to synchronize clocks these days. A GPS receiver will only get you a time reference accurate to within tens of nanoseconds.
"This is an interesting situation where the speed of light may factor into the legality of their action."
I don't see what's "interesting" about it. They broke the law. Physics proves it pretty clearly.
Washington to Chicago is 596 miles via a great circle, however the Earth's curvature will reduce that, but only by about a mile.
Light travels at 186 miles per second, thats 3.2ms
In the case of antipodes, you certainly see the effect
Auckland to Malaga, 12392 miles (67ms) as the great circle goes, but dig a hole through the earth and you can do it in under 8,000 miles (42.5ms)
Physicists will claim that an event occuring at 1400UTC in Auckland will not have occurred until 1400+42.5ms in Malaga, however there's no way for anyone in malaga to receive data until +67ms at the earliest. If I executed the trades at +50ms, technically it's happened. At +40ms, we have arguments about whether it's happened or not (an impartial observer who is equidistant from both points will agree that the rate changed, then my trade was executed). Even at -40ms there's no way for me to impact the event.
However on a more practical scale, as we can't encode data in neutrino bursts, the only way for a trade at +60ms in Malaga would be to have pre-knowledge of what happens in Aukland. But from a physics point of view, you could theoretically know.
So you've got the following key points
135959+933ms last time I can practically* do something in malaga to affect the auckland release
135959+957.5ms last time I can do something in malaga to affect the auckland release
140000+0 event occurs in Auckland
140000+42.5ms theoretically I could know about it
140000+67ms I could know about it
you do realize that 50% of all current trades last less than 100ms right? That is right folks 50% of the current volume is nothing but hot air. that is why the stock market has had 10-15% growth but the economy is barely doing 1-2%.
The fact this happened shouldn't surprise anyone. this is the problem of HFT.
i thought once I was found, but it was only a dream.