On May 28, 2013, about 1/4 second before the expected release of the Consumer Confidence number, trading
exploded in SPY, the eMini and hundreds of other stocks. Even more interesting, activity
exploded just 1 millisecond earlier in the futures (traded in Chicago) than stocks (traded
in NYC). The speed of light separates information between Chicago and NYC by at least
4 or 5 milliseconds. Which means this was more likely the result of a timed trade in
both futures and stocks, rather than an arbitrage reaction between the two.
When the ISM Manufacturing number was released a week later, SPY traded 15 milliseconds early, but
the eMini lagged by about 7 milliseconds.
We found no other instances of early trading in the 11 previous monthly releases of
the same Consumer Confidence data.
A person familiar with the matter told us: The Conference Board said the information
was released to the newswires in NYC at exactly 10am, not a nanosecond before.
When we looked at trades in IWM, another major index ETF, we found that there was a
second surge of trading at 9:59:59.964, just 36
milliseconds before the official 10am news release. Plotting trades from all (about 8000)
NMS Stocks (chart 1 below) confirms this second
surge.
Update: June 20, 2013
We now believe the Consumer Confidence number was given early to several news organizations.
Then as 10 am approached:
One of those news organizations released the number simultaneously in NYC and Chicago
at 09:59:59.754.
Another news organization released the number at 09:59:59.964 (210 milliseconds
after the first news organization released the same data).
Responsible news organizations released the numbers at 10:00:00.
It is possible that a high frequency trading firm received the Consumer
Confidence number in advance and traded simultaneously in NYC and Chicago at
09:59:59.754. This
is a likely
scenario, because few news organizations have the technology and
experience to do timed releases in multiple locations.
Furthermore, because the release at 09:59:59.754 hit both NYC and Chicago within a millisecond,
it is unlikely this early release was caused by a clock synchronization
problem. Because for that to happen, two different clocks running at two different
locations (unless they were linked to a private master clock)
would have to be off by the same amount of time: a whopping 246 milliseconds.
That leaves us with the possibility that either a news organization consciously or negligently released
the information early, or someone gave it to a high frequency trading firm capable
of executing simultaneously in NYC and Chicago.
Update: June 21, 2013
Dow Jones has admitted the clock they used to release this news report
was not properly synchronized and caused the release to go out 1/4 second early. This
would indicate the first surge was from Dow Jones. For this to happen and fit
the known facts, a master clock at Dow Jones that updated servers in NYC and Chicago
would have had to be off by 246 milliseconds.
1. Trades from all NMS Stocks priced below $180 over one second of time.
Here's a video showing quoting and trading in SPY for 1/2 second - just before 10:00:00. Activity explodes at 9:59:59.755. (Or
watch it on youtube) 2. IWM - Trades color coded by reporting exchange and NBBO. One Second of time.
The second surge became apparent when looking at the symbol IWM - an ETF based on the Russell 2000 index.
3. SPY - Trades color coded by reporting exchange and NBBO. One Second of time. 4. June 2013 eMini Futures trades and quote spread. One Second of time. 3b. SPY - Trades color coded by reporting exchange and NBBO. About 1/10th second
of time.
Trade activity starts at 09:59:59.755 3c. SPY - Bids and asks color coded by reporting exchange and NBBO. About 1/10th
second of time.
Quote activity starts at 09:59:59.754 - 1 millisecond earlier than trades. 4b. June 2013 eMini Futures trades and quote spread. About 1/10th second of
time.
eMini futures trading explodes at 09:59:59.753 - just 1 millisecond before activity
in stocks. Nanex Research
