On Tue 2003-12-23T10:13:10 -0500, [EMAIL PROTECTED] hath writ:
> The whole point of arbitrage is to know (and act on) what other people
> don't know, or or don't know yet.  In the 1s-precision regime, people
> who have access to data that comes to them less than 1s faster than
> others don't gain any benefit from it.  With more accurate timestamping,
> more accurate reflection of the value of timely data is possible.

How likely is it that some sort of leap second arbitrage is an
unspoken motivating factor for discontinuation?

It would obviously be something that a system operator would not
want to advertise.

The interactions of the systems are the point to ponder, especially
given that there have been several different ways in which systems
handle the timestamps around leap seconds.  Basically, there have
been schemes like any of the following

UTC     oldBSD  newkernels      duplicate
59.00   59.00   59.00           59.00
59.20   59.10   59.20           59.20
59.40   59.20   59.40           59.40
59.60   59.30   59.60           59.60
59.80   59.40   59.80           59.80
60.00   59.50   00.00           59.00
60.20   59.60   00.00+eps       59.20
60.40   59.70   00.00+2*eps     59.40
60.60   59.80   00.00+3*eps     59.60
60.80   59.90   00.00+4*eps     59.80
00.00   00.00   00.00+5*eps     00.00
00.20   00.20   00.20           00.20

Does the system which actually duplicates timestamps permit the
appearance of knowing the future for the sake of a transaction?

There were several Jun/Jul leap seconds during the 1990s.
The Dec 31 leap seconds happen on Jan 1 in AustralAsia, and that is
likely to be a holiday, so no big problem happens.
But the Jun 30 leap seconds happen on Jul 1 in AustralAsia, and that
is a business day more often than not.
The Sydney market is definitely open, and the Tokyo markets seem to be
in the pre-trading phase at 0h UTC.
Obviously the situation grows worse as markets expand their trading
hours toward 24x7 operation.

Another question is, can you really move enough capital through the
Sydney markets in one second to make the effort worthwhile?

What would the response of the IERS be if some market executives
asked them to avoid June 30 leap seconds whenever July 1 was
a business day?

Steve Allen          UCO/Lick Observatory       Santa Cruz, CA 95064
[EMAIL PROTECTED]      Voice: +1 831 459 3046     http://www.ucolick.org/~sla
PGP: 1024/E46978C5   F6 78 D1 10 62 94 8F 2E    49 89 0E FE 26 B4 14 93

Reply via email to