Progress - past, present and future
32 Best Execution Guide to Low Latency 2010
LOWER LATENCY TRADING
We are now past the first anniversary
of the latest black September and we
are still in business! In fact, quite the
contrary, volumes are up, shape sizes
down, and the proof is that, after a
short interlude of potential meltdown,
the markets are still with us and they
will trade going up or down.
For the trading market it could be
argued that the last 18 months have
done us a service. 2006-2008 was the
latency arms race. The learning curves
across the market, from technology
vendor to buy, sell-side and new
liquidity providers were colossal as
we all invented new business models
on the fly. As we sit looking forward
to 2010, it looks like rationalisation
has set in. We know who needs low
latency, how it’s affected and the
remedial causes. So let’s see how we
have arrived at today, what we now
have and what we might usefully do
with it next year and beyond.
We have come through a phase
of almost unprecedented hype in
technology marketing around low
latency. This has often been at the
expense of understanding where the
causes of latency are likely to exist,
and the practical options to cure the
symptoms – from tuning the engines to
wholesale replacement of aging parts.
In this context the task is multi
dimensional. One has to look across
the spectrum of processes that make
up the pre-trade life cycle and the
supporting technologies at each phase,
from receipt of market data and the
firing of the starting gun, through
intervening analyses and checks,
to the trade instruction hitting the
execution venue. Technically a variety
of components interoperate to enable
this, which is where the leading edge
computer engineering skills exist.
The Formula 1 racing analogy plays
amazingly well, where fast is never
fast enough. Here CAPEX (capital
expenditure) is not an issue and only
the best and latest components are
used, if necessary custom made, and
simulated on massive supercomputers
by a singularly focussed technical team
working to get the car round the circuit
fractions of a second faster than the
nearest rival.
Progress – past,
present and future
Jonathon Traer-Clark of Hewlett-Packard,
Nigel Woodward of Intel, and Jerry Brunton of BT