The Gateway to Algorithmic and Automated Trading

Automated Trader Magazine Issue 04 January 2007

Latency has inevitably become a hot issue in the auto/algo trading world, with some of the more excitable marketing hyperbole already talking in sub-microseconds terms. So far much of the chat has focused on network latency – in general terms, how to minimise the time between sending an order to a trading venue and it getting there. The snag is that this now looks like a case of rapidly diminishing returns; once everybody has bought the same ultra fast connectivity and/or put their execution servers as close as possible to the venue, a point of equilibrium is reached where nobody has an advantage.

That point has not escaped some of the interviewees in our feature on colocation (page 40) who have appreciated that while network latency gains may be increasingly hard to find, there is still scope for improving internal application latency. That point has also not been lost on vendors such as RTS Group, who have made it a major focus for their RTD Tango product (Spotlight, page 62).

Another theme in this issue is Asia. For a while we’ve been noticing an easterly migration of experienced sellside AES personnel, so we thought we’d better take a look at what they were getting up to in our feature “Algo in Asia” (page 78). While the general sense that emerges is that Asia is still playing algorithmic catch up, there are encouraging signs also borne out in the Exchange Views (page 32) interview with David Stocken of the ASX.

After several years of frenetic innovation, a third theme we have been picking up from buyside readers is algorithmic reality – i.e. what is really required to get the job done? According to our My Machine (page 84) interviewee, Clive Williams, education is near the top of the list. In Clive’s view, if you don’t fully understand how an algorithm works then you shouldn’t be using it. To judge by the responses in the AT Round Table (page 50) the sellside seem to be increasingly aware of this and accept that showing the buyside how existing tools can be put to best use is perhaps more worthwhile than blinding them with yet more snazzy algos.

Finally, Chris Donnan’s Strategies article (page 16) touches on something of increasing importance in the auto/algo world – optimisation. As the number and complexity of models continue to increase, so does the problem of optimising those models. Chris gives a thoroughly pragmatic and accessible outline of some possible solutions.

Andy Webb - Editor Issue 04

  • Copyright © Automated Trader Ltd 2019 - Strategies | Compliance | Technology

    Prepare failed: (144) Table './ATStatistics/logger_ads_imp' is marked as crashed and last (automatic?) repair failed