ROI vs low latency infrastructure costs in HFT

(Last Updated On: May 23, 2012)

ROI vs low latency infrastructure costs in HFT

Any studies, views or information on evaluation of ROI vs latency reduction would be appreciated.

i.e. Spend a million pounds to reduce execution latency by 2 milliseconds with expected returns of ????



All depends on the fail overs and DR procedures you have within your infrastructure already. How fast can your connections roll over upon failure and how long will it take for your system to be back on online and trading. This is the simplest form of working this out. Calculate those times and compare with lost business within your company to find out the RO, and if you can afford to carry on with your current system or need to new investments.


I think you missed the point with all the resilience discussion.
I think the ROI depends on exactly how the latency reduction gains are being leveraged, which is highly variable and subject to a wide range of factors, however I am sure there are a large number of Quants and Statisticians in this group who are standing by to tell us otherwise.



I’ve searched for this kind of studies, but found nothing interesting. This isn’t surprising since results are (very) highly dependent on the underlying strategies… That being said, why not study it yourself on your strategies ? What you can do is to try to estimate it by analyzing your own orders in the market data and then guess what would be the difference (how many more executed or canceled before executed etc) if your orders arrived x microseconds faster. This is just an estimation, since the market data is always biased by latency on exchange (and potentially yours) infrastructure, but based on my experience, you can get a quite good estimation.



In 2008 The research company Tabb Group publishd a report on the true commercial cost of poor latency management. They estimate that if a broker’s electronic trading platform is 5 milliseconds behind the competition, it could lose at least 1% of its flow – an estimated $4 million in revenues per millisecond. Up to 10 milliseconds of latency could result in a 10% drop in revenues.


Hope this helps.



we have those exact stats let me get my hands on them, it was shocking i know because we will spend anything now no questions asked. in fact it’s more about the additional opportunities you have the faster you are. mark



Thi is for support purposes, may we need some decimals places to ensure that order is on the agreement level, and can we sell that service, agrees ?
We can consider PTP boards and NTP(Stratum 0) to serve a better mili/nano second.
It will show the status of your real-time health also. Can you imagine when some costumer see that timestamp :

Performance for 1000 operationonds:
in milliseconds: 0
in nanoseconds: 144991
Performance for 1000000 operationonds:
in milliseconds: 62
in nanoseconds: 48716806


NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Subscribe For Latest Updates

Sign up to best of business news, informed analysis and opinions on what matters to you.
Invalid email address
We promise not to spam you. You can unsubscribe at any time.


Check NEW site on stock forex and ETF analysis and automation

Scroll to Top