Announcement

Collapse
No announcement yet.

Some number crunching: how much latency affects profit

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Some number crunching: how much latency affects profit

    This month I ran an experiment to answer the following question:

    Q: Is it worth me paying for a ForexVPS instance, compared with the free VPS I had access to?

    TL;DR: Yes. 80ms of latency appears to be costing me about 12 pips a week, or 4-6% of FXViper and SmartScalper profit, and 0.6% of SmartTrader profit.

    I have two VIP accounts (one for retail trading, one for my SMSF), and both with a similar balance, taking similar risk levels. Both use the same broker (IC Markets). One account is on server Live03, the other on Live04.

    This month, one VIP copier ran on a VPS leased by a friend (so I have free access) in Canada. Ping time to each IC Markets server is 79ms. The ForexVPS instance was 1ms away. Traceroute from the Canada VPS to each ICM server was identical. The traceroute from ForexVPS to each ICM server was different, but being ~ 1ms, I assume it makes no difference.

    I dumped all the trades from 1 May to 23 May to CSV and imported into Excel. Here are the crunched results.

    Trades running on ForexVPS averaged:
    * FXViper: 5.69 pips
    * SmartScalper: 9.33 pips
    * SmartTrader: 85.6 pips

    Trades running on Canadian VPS averaged:
    * FXViper: 5.44 pips
    * SmartScalper: 8.76 pips
    * SmartTrader: 85.0 pips

    The average of the differences (not the difference of the averages):
    * FXViper: 0.25 pips, stddev 0.95 pips
    * SmartScalper: 0.56 pips, stddev 0.99 pips
    * SmartTrader: 0.64 pips, stddev 1.34 pips

    As total pips, this is:
    * FXViper: 16.5 pips
    * SmartScalper: 9 pips
    * SmartTrader: 10.8 pips
    * total: 36.3 pips

    Remember this is across 15 trading days, so it's about 12 pips a week.

    As a proportion to the profits, therefore, using my own distant VPS cost me:
    * FXViper: 4% of profit
    * SmartScalper: 6% of profit
    * SmartTrader: 0.6% of profit

    So if US$40 is less than the sum of these proportions, then it's worth me paying for the VPS.

    Considerations:

    1. Perhaps it's not the latency. It might well be the difference in the SimpleTrader servers. Hard for me to test that.
    2. It might be some difference between the IC Markets servers. One might be more CPU-bound than the other, and that causes a latency in orders
    3. Might be statistically insignificant. What would I know.

    But I hope this helps scratch an itch that others might have in deciding on whether to pay for a VPS. For our fast-moving traders (Viper, Smart Scalper) it's significant.

  • #2
    Thanks for that analysis, it's really helpful.

    Have you done any study on the latency effects on profit between a broker server in NY and the SimpleTrader servers in the UK?

    Comment


    • #3
      Originally posted by saint023 View Post
      Thanks for that analysis, it's really helpful.

      Have you done any study on the latency effects on profit between a broker server in NY and the SimpleTrader servers in the UK?
      No. My research was purely motivated to optimise profit for my situation, but I wanted to share for the benefit of others. If you're going to pay for the VPS, why would you choose one further away from the servers you're talking to?

      Comment


      • #4
        Unfortunately there is no choice as SimpleTraders servers are in the UK only. There is discussion about having a server in NY also, however there is no action yet.

        The signal has to travel from the trader to the Simple Trader servers to our VPS/computer and then to the broker. It would be great if they were all in the one location.

        As most brokers seem to have their servers in NY, until there is a SimpleTrader server in NY also, we have to leap across the pond from the UK if choosing that VPS because it's where the SimpleTrader servers are based.

        I imagine that makes it tough on the traders who are scalping

        Comment

        Working...
        X