Measuring Execution Speed

Discussion in 'Order Execution' started by OptionsWizard, Apr 13, 2006.

  1. I finally figured out the way to ping IB servers and the roundtrip delay is about 85ms.
    So the time IB took to execute the orders was 25ms-105ms!

    Way to go IB!
     
    #11     May 6, 2006
  2. kjsnow25

    kjsnow25

    you guys are aware there are brokers out there that speciaize in low latency pipes, and that 100 milliseconds is basically a nightmare for their customers trading those strategies, right?

    There certainly is a way to verify latencies, and it's not just ping times. Beyond that, it's not just times for single orders, it's how fast as well as how fat the pipe is.

    There are a lot of ways to llok into latencies and throughputs.
     
    #12     May 9, 2006
  3. IMHO, lots of hot air.

    Would somebody participating in this thread give us an estimate of the percentage increase in his average daily profit under the assumption that his overall latency time would go down by 100ms.

    We'll resume our discussion from that point on.
    :D

    nononsense
     
    #13     May 9, 2006
  4. kjsnow25

    kjsnow25

    that's a bit goofy, because there are plenty of traders with a co-located box at brokerages clipping speeds of, let's say for argument's sake, round trip, with internal processing time for checks vs. easy to borrow lists, margin, capital, out to ECN, and back, at less than 15 ms to ARCA, less to INET (for example). To think that those guys don't have an inherent advantage....firing their algo out there? The algo is key, but execution speeds (if that's part of your strategy) are key.

    To isolate your "profit" as a percentage of speed is silly, but you all know you're in ther same ECN's with guys trading like this, right? It might be hot air to those that Don't trade for speed, but fo rthose that do (they'r eon the board) it does matter.
     
    #14     May 9, 2006
  5. ids

    ids

    We are talking about retail trading over Internet here. You are comparing apples to oranges.
     
    #15     May 9, 2006
  6. Well, that depends on your trading timeframe. If you trade daily swings, it's completely irrelevant.
    If you day trade the ES and average 4 ticks (1 point) per trade, than losing 1 tick on entry and exit is 50% of your profits.

    The numbers I measured, 100-200ms roundtrip, are in my comfort zone. I don't see any reason to look for better , but above, let's say 500ms, I'm sure I would start losing ticks on entry and exit.
     
    #16     May 9, 2006
  7. kjsnow25

    kjsnow25

    here's some news as well: retail traders do and can co-locate boxes at brokers. It's not an expensive proposition.

    I'm just pointing out who you are up against.
     
    #17     May 9, 2006
  8. sccz97

    sccz97


    as i trade forex around announcements it is absolultey essential that my overall latency is as small as possible and under 100ms can make a huge difference, and I'm not talking just a couple of basis points.
     
    #18     May 9, 2006
  9. Your figures seem to make sense to me. As a fast mini-index trader, I also would start to feel uncomfortable with 500+ ms.
     
    #19     May 9, 2006