backtesting futures

Discussion in 'Automated Trading' started by Kris, Jan 25, 2009.

  1. Kris


    As I mentioned in another post I am unfamiliar with trading futures, I've only ever traded equities and options. Now, when backtesting futures data, do I want to use the current contract or a continuous contract? I did some testing on ES H9 which is the March contract S&P emini I believe... is this correct?

    Also, are the emini's traded on a Level 2 screen like stocks? It looks like 1 contract on the ES nets $50 per point, is that right?
  2. This question i also would like to see answered as
    i am in the same position, new to future's; doubting about
    what to use for testing.

    only thing i can help you with is... ( you will find more information
    at CME's website about the specifications )
    1 ES point = 4 ticks
    1 tick = 12.5 usd
    makes 1 point 50 usd
  3. Kris


    Okay, that makes sense.

    Maybe someone else can help us out on the other points. What I'm wondering is if I can use the latest contract for backtesting or if I need to use previous contracts for backtesting previous months? ie for testing Nov/Dec 08 do I need to use the Dec contract or can I just use the March 09 contract?
  4. Surdo


    Use the "continuous" contract for backtesting ES.
  5. Hi all. I use amibroker for my charting. You should be able to use the current contract and within your testing/charting software be able to set the retention level/how far back in time you can look at that specific contract. In amibroker when you create a chart it asks how many bars and a certain amount of bars=certain amount of days. Hope this helps
  6. if you need long
    futures (continuous) history free available here

    how do you calculate commissions/spreads/slippage
    for ES ?

    in my testing i have used 5$ commission rt
    1 tick spread 1 tick slippage x 2 = 4 tick rt
    makes 55usd RT / contract

    would this be enough and close to what you are
    using in your calculation?
  7. Kris


    Yes, I'd like to know this as well.

    Also, futures is a 24 hour market? Or do the North American stock index futures trade at the same time as the index?
  8. As far as backtesting goes, the longer the time period you test over, the more reliably your system will work. So, obviously, continuous contract is going to provide a longer time period than a single contract.

    I would recommend testing on at least two years of data, to cover many different types of market conditions.