Back Testing vs Real Trading

Discussion in 'Strategy Building' started by dan05, Aug 4, 2006.

  1. dan05



    I'm testing a system, that provides predictions based on the continuous series of the Indexes. I'm back testing the trades provided by the system, with the e-minis future contracts.

    For each day, and each market, the system gives the entry time, and exit time.

    To compute the P&L of the trades, I'm using a tick by tick database of the future.

    The entries so far are market orders, at the specific entry time.

    I've noticed that on the tick by tick database, I can have for a single minute 6 different prices. Bid, Asks Settles.

    If I'm modeling a Market Order, what would be the most real price to choose for the entry and Exit?

    I thought of 3 alternatives.
    1. If going LONG at the Market, pick the FIRST ASK Price.
    2. If going SHORT at the Market, pick the FIRST BID Price.
    3. FIRST selltled or traded price. (Either Long or Short)
    4. The most traded price for that minute.

    Anyone had the same dilema? Can someone help me with this petit detail?

    Thank you
  2. I am subscribing as I am new to backtesting automation and would like to learn how.
  3. When buying use the first ask price+1 tick slippage on liquid markets. Opposite for selling. You may want to increase the number of ticks slippage in less liquid markets.
  4. dan05


    Great Thank you for your help.

  5. You don't predict.

    You assume that the market doesn't change.
  6. dan05


    Hi TSGannGalt,

    I used to think that till I found this system.

    This company seems to be predicting the markets.

    Check this.

    The black line, is what the system predicted for the day. The red line is what really happened.

    The point where the lines split, is the time of the day when the prediction was computed.

    It looks to me as if they are really predicting. And I've been following the system for some months, and results where really good. Now I'm running some further back tests, to find out more about the system So far, so good.