Backtest results for listed vs likely price

Discussion in 'Learn how to trade or invest by asking questions' started by BKaufman34, Jun 15, 2010.

  1. BKaufman34

    BKaufman34 New Member

    I have a an EOD trading system that backtests with good profit and
    smooth equity curve over the last 12 years. Average bars held per trade
    is 19. The system trades next day AtClose. Either ETF IWM or MDY is
    held for up signals or its inverse RWM or MYY for down signals.

    Backtesting results are for Yahoo listed historical closing price, but
    my real transaction could almost never be the last trade of the day.
    I could however consistently place orders say near 10 minutes before
    closing time. My orders are always "market", never "limit".

    I would appreciate opinions from those who know market internals, if
    there would be significant difference on-average between backtest
    results using the listed historical closing price, versus placing orders
    a few minutes before market close? I focus on CAGR%, maximum DD and
    Sharpe ratio.

    Here is same question for same system and ETFs if trading next day
    AtMarket. My chance of getting the first trade of the day is slim, but
    would difference between backtesting listed historical versus real
    execution price on average be small?

    Thanks, Barry
     
  2. LongArm

    LongArm Member

    You're right, it's likely that your backtesting prices and what would be your REAL entry prices are not going to be the same, especially at the beginning and end of the day. My feeling on it though is, sometimes the discrepancy will be in your favor, sometimes not--it should even out in the end (to some degree).
     
  3. BKaufman34

    BKaufman34 New Member

    What I hope is in my favor are:

    The average 19 trading days in the market. Slippage should cause less
    error as this # increases. I hesitate to trust backtested systems that
    are in the market only a couple of days.

    Hopefully high liquidity of the ETFs I use. They are based on the
    popular S&P 400 Mid Cap and Russell 2000 Small Cap indexes.
     
  4. Tradetowin

    Tradetowin Member

    Yes, but bear in mind that "past performance is not indicative of future results". since backtesting is using past data, you should allow some errors in actual trading which may or may not be small, especially if the market is volatile.

    Other factor would be slippage, actual trading is subject to this and can further increase the differences between backtesting and actual results.
     
Loading...

Share This Page