how good is good enough?

Discussion in 'Strategy Development' started by otayah, Apr 29, 2011.

  1. otayah


    I've been backtesting and optimizing a simple strategy that averages a 0.55% gain per trade. It'll be an automated system making about 10 trades per day in 200-500 share blocks in mid-priced stocks with $0.01 spreads. Is a 0.55% gain per trade from backtesting good, mediocre, or useless in the real world?
    The backtesting doesn't account for slippage and the strategy mostly takes place at the open when things are moving quickly. What should i expect? I've read that $.07 of negative slippage on $50 stocks and $.03 for $20 stocks are average for filling market orders of this type at this time. I'll eventually experiment with limit orders but I'd rather use market orders to make sure that every possible trade is entered in case the hardest-to-enter trades are the most profitable. If $.07/.03 is the worst-case scenario, the strategy would still be decent if it lost that much on the fill and maybe half that much on the exit. Is this a reasonable expectation?
  2. Tell you nothing about the real world. Did you test it out of sample?
  3. Forward test it , slippage and commish are a bitch in the real world to be frank.

    Best wishes
  4. otayah


    Thanks for the replies.

    Yeah, I'll have to forward test it but I feel like the IB demo account probably isn't realistic and was looking for some reassurance before spending actual money.

    Related to the forward test and slippage:
    1. Could I test it with just 20-50 share blocks instead of 200-500 share blocks, or will execution of these amounts not replicate at the higher 100-block amounts?
    2. If I eventually wanted to keep each trade in equal $10k amounts, will it introduce lots of extra slippage to have odd lots? Say I program it to buy $10k of a stock that translates into 268 shares. Will that mean the 68 shares take a longer time to get filled/sold and introduce additional slippage, or do they trade just as quickly or more quickly than rounded 100-share blocks?