backtesting

Discussion in 'Strategy Development' started by Kris, Oct 10, 2008.

  1. Kris

    Kris

    I've never done any sort of backtesting, could someone point me in the right direction to get started? Is tradestation or trade-ideas a good place to start?
     
  2. I've played with Tradestation, Multicharts, Ninjatrader and Openquant. They are all great in their own right but I had one experience that was an eye opener. I tested a system by looking at charts and noting the trades, I tested the same system with Ninjatrader's backtester and also with Openquant's backtester and I got three entirely different results. I never could account for that but it changed the way I do testing. I do some manual testing, then I write an indicator [not a strategy] that performs the same work I did manually, then I proceed to a strategy that I can backtest [currently that is done in Openquant] and I also run it forward in realtime against a simulated account and I compare all the results. Anything less than that is not satisfactory to me considering that the backtesting industry is sort of in a stage of infancy and you can't really predict slippage and missed trades in a backtest...
     
  3. I remember using Excel spreadsheets at first. Now I write my own c++ programs. There are many commercial trade simulator software programs available and some companies have free trial offers. I find back testing useful for optimizing the risk level. I recall for years I was trading too small. If the risk is too small then winning trades do not show enough profit to outperform a money market fund.
     
  4. Kris

    Kris

    Okay, now where do you guys get your data feed from? Which feed source offers the most intraday historical data for backtesting? I'm looking to backtest day trading/scalping methods.
     
  5. Opentick and IB. Opentick is free but it's closed to new accounts currently or was the last time I looked.
     
  6. Kris

    Kris

    Anyone use esignal?
     
  7. Techy

    Techy

    You can write Expert Adviser in MetaTrader4 and then backtest it in historical data. If you know C++ programming language it will be easy for you. If you see that backtesting results are good you can use the Expert you wrote for regular trades without your interaction.