Trailing Stops in Backtest & Live Trade

• 04/04/2016 at 10:03 PM #4864
  GraHal

  Hello

  Some 6 months ago I had email correspondence with IT-Finance re Trailing Stops and how Backtest results are nothing like the results one can expect in Live trading for the following reason …

  Backtest: 9 point Trailing Stop and 1 Hour Candle

  1. In ProBacktest the stop is triggered by a 9 point pull-back of the 1 hour candle only (ProBacktest does not look at each price tick within the 1 hour candle).

  Live Trading: 9 point Trailing Stop and 1 Hour Candle

  2. In Live Trading the stop is triggered by a 9 point pull-back of price ticks (as one would expect).

  As you can see (from difference in above 2 statements) that results from Backtest will be significantly different from Live in relation to the price / value that the Stop is triggered and thus overall result.

  In my opinion, the difference is so significant that Stops may as well not be considered / used / Backtested at all.

  Does anybody know if the situation has changed / is likely to change and when that might be regards dates … to nearest year / quarter will be fine (3rd Qtr 2016, 4th Qtr 2016 etc).

  Thank You

  GraHal

  

  GraHal

  Participant

  Topics: 211

  Replies: 8773

  Been thanked: 2179 times
  04/05/2016 at 9:44 AM #4876
  GraHal

  I thought of a bit more info to elaborate / make the point.

  So let’s say both scenarios START at TIME = 120000

  Backtest: 9 point Trailing Stop and 1 Hour Candle

  At the end of a 1 hour candle the system asks … “was there a 9 point pullback during this 1 hour cancdle”? Answer Yes so exit the trade … Time = 130000.  You may actually have pulled back 109 points during that same 9 hour candle but no check by the system is done until the end of the candle at Time = 130000.

  Automatic Live Trading: 9 point Trailing Stop and 1 Hour Candle

  2 minutes into the trade there is a 9 point pullback so the system exits the trade … Time = 120200.

  Hope that makes it clearer?

  I did a lot of testing 6 months ago (drove me mad! ha) and can’t bring myself to repeat it so if ProRealTime BackTest has changed then please let me know?

  Thanks

  Graham

  

  GraHal

  Participant

  Topics: 211

  Replies: 8773

  Been thanked: 2179 times
  04/16/2016 at 3:24 AM #5478
  Stef

  Hi @GraHal,

  This is called Look-Inside-Bar Back-testing and is supported by platforms such as TradeStation.

  Here is a description from their help file:

  Back-Testing Resolution

  1. Place a check mark next to Look-Inside-Bar Back-testing to enable the use of a sub-interval to identify price action occurring within each charted strategy bar.

  Look-Inside-Bar-Backtesting has no effect on real-time data or strategy automation; it is a feature of TradeStation’s strategy backtesting only. It is designed to report more accurate backtesting results; it allows a strategy to examine a finer data interval than appears in your chart so as to more carefully reconstruct the historical trades as they might have occurred in real time. It is not necessary to use Intrabar Order Generation to take advantage of the improved back-test accuracy afforded by Look-Inside-Bar Backtesting. For more information on Intrabar Order Generation, see About Intrabar Order Generation or Format Strategy – Calculation.

  If checked, the back-testing resolution is automatically selected based on the type of chart being used.

  For weekly and monthly charts, Day is selected

  For daily and intraday charts:

  If you select Tick you’ll be able to identify price action occurring within each charted strategy bar at the tick level.

  If you select Intraday you change the number of minutes for each sub-interval used for back-testing.

  2. Enter a value in the Maximum number of bars study will reference box to change the number of bars to reference.
  3. Click OK when you have completed your changes or go on to the next properties section that you want to format.

   

  

  Stef

  Participant

  Topics: 20

  Replies: 52

  Been thanked: 15 times
• Author
  Posts