real-time backtesting

[koalboy]

Hi!

I created a strategy with two timeframes (primary is on 1-tick, secondary on 1 Min) on the same instrument what allows me to update the value of a dynamic stop-loss (takes its value from a custom indicator) every tick.

This is needed, because the StopLoss should take the maximum value of the indicator ever reached (calculated on every tick) since a position was entered.

In Market Replay environment the strategy behaves like expected without any problems.

But when I start a backtest with the same setup, I don't get any results and the log-tab shows me following message:

Failed to call mehtod "Initialize" for strategy "AATrailProb"; "Position" property can not be accessed from within "Initialize" method.

1. What does this mean?
2. In general: Is my logic right that I should get "near to live-trading"-behaviour of my strategy when I backtest this strategy calculating indicator values on a timeframe that should represent the live-trading-primary-bar-intervall (eg 1 Min) but letting the strategy run on a one-tick-primary-timeframe when backtesting?

I'm using the Position-method for resetting purposes within the strategy:
Here is my code:

 [INDENT]#region Initialize
protected override void Initialize()

{
Add(PeriodType.Minute, xminutes);


CalculateOnBarClose = true;

StopValueSCalc = new DataSeries(this);
StopValueLCalc = new DataSeries(this);
}
/// <summary>
/// Called on each bar update event (incoming tick)
/// </summary>
/// 
#endregion


#region OnBarUpdate

protected override void OnBarUpdate()

{ 
if (Historical)
return; 
if (CurrentBar < 10)
return;


if (BarsInProgress == 1)
{
StopValueSCalc.Set(ATRTrailingStop(10, 1, 20, 1).ATRTrailup[0]);
StopValueLCalc.Set(ATRTrailingStop(10, 1, 20, 1).ATRTraildown[0]);
ResetSLL = ATRTrailingStop(10, 1, 20, 1).ATRTraildown[0];
ResetSLS = ATRTrailingStop(10, 1, 20, 1).ATRTrailup[0];
}


if (BarsInProgress == 0)
{ 
#region LONG


if (Closes[0][0] > Highs[0][1])
{
EnterLong(DefaultQuantity, "ENTERLONG");
}

if (Positions[1].MarketPosition == MarketPosition.Long)
{
StopValueL = Math.Max(ResetSLL, MAX(StopValueLCalc, BarsSinceEntry())[0]);
SetStopLoss("ENTERLONG", CalculationMode.Price, StopValueL, false);
}
 

if (Positions[1].MarketPosition == MarketPosition.Flat)
{
SetStopLoss("ENTERLONG", CalculationMode.Price, ResetSLL, false);
}
 

#endregion

thx for help!!
[/INDENT]

[NinjaTrader_Dierk]

- please make sure you are on latest 6.5.0.8 or 6.0.1000.9
- somehow you access the Position property directly or indirectly in the Initialize method in your code. This is not allowed. Please revisit your code
- any backtest is different than the "real" thing. Please read more here why: http://www.ninjatrader-support.com/H...sBacktest.html

[koalboy]

Hi Dierk!

1.) I updated from 6.0.1000.7 to 6.0.1000.9 but I still get the same error in the log tab.

2.) you wrote
"- somehow you access the Position property directly or indirectly in the Initialize method in your code. This is not allowed."

If you take a look at my code, you can see, that I only access the values of the custom dataseries "StopValueSCalc" and "StopValueLCalc", which have to be defined within the Initialize-Method, havn't they?
Anyway I moved the DataSeries-definition to the beginning of the OnBarupdate-Section and changed the code in the Initialize-section to:

 
{
Add(PeriodType.Minute, 1);
 
 
CalculateOnBarClose = true;
 
 
}

and I also tried the variations with "Position" instead of "Positions[1]" but without any change of result.

I still get the same error message after starting the backtest-procedure. Does that mean, the Position-method cannot be used within a MTF-Strategy?

3.) Of course I already read the chapter in the NT-help-section and some other articles on backtesting-problems in general.

Historical tick-data is provided by my data-supplier so I try to come as close as possible to real-time-conditions (of course there are systemical restrictions).

The Market-Replay-Tool of NT is the only alternative to this, though I don't have enough replay data yet to get valid results. Also optimization of input parameters of a strategy can only be done on historical basis economically (instead of trying several set ups manually).

Would you agree that my approach of a 1-Tick-MTF-Strategy is right concerning that target or would you suggest something completely different? I'd be glad for just a little push in the right direction!

thx

[NinjaTrader_Dierk]

Could you please mail be your complete strategy and a screenshot of any settings needed to reproduce the problem to "dierk AT ninjatrader DOT com"? Thanks