Backtesting resolution

[PHOLAN]

I have done intensive reading behind your backtesting philosophy. And while it makes sense for the most part, it errs by ommission.

Yes, it's true that live results will differ from backtest results. However, although you can not know the bid/ask during backtest, you can always come down to the lowest possible timeframe in order to provide a more accurate result.

Let me give you an example, I have a TF 3min strategy with a $280 TP and a $190 SL. My backtesting results are completely off becuase in those cases where a long bar is encountered, NT cannot discern what came first between the OHLC and mmy entry gets tagged as a loss instead of a profit. And, although market replay is a nifty feature, it does not work for log term backtest of a strategy.

A more accurate approach would be for a smart backtest, in which it looks at the bar and, if there is an entry, then default to 1 min data (for example) to see what happened inside it. That way, the problem with the OHLC would be avoided and more accurate results would be provided.

[NinjaTrader_Jason]

Hello PHOLAN,

I will forward your suggestionto our development team and ask them if they can add this to the list of future considerations for the software.

Thank you for your feedback.

[NinjaTrader_Ray]

You can backtest intra-bar by utilizing a secondard more granular series. Here is a reference sample.

http://www.ninjatrader-support.com/vb/showthread.php?t=6652

[PHOLAN]

Thanks Ray,

I thought of that. Will try it.
Hopefully V7 will include this as part of it's built in backtest.

Regards,


Philip

[PHOLAN]

Ray,

I'm executing the logic on the primary bars (3min) and placing the trades on the secondary ones (1min). The graph, a 3 min bar, still shows the SetStopLoss on the 3 min bar incorrectly. The position enters a short on the 3 min bar and it closes at a stop loss on the same bar, when in fact, the High of the bar was prior to the order being placed.

P

[NinjaTrader_JoshP]

Not sure what you mean. Please see executions tab to see what your orders are doing. Also, it helps tremendously to open a second chart of your 1min and match up the time frames to see the additional granularity.

[PHOLAN]

The problem occurs when you look at historical trades on the 3 min bars when placing the trades in a secondary 1 min bars.
Basically, if I look at the trade in a 1 min chart, its a winning trade but, it shows as a stop loss in the 3 min chart. Even though I'm placing the trades in the 1 min chart.
It seems that the SetStopLoss is being executed on the 3 min chart instead of the 1min one.

[NinjaTrader_JoshP]

You need to look at the executions tab to see what or use TraceOrders = true to see what your orders are really doing. If you find this to be the case from the executions tab then your only option would be to not use SetStopLoss() and opt for an ExitLongStop() order.