Backtesting strategies with trailing stops

[betcashrun]

What is the cleanest way to use trailing stops in a strategy (and backtest it) given that:
- the strategy is based on the 5 minute Bars
- the function SetTrailStop() works only on a tick by tick basis
- the data provider only provides historical 1 minute data

I think the best is to use a multi-bars strategy (primary bars = 1 minute, secondary bars = 5 minutes) with a stop loss updated every minute to trail the profit.
Or do you think otherwise?

[NinjaTrader_JoshP]

That could be one approach that would work.

[betcashrun]

Thank you for your reply Josh. What are the other alternatives?

[NinjaTrader_JoshP]

You could just use SetTrailStop(). Whatever information it has it will use. In backtesting if you are on 5 min bars it will not have the full granularity of a 5min bar trail stop, but if the price is breached it will stop you out.

[betcashrun]

My very limited experience with NT shows that the use of this function with 5 minutes bars can imply inaccurate results. For instance, let's say we have this 5-minute bar OHLC = { 100;101;95;96} with a buy @ 96.5 et trailing stop of 2 full points. My backtest results have shown something like this:

Buy Limit @ 96.5 => (filled)
SetTrailStop(2.0) => (sell order filled @ 99)??? for a magical profit of 2.5

Due to the lack of information it's not possible to conclude whether the trailing stop could have been triggered at a price of 95 to 99 or simply not triggered.

That's why I think SetTrailStop() should only be used on a tick by tick basis as stated in the user guide.

[NinjaTrader_JoshP]

It can be used anywhere. Whether it is as accurate as possible is not going to be the case if you just don't have intrabar information.