I found out the following;
When backtesting with 2 data series of the same symbol, data1 based on daily bars and data 2 based on 15 minute bars, the backtesting routines use the daily close during the day(in simulation). The problem is that this close value in reality is not know before the end of the day. Instead of using the daily close value for the daily bars it should use the close of the intraday bars for the daily bars. If you don't do it this way you get a big difference between backtesting and real trading.
serious backtesting problem
-
- Posts: 176
- Joined: 02 Feb 2008
- Location: Lausanne, Switzerland
- Has thanked: 9 times
- Been thanked: 5 times
Re: serious backtesting problem
You need to reverse your instruments.
Put the higher frequency timeframe as data1, and use intrabarpersist for all indicators that refer to data2 or above.
Put the higher frequency timeframe as data1, and use intrabarpersist for all indicators that refer to data2 or above.