Strange back-test results..

[heech]

Hi there,

This is probably going to be one of those issues that you guys won't be able to help me on... but I thought I'd throw it out there and know if you guys had any input.

I have two versions of code that I'm running on the same symbol. The results of the backtest (identical parameters) *look* the same, but aren't. Here's what I mean:

Backtest 1
1 IOC Back101 ProtectDown Long 1,000 13.19 13.11 10/30/2008 12:30:00 PM 10/30/2008 12:30:00 PM Buy Sell -0.0064442759666413253 -0.0064442759666413219 0 0.0064442759666413253 0 0 1
2 IOC Back101 ProtectDown Long 1,000 14.06 12.93 10/31/2008 8:00:00 AM 11/6/2008 7:00:00 AM Buy Sell -0.0803698435277382 -0.08629619404329103 0 0.0803698435277382 0.10170697012802288 0.18207681365576106 51
3 IOC Back101 ProtectDown Long 1,000 13.37 13.11 11/10/2008 8:30:00 AM 11/10/2008 8:30:00 AM Buy Sell -0.019820493642483082 -0.10440625452036856 0 0.019820493642483082 0 0 1
4 IOC Back101 ProtectDown Long 1,000 13.85 13.26 11/10/2008 12:00:00 PM 11/11/2008 6:40:00 AM Buy Sell -0.042599277978339338 -0.14255790143971747 0 0.043321299638989147 0 0 4

Backtest #2
1 IOC Back101 ProtectDown Long 1,000 13.19 13.11 10/30/2008 12:30:00 PM 10/30/2008 12:30:00 PM Buy Stop loss -0.0060652009097800072 -0.0060652009097800219 0 0.0060652009097800072 0 0 1
2 IOC Back101 ProtectDown Long 1,000 14.06 12.93 10/31/2008 8:00:00 AM 11/6/2008 7:00:00 AM Buy Sell -0.0803698435277382 -0.0859475851894349 0 0.0803698435277382 0.10170697012802288 0.18207681365576106 51
3 IOC Back101 ProtectDown Long 1,000 13.37 13.11 11/10/2008 8:30:00 AM 11/10/2008 8:30:00 AM Buy Stop loss -0.019446522064322964 -0.10372272564199625 0 0.019446522064322964 0 0 1
4 IOC Back101 ProtectDown Long 1,000 13.85 13.26 11/10/2008 12:00:00 PM 11/11/2008 6:40:00 AM Buy Sell -0.042599277978339338 -0.14190349039804107 0 0.043321299638989147 0 0 4

The part that's absolutely confusing me is why the "profit" column are slightly different in these two executions (0.006444.. in backtest 1 versus 0.00606.. in backtest 2), when the price/timing of entry/exit are exactly/absolutely the same.

This is repeatable, and tied to my two versions of code. The changes between the two versions are really minor (nothing obvious, like commissions or slippage or...)... there's one of the two that I prefer, but I can't figure out where in my logic I'm leading to this difference in "profit".

[NinjaTrader_Bertrand]

Hi heech,

Hard to judge, there could be multiple scenarios at play...

1. Make sure you test both versions under the exact same settings and conditions.

2. Use the Print() command to debug and see if your trading logic hits home, see this thread here, too - http://www.ninjatrader-support2.com/...ead.php?t=3418

3. Make sure you test with the same fill algorithm - http://www.ninjatrader-support.com/H...AStrategy.html
See under 'Historical Fill processing'

Have a good Thanksgiving!

[heech]

Hi Bertrand,

I think I decided what the issue is. In both cases, the difference came when a stop was hit in the same bar as the entry. I think I must've very subtly changed the timing (in the backtest) of when the stop loss is placed... perhaps not even something on the scale of many minutes? 30 seconds? Who knows.

Thanks for looking at it.