best way to validate optimizations

[edstaffin]

Hi,
I am using range bars to do my optimizations. I also am running my strat on a sim account to see what it does in the "real" world. I am finding that the 2 results don't always match up enough for comfort. Sometimes one or the other takes a different trade or no trade at all. Sometimes the entry times and exit times are not that close (read several minutes or more apart).

My question is, is there a best practice method for validating an optimization? Should I be using a minute data series to optimize and compare it to a minute dataseries run on the sim account for the same time frame? I need to have confidence in my optimizations and if I can't match up *exactly* or very close to exactly, I get skittish. I am anxious to turn my strat loose and suck up a few dollars, but I need to know that what I see in my optimization is the same as what I see in the sim.

So, for example, if I run my back test for 3/1/16 to today and compare it to the results of my strat running for the same period. I would like to see very close results.
BTW, I am running 1 contract for one instrument for one direction (long or short) in one account. So that means that I have to separate strats in 2 separate accounts ... one long and one short. This keeps things nice and simple.

Inquiring minds want to know ... how best to get to the promised land!

Thanks ... Ed

[NinjaTrader_PatrickH]

Hello edstaffin,

Thank you for your post.

While there is no guarantee that historical performance will match a backtest run there are items you could look into to help improve the consistency.
Try using the High Fill Resolution in the Strategy Analyzer. Play with the different bar types and values.
In the end the backtest will not be able to simulate what live executions did 100%.

For information on Fill Resolution please visit the following link: http://ninjatrader.com/support/helpG...ical_fill_.htm

Please let me know if you have any questions.

[edstaffin]

Hi Patrick,
Thanks for the quick reply. I started to look at this, but got stopped because the indicator that I am basing my strat on requires me to add a data series and when I try to do a backtest or optimization I get a message saying that the High order fill res is only available for single series strats and to please program directly into the strat the more granular res I want to simulate order fills with.

I don't know where first base is in this case. Is there some sample code laying around or some doc you can point me to that would get me off the dime?
Thanks ... Ed

[NinjaTrader_PatrickH]

Hello edstaffin,

Thank you for your response.

So in this case you can add the needed series to your code to submit orders to a more granular series for executions in backtesting. You would need one for each of the instruments in your script.

You can find details at the following link: http://ninjatrader.com/support/forum...ead.php?t=6652

Please let me know if you have any questions.