Problem moving OpenQuant strategy to NT

[bluelou]

I've recoded a strategy that I've used in both TradeStation and OpenQuant for use in NT. I haven't added anything fancy yet, no DLLs or anything truly proprietary. Basically, it's just a moving average crossover strategy in NT at this point.

It's about 1000 lines of code and will probably pick up another 500-700 lines by the time I'm done.

Anyway, when I try to backtest it on tick data, say, 3 months worth for instance, memory and VM blow up to >1,500,000 kb each and then after a few minutes the program crashes. I was able to run this in about 2 minutes in OpenQuant.

However, the strategy will backtest in NT on 1-minute bars. It takes a few minutes, but it runs. But, since it's a tick-bar based strategy, this is of no use to me.

Can someone from NT take a quick look at my code and tell me what I can do to get the strategy to run properly? Or, is this what working with tick data is like in NT?

Thx,
BlueLou

[NinjaTrader_Dierk]

That's the issue where Ray and I mentioned that you probably will hit the wall just on RAM excess. Do double check that it's not your strategy you could try running SampleMACrossOver using the same interval/period settings.

[bluelou]

Dierk,
I think you may be right. This is a real disappointment. There's alot to like about NT. Especially a help forum staffed with qualified help. I'm hoping NT memory management is vastly improved with NT7. 3/4ths of my code is converted for NT but the memory management is a deal breaker. I think I'm back to using OpenQuant for trading/backtesting and TradeStation for charting. Ugh.

Thank you for your help.

-Lou

[NinjaTrader_Dierk]

- to get a better understanding you could compare to SampleMACrossOver
- what particular interval settings are you using on which instrument?

[bluelou]

Dierk,
I'm testing on 6E 09-08 with 300 tick bars. I've tried testing using anywhere from 1 month to 3 months of data history but the length of the data history doesn't seem to matter, NT freezes or crashes regardless. I'll try the MA cross-over tomorrow.

In this particular case I think it may be something about the strategy that isn't agreeing with NT. Though, I have no doubt that memory management will be a problem down the road.

It's 1:45 AM here in New York and I'm done for the night. I'll let you know tomorrow what comes of the MA cross-over test.

Thx,
Lou

[NinjaTrader_Dierk]

My guts feeling tells me 300 ticks should be doable for a few months -> I suspect there is something in your strategy code. I suggest
a) cross checking with SampleMACrossOver
b) isolating by eliminating