Code efficiency question

[RodS7]

If a complicated indicator is called from a number of strategies, would there be any positive or negative if that logic is imbedded directly into each strategy rather than calling it?

[NinjaTrader_Bertrand]

Hi Rod, would you call the same indicator method on the same instrument / timeframe and with exactly the same parameters?

[RodS7]

Same timeframe, same parameters, different instrument, but called multiple times in each strategy

[NinjaTrader_Bertrand]

Thanks Rod, then you would not be accessing a cached instance with your call, so I would not expect to see any harm on performance if you directly embed as method for example.

[fx.practic]

I tested 1 startegy that called indicator that called another indicator.
Strategy with imbedded both indicators calculations was 2.5 times faster.

[RodS7]

fx.practic

now that's definitive! Thanx

I am curious how you measured the speed....

Rod

[fx.practic]

Well, the methodology.

1) Write 3 strategies (attached):
sCodeEfficiencyTestSmaExternal - calls SMA(200)
sCodeEfficiencyTestSmaImbeddedFast - calculates internaly only current value of SMA(200)
sCodeEfficiencyTestSmaImbeddedSlow - calculates internaly value of SMA(200) ant put it to internal DataSeries

2) Load CL minute chart from 01.01.2012 to 26.09.2012.

3) Disconnect NT from quotes provider.

4) Disconnect internet and close almost all programs, exept NT and notepad.

5) Run optimisation of each strategy (see screenshot):
chart: CL 1 mint from 01.01.2012 to 26.09.2012
optimize: variable Iteration from 1 to 1000, step 1

6) measure the time of each optimization by CASIO W-S200

Results:
3 minutes 08 seconds - sCodeEfficiencyTestSmaExternal
1 minutes 35 seconds - sCodeEfficiencyTestSmaImbeddedFast
6 minutes 11 seconds - sCodeEfficiencyTestSmaImbeddedSlow