Execution Efficiency

[RodS7]

Does anyone know which approach executes more efficiently:

1. Have strategy call an indicator multiple times during execution, or
2. Have the strategy call the indicator once, store the result and use it during its calculations?

Rod

[NinjaTrader_AdamP]

Rod,

The latter would be more computationally efficient overall, though with the speed of modern processors it wouldn't necessarily use up much more resources to do it the first way, but this would depend on the indicator you are using.

1 would probably use more processor time, 2 would maybe use a bit more memory. Typically this is a trade off in different algorithms people use, i.e. if you increase memory usage you can decrease processor usage, if you increase processor usage you can decrease memory usage. Its not always this way, but a rule of thumb. In either case I believe the difference would be negligible unless the indicator is computationally intensive.

Please let me know if you have any other questions.

[sledge]

2 would always win, but you might lose code readability ...


I would do #2, but only if it is slow to compute, like for some custom indicator.

It is probably best to get something working, then optimize after.

Does anyone know which approach executes more efficiently:

1. Have strategy call an indicator multiple times during execution, or
2. Have the strategy call the indicator once, store the result and use it during its calculations?

Rod

[RodS7]

thanks to you both - my instincts said that '2' would be optimal