Calling Update() not from the property

[savekad]

If indicator I1 is accessing two vars of indicator I2, is it OK to avoid a redundant call to I2 Update() from the vars get properties and instead just call I2 Update() directly?

From this (I1 code):

// I2 vars properties contain Update()
Print(I1().var1);
Print(I2().var2);

To this:

// I2 vars properties don't contain a call to Update()
I2().Update();
Print(I1().var1);
Print(I2().var2);

[NinjaTrader_AdamP]

savekad,

I can't think of any reason that would cause any issues. Let us know if you run into any.

[koganam]

Originally Posted by savekad

If indicator I1 is accessing two vars of indicator I2, is it OK to avoid a redundant call to I2 Update() from the vars get properties and instead just call I2 Update() directly?

From this (I1 code):

// I2 vars properties contain Update()
Print(I1().var1);
Print(I2().var2);

To this:

// I2 vars properties don't contain a call to Update()
I2().Update();
Print(I1().var1);
Print(I2().var2);

I would like to be corrected but I doubt that that would work correctly. For one thing I2().Update(); seems to be what happens with OnBarUpdate(), so looks redundant.

The way I understood it, we need to Update() in the Property getter because the asynchronous, event-driven nature of NT means that the value may not yet have been updated when we try to access it, so we need to force an update. Somehow, I think this will probably be more of an issue if COBC is false, but I can still envisage a scenario where it could still be an issue with COBC as true.

[savekad]

koganam, can you elaborate?

I think that even if COBC is false, it would still be faster to call OBU just once for any further var accesses than to fire it every time you access a var from the same OBU call of the original indicator (I1 in our case), wouldn't it?

[koganam]

Originally Posted by savekad

koganam, can you elaborate?

I think that even if COBC is false, it would still be faster to call OBU just once for any further var accesses than to fire it every time you access a var from the same OBU call of the original indicator (I1 in our case), wouldn't it?

It will be faster, as you are evidently calling updates less often. I doubt if it will be accurate, again because one may be accessing a value that has not yet been updated, because the system was off doing something else. Such is the nature of asynchronous events.

[savekad]

So you say that by removing Update() from a var getter, it might be inaccurate to access it even if the access is made right after an explicit call to Update() ?

[NinjaTrader_Brett]

Hello,

Update() will only trigger if it needs an Update to get "up to speed". So you wont waste unneeded CPU cycles if the value is already up to date. Its always a good idea to have Update there as an insurance policy to make sure you always have up to date and correct values.

-Brett

[sh_daggett]

Originally Posted by NinjaTrader_Brett

Hello,

Update() will only trigger if it needs an Update to get "up to speed". So you wont waste unneeded CPU cycles if the value is already up to date. Its always a good idea to have Update there as an insurance policy to make sure you always have up to date and correct values.

-Brett

Hello Brett,

I have been wondering about this because I am refactoring some code into multiple indicators to be used by other indicators and strategies. Can you elaborate on your statement, "Update() will only trigger if it needs an Update to get "up to speed"."?

I take this to mean that if I call Update() from within a property or from outside the indicator, the OnBarUpdate() method will only be called if it needs to "get up to speed". How does it know whether or not it needs to get up to speed?

I was thinking that, for when COBC is true, I could put something like the following in the beginning of the OnBarUpdate() method, but it sounds from your comment that this would be redundant; that NT already manages that.

private int lastBarCalced = -1;

protected override void OnBarUpdate()
{
     if (CurrentBar = lastBarCalced) return;
     lastBarCalced = CurrentBar;
     // continue with the method...
}

If my understanding is correct, does NT also "know" when an update is not needed for when COBC is false?

Many Thanks.

Scott

[NinjaTrader_Brett]

Hello,

Sure basically its a good idea to add Update() in the property so that when access a value it makes sure that you have the most update to date variables as possible.

How it works is it will update to the current CurrentBar index if it is not already there.

If COBC=False however this would always force an update since we would have to assume we are not up to date since we have no index to check against as your question lead me believe that is what you are asking on.

There is no cost saving measures we could do on COBC = false like what we can on COBC = true.

-Brett

[koganam]

Originally Posted by sh_daggett

...

private int lastBarCalced = -1;
 
protected override void OnBarUpdate()
{
     if (CurrentBar = lastBarCalced) return;
     lastBarCalced = CurrentBar;
     // continue with the method...
}

Scott

That sequence seems to be identical to just using "if (FirstTickOfBar)...". No?

[sh_daggett]

Originally Posted by koganam

That sequence seems to be identical to just using "if (FirstTickOfBar)...". No?

Hmmmm....well, I guess I was thinking that Update() might be called as a result of OnMarketData() accessing a property of a hosted indicator, in which case OnBarUpdate() of the hosted indicator might be called intrabar, even if COBC is true. I think Brett is saying that I don't need to worry about it because NT handles it in such a way that the extra calls to OnBarUpdate() in the hosted indicator don't happen. I'm probably just confused. What am I missing? (And thanks for the reply and your interest. I appreciate it!)

Scott

[NinjaTrader_Brett]

Hello,

Right if COBC = true you should not have to worry about the extra calls as NT wont force OnBarUpdate to run if it already has processed that bar. If COBC = false however then it would recalculate each time.

-Brett