Anything like "lowest 95%"?

[heech]

Hi,

I'm looking for an easy way out. I'm currently using High/Low within my logic, but I'm working with 30 minute bars... and sometimes these values can be skewed by a really short burst in one direction or the other.

So, I want to filter some of this out... by looking at the lowest value for "almost all" previous bars.

Is there an easy function that I can call, within my 30 minute OnBarUpdate()... that would give me the value that X% (say, 95%) of previous 1-minute bars were higher than?

[NinjaTrader_JoshP]

You can try the real-time tick filter if you feel the data is erroneous. Tools->Options->Data.

If you feel the data is valid and still want to proceed you will need to roll your own function.

[heech]

Josh,

The data is valid, I just wanted to find an easy way out. I was hoping DataSeries would have an easy function for this.

[NinjaTrader_JoshP]

Unfortunately not.

[michaelangela]

Hi there, it may not be a built in function, but I am fairly confident C# can handle it and it sounds like you're coding this in C#. And this isn't a precise thing, just something to eliminate the extreme data. The basics are simple: if you wanted to eliminate the most extreme data, get your data in an array, sort it, and remove the highest value.

I am learning C# now so I don't have the syntax on this but basically it would be something like:

get an array of values you want to check
sort that array
take off the top x values or lowest x values
read the last hightest or lowest value

As an example, and for easier math, use 100 minutes, or 100 bars of 1 minute data. Copy that into an array. Sort that array say from low to high so that the lowest value is at 0, i.e. sortedLows[0] is the lowest value. sortedLows[5] should then roughly represent the 95% low. If you didn't take 100 bars of data then you'd have to figure out which bar you would want.

Just a thought.

Michael

[heech]

Hi there, it may not be a built in function, but I am fairly confident C# can handle it and it sounds like you're coding this in C#. And this isn't a precise thing, just something to eliminate the extreme data. The basics are simple: if you wanted to eliminate the most extreme data, get your data in an array, sort it, and remove the highest value.

I am learning C# now so I don't have the syntax on this but basically it would be something like:

get an array of values you want to check
sort that array
take off the top x values or lowest x values
read the last hightest or lowest value

As an example, and for easier math, use 100 minutes, or 100 bars of 1 minute data. Copy that into an array. Sort that array say from low to high so that the lowest value is at 0, i.e. sortedLows[0] is the lowest value. sortedLows[5] should then roughly represent the 95% low. If you didn't take 100 bars of data then you'd have to figure out which bar you would want.

Just a thought.

Michael

Great idea. I hadn't started to tackle this yet, but what you're saying sounds absolutely like the way to go.

Here's an easy to use quicksort routine:

http://en.csharp-online.net/Quick_Sort