Estimate Two Scale Realised Variance TSRV in Matlab?

Hi

I need to estimate Two Scale Realised Variance using 1 min and 5 min data. Kindly suggest how to do the same in matlab? I am not able adjust the timescale as givin in TSRV in MFE Toolbox by Kevin Sheppard https://www.kevinsheppard.com/MFE_MATLAB .I have 1 minute data series with two colums - first column is DATETIME column given in format YYYY-mm-dd HH:MM:SS and second column is PRICE column. I have series for 2months. How to calculate TSRV for 1 min and 5 min series.

I have attached code file in Matlab.

[NS]

Any help on this?

[Adam Danz]

There's probably a lack of familiarity with the topic across most volunteers here.

Could you provide an equation for TSRV?

2nd best would be to provide the code or a link to the code that calculates this in the toolbox you mentioned.

[NS]

Sure Sir I have attached the code file in Matlab.

[Adam Danz]

That's helpful.

Could you explain what's preventing you from using this function? I know you mentioned, "I am not able adjust the timescale as givin in TSRV" but I don't understand that.

[NS]

Sir

1. Firstly, This code is developed for Calendar times in seconds as you can read in its description "SAMPLINGTYPE " . I have minutely sampled data.
2. Secondly, in "TIMETYPE" it has only time format. I have DATETIME .
3. My data may not have all minutesin a day after cleaning them for outliers.

I have a dataset with 1 minute sample price series for one month for which i need to estimate TSRV for each day. I am not able to use this code as when i split my DATETIME series int each day in cell array. I get may errors as some commands within the code in not applicable to cell. Secondly, how can i ensure that since each day may not have all minutes , this code is able to run TSRV for 1 min and 5 min?

[Adam Danz]

But you can pre-process your data so that it fits these requirements.

It's not difficult to transform minute data into seconds data.

It's very easy to get clock time from datetime.

There are lots of way to fill in outliers so your data are continuous across all minutes.

If your data are in cell format and the code requires a different format, that's a very common problem with lots of solutions.

These are all very common steps people take to transform their data so that it meets the requirements of a function. There are 3 options I can think of.

1. keep searching for a function that is designed to work with your data
2. write your own function that is designed to work with your data
3. transform your data so it works with the well-established function you already have.

I can assure you that option 1 is a very unlikely scenario. Option 2 might work if you are very familiar with the algorithms you need to implement. Option 3 is the most common option in data analysis.

My suggestion is to start transforming your data so it meets the requirements of the function and you can get lots of help with that in this forum and by googling these very common steps.

[NS]

Sure thanks.Let me try what all i need to transform to use this code.

[Adam Danz]

Good! You can get lots of help with those steps here in the forum.