How to overcome this issue (weeks contains holidays )

[huab]

I wanted to make a study that divides last week volume to the week before.. (timeaggreg: week), the problem is that there are 9-10 weeks during the year with 4 trading days, so the comparision will not be logical..
Can I use calculations from getweek(), getdayofweek() -and any other data- functions, or there are an easier/straightforwad ways?

 

[Joshua]

Can I use calculations from getweek(), getdayofweek() -and any other data- functions, or there are an easier/straightforwad ways?

Can you use them... to do what? What is your plan on how to handle the four day weeks?

 

[mashume]

I think, the most straight forward approach would be to compare the sum of the preceding 5 days to the sum of the 5 days before those:
sum(volume(aggregation daily), 5) / sum(volume(aggregation daily)[6], 5)
or something like that.

it's a trade off (no puns intended) that will give you a rolling week rather than a static Monday to Friday week, but it may prove useful.

The most important questions, and the one that @Joshua tried to get at is what will you do with your statistics when you find a four-day week. That is a question whose answer will determine the approach you take in mitigating the bad data that a 4 day week presents in analysis.

If you could determine the four day weeks, would you simply multiply the volume for the week by 1.25?

Another random thought is this:
https://tlc.thinkorswim.com/center/reference/thinkScript/Functions/Date---Time/CountTradingDays
You might be able to use some magic to determine if a given number of days in the past is the same number of trading days in the past. Just a thought.

Perhaps it's helpful, perhaps not so much so
-mashume

 

[huab]

Can you use them... to do what? What is your plan on how to handle the four day weeks?

I imagined if i used getyear(), then we know from the calendar holidays fall in which weeks, getweek() returns the week number from 1 to 53, then make a condition if getweek equal 3, for example, and that was 4 trading days, then multiply the volume by 1.25 ..
I mistakenly typed getdayofweek() as I was thinking of other way but it required multi aggregation times (then, for example if the volume of the symbol in day 4 of the week is nan then multiple the previous one by 1.25 ..
(I'm still a beginner)

 

[Joshua]

Yes, if you actually wanted to check the calendar and do it manually like that, it could work. A better question might be, what type of chart are you planning to use this indicator on, and-or how do you intend on displaying the data? The most effective method I could think of would require the script to be used on a daily chart, which would reconstruct the weeks from that perspective, and automatically detect the four day weeks.