Chapter 8 - Statistical Correlation models

[ag0511]

Hi

I had certain queries related to the following

1.Pearson's correlation with expected values

Consider sample data set

Probability Returns ( X) Returns (Y)
O.1 15 % 8%
0.2 - 10 % -5 %
0.3 10 % 7%
0.4 5 % 3 %

How does one calculate different variables used in the formula of Pearson correlation for expected values ? (May use different data set for explaining calculation of different variables used in formula)

 

[ag0511]

2.Spearman's Rank correlation and Kendall' Tau

For these ordinal measures,if one encounters a situation in which say subsequent X or Y values are equal.Then after assigning similar rank to both the observations ( correct me,if I am wrong in such a treatment,or if there is any alternative treatment in such cases ),will the next observation will be ranked,n+1 or n+2,skipping the n+1 rank altogether.For example,if 2 observations for X has been assigned rank 2,next observation (higher value) will be ranked 3rd or 4th ?

 

[ag0511]

3.Kendall's Tau

The value of Spearman's rank correlation remains unaffected,even if one straightway assigns rank to raw data or raw values of X and Y,instead of first,tabulating them rankwise (as per increasing values of X).But will it affect Kendall's Tau correlation,if pairs are formed based on raw data or in others words will it affect Nc or Nd ? (Incidentally,both approaches gave similar answers,for a particular question)

 

[Nicole Seaman]

Hello @ag0511

We ask that you please search the forum before posting new threads and that you do not post three different times in the same thread asking about different concepts. All of these concepts have been discussed MANY times in the forum. We are happy to help answer questions, but it saves a lot of time if you search the forum first so we are not answering the same questions over and over in different areas of the forum.

• The search function is here, where you can search for basic terms like this: https://forum.bionicturtle.com/search/?type=post.
• You will see that I added tags to your post to show you how to use the tags. You can search the tags here: https://forum.bionicturtle.com/tags/. If you click above on the tags that I added, threads will come up that are tagged with the concepts that you are asking questions about.

If you still have questions after searching the forum and reading through the discussions, we are happy to help. Here are a couple of links that provide information on using the forum:

• Utilizing the Bionic Turtle Forum: https://www.bionicturtle.com/utilizing-the-bionic-turtle-frm-forum/
• IMPORTANT!! Read Before Posting Questions in the Forum: https://forum.bionicturtle.com/threads/important-read-before-posting-questions-in-the-forum.9071/

Thank you,

Nicole

 

[ag0511]

Hi Nicole

I have three different queries related to this particular chapter,therefore I had posted three times in the same thread,so as to encourage even that person to reply (not necessarily David,initially) who may be having answer to either of the three question,but not all.


Before posting,I did search for the answers to my queries in the threads on the first five pages of the section for market risk.After your reply,I have performed an extensive search,using tags,looking at all the threads in the 24 pages of the P2.T5.Market risk section,using search everywhere function and following the discussions in those threads,where the subject mentioned any of the tagged topics.However I have been unable to find answer to any of the specific queries that I have posted.I may have missed something or may be the discussion has happened in that part of forum,where there might have been restricted access.Can you post the links,here ?

A possible way of dealing with issue raised in your post is to create sticky chapterwise threads in each subject/section.Hopefully it should encourage all students to post their doubts related to any chapter at a single place,doing away with a lot of unnecessary threads.The links to threads created in the past,can be offered as solution,if required.It make take some time to implement or may be time consuming at the onset.But will be much rewarding and efficient in the long run


Thanks