I want to make Fama-French time series analysis.But to make time series for a time span (let's say 10 years time horizon), lets say X firm must be listed in stock market 10 years, right? But, some firms does not there for ten years. Therefor, I decided to take firms listed (stayed) in market for ten years. But, I occurs survival bias problem, I guess. Because, I take firms that survived. So,what can I do about it?Is my approach true?I appreciated it if you can help me.
Dear Canan,
I can assure you none of the stock market theories will work experimentally unless you are certain the stock exchange has reached steady state at least over the long run in mean-variance space of returns necessary for quantum mechanical factorisation and integration of stock market information systems. See for e.g. Mallick (2014), Mallick, Hamburger & Mallick (2016, 2017, 2018).
S.K.Mallick,
for S.K.Mallick, S.Raychaudhury & S.Mallick ,
of The RHMHM School,
USA, Japan, India
Szymon Stereńczak thank you for your answer. Index sample (NYSE for instance) composition can change every year but not the total number listed in NYSE. So,lets take SH(small cap and high book-to-market portfolio for example while calculating SMB(small minus big). SH portfolio is constructed from different samples of NYSE because changes in NYSE composition.I thought that data sample that I pull to make time series for SH changes and it is not right.It is not the same independent value (for SMB ).But I think I have to do like this because the other option is totally wrong.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series