I would like to ask whether there is a good way of analysing stock indicators that have no value limitations. For indicators like RSI we have a closed range ( 0 - 100 ) but in case of indicators like for example OBV we may have two stocks that have 10000000 and 1000 value of this indicator respectively in the same period of time. Does it make sense to discretize values of such indicators, by creating the same range (for example from 0 to 100) for each stock or it is completely senseless?
2
-
$\begingroup$ Thinking about this concern from a more statistical POV may help. For example your statement about RSI having "a closed range" can be expressed in statistical terms as "RSI is standardardized to fall in a range between 0 and 100." Similarly for the OBV values of the two stocks -- expressing their OBVs as a rate per something would normalize their values, thereby making them comparable between stocks. @michaelhartman (below) has suggestions regarding how to best do that. $\endgroup$– DJohnsonCommented May 28, 2018 at 12:28
-
$\begingroup$ A common trick in Statistics (actually the foundation of Nonparametric Statistics) is to replace the observations by their rank. So you might look at the cross-sectional rank of OBV rather than OBV iitself, and fr example you might buy stocks with "OBV in top quartile" rather than OBV above a fixed value. $\endgroup$– Alex CCommented May 28, 2018 at 14:22
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Technical analysis is not quite in my wheelhouse, but it's been an interesting topic to me, so hopefully I can lend a hand. Let's start with some basic assumptions: Because OBV is based on volume, there is obviously a huge range as you've pointed out. This makes comparison straight across companies impossible.
To compare companies, you need to take out certain factors that could affect volume. My first inclination would be to adjust for the size of a company. Common markers of size would be revenue (REV) and market cap (MKCAP). I wouldn't have much hesitancy in using size as an adjustment for any company data.
Thinking through it a bit more, because OBV is often seen as an indicator of how "smart money" is moving in an economy, you might want to adjust for price, as small investors will be slow to buy really expensive stocks and quicker to buy cheaper stocks. You may want to rethink this for other indicators, but it is an option.
Looking at these two options I'd be more inclined to adjust an indicator for size. This way you can look at companies across a range of sizes and sectors. This might work better for some indicators and worse for other. I don't know how well these would work but it has sparked some interest for me, hopefully it helps you find what you're looking for!
