The Kolmogorov–Smirnov test aims to tell you if the distribution of prices (or log returns) tends to follow a normal distribution or not. You can read about this test on Wikipedia. It seems to be a basic but trusted measure in the quantitative trading world.
When KS-t columns are blue, then it's safe to assume normal distribution. When they are red, the normal distribution assumption is proven wrong by the magnitude of the KS-t value.
In the plotting tab of the script, you can activate another option that displays the probability of the distribution being actually normal. It's values are bounded between 0 and 1, like all probabilities.
This test can be useful when using statistical concepts for trading markets, like standard deviations, z-scores, etc because they all depend on the assumption of prices (or log returns) being normaly distributed.
If you see something wrong, don't hesitate to message me.
@RicardoSantos, hello and thank you for your remarks. It seems that the first link you provided is about the Two-Sample KS-test, but my attempt here is about the One-Sample KS-test, as it tries to compare the observed cumulative distribution function with its specified theoretical distribution. The second link seems to be a nice read, that I've yet to begin ;)
RicardoSantos
⋅
@Hachann, perhaps, i was expecting output to be 0<KS<1 but i may just be overthinking it or mixing it up with the rejection probability :) anyways thank you for sharing it
github.com/daithiocrualaoich/kolmogorov_smirnov/blob/master/src/test.rs
further reading:
daithiocrualaoich.github.io/kolmogorov_smirnov/