I've plotted histogram of returns of several time bars (i.e. various instruments) and they all seem to be normal. But in the ML track "Section 13 Unit 1" it is said that distributions returns of tick bars are more close to normal, and in addition, it is showed in slides somehow that returns of time bars are NOT normally distributed.
Clarify this topic for me please.
Thank you.
Hello Mohammad,
Apart from the visual method to identify a distribution as normal - it is also recommended to use a mathematical method like D'Agastino-Pearson's or Chi-squared test to figure out if a distribution is actually normal. Yes, it is true that as against time bars tick bars are closer to a normal distribution although distributions like dollar bars are even closer.
For the returns you have simulated you can test the null hypothesis that they were sampled from a normal distribution using this function in the scipy library.