Questions regarding 2 homework

[OlgaKuzminichna]

My questions regarding second homework:

1. Is it fine to publish homework in py format rather than Ipub?
2. Task 2 with histogram, what could be the best advice on picking right bins
3. Why we don't smooth the plot with returns,
   for me it seems to be irregular fluctuations and I would prefer to use rolling for a better picture?
4. Task with heat map. I got confused why it is:

 returns.iloc[:10, :10].corr() 

but not

returns.iloc[:, :10].corr()

The first option means we take first 10 companies and only first 10 dates, so we ignore another 240 dates

5. Why SIEGn.DE .GDAXI is not picked as highest stock pair as 83 is not 85
   

[OlgaKuzminichna]

The answers for my questions:

1. Fine
2. Mostly depends on research question, but there is a general advice available on the internet
3. It is fine to smooth for a homework, it could result in the loss of variability and heteroscedasticity
4. it is typo- correct option is returns.iloc[:, :10].corr()
   alternatively we could calculate correlation first and then filter rows and columns by 10
5. Because .GDAXI is not a stock

[cafawo]

Regarding Q2, the optimal number of bins in a Histogram:

You could consider the Freedman–Diaconis rule, which is based on this paper: https://link.springer.com/article/10.1007/BF01025868

Technically Freedman–Diaconis gives you the optimal bin with $h$, if you want the number of bins $n$ you can covert by

$$n = \frac{\max - \min}{h}$$