You might have a look at Patrick Burns' (frequent contributor on Quant Finance Stack Exchange) Portfolio Probe blog for code to apply shrinkage methods to that matrix. Raw historical covar matrix is pretty unstable. What you really want is the expected instantaneous covar matrix -- the historical matrix, no matter what the look-back, is a poor estimator of that.
Can we generalize to "All TRADING books are rubbish?" What a waste of time and money and hope compared to diligently watching markets evolve in real time.
95 % fail after read crap from these trading books!After reading crap , place a bet using the junk science material , then lose .
the op is doing the correct action in posting the list of books he has read. you give info you get info. there are too many takers on ET. they usually end up being ignored.
most of the contents are rubbish but if you get a good idea or two from it it becomes a worthwhile read.
What about the crap 98 % , the shit put inside your phsyche from theses crap writers which costs you dearly?
Believe nothing, test everything and above all look to amalgamate good ideas to develop an approach or method that works for you as an individual. Unfortunately I do not know a way to instant riches.