Skip to content

2019 New Ideas

.

Books Currently Reading

Some Ideas To Resolve Soon

  • Poisson distribution
  • What is PCA?
  • What is irvega rho?
  • Some notes on financial markets
  • How to use TensorFlow?
  • How to use pivottable.js?
  • How to use metplotlib for charts?
  • Backtest strategy to buy on n% dips?
  • Build pages on individual tickers
  • How to host python script on AWS?
  • Why use NumPy?

New Investment Ideas

A new proposed idea is to buy on n% dips. Would this work? Or buy an index instead of individual stocks on such dips.

PCA

PCA - principal component analysis sounds very much like fourier transform. It is a statistical proceudre that reduces 100 variables to a few important- orthogonal (meaning unrelated) variables- principal components- that explains 95% of the movements of the original wave.

PCA is used primarily for making predictive models.

Rho

Rho is the sensitivity of the option value to changes in the risk free interest rate.

PivotTable.js

PivotTable.js looks pretty powerful.

Why use NumPy?

  • ndarray- a fast and space efficient multi-dimensional array providing vectorized arithmetic operations and broadcasting capabilities
  • functions for reading and writing to disk
  • tools for integrating with C, C++ and Fortran
  • math operations with no loops
  • linear algebra and fourier transform functions

But what functions does NumPy provide and what from Pandas? I believe data analysis primarily use math functions from Pandas and use NumPy primarily for its fast ndarray.

Pandas runs on top of NumPy- so Pandas depends on NumPy. Pandas provides datatable-like functions- as well as stats, groupbys, merge and join methods.

Pandas was built starting in 2008 by Wes McKinney at AQR.

Matplotlib for Quick Charts

Quick charts demo

Poisson Distribution

Poisson Distribution is for modelling the number of times an event occurs in an interval of time or space- for example- the number of patients arriving in an emergency room between 10 and 11pm.

The following assumptions must be true in order to use Poisson Distribution

  • k is the number of times an event occurs in an interval- like 0, 1, 2 …
  • The events are independent.
  • The rate at which events occur is constant.
  • Two events cannot occur at exactly the same time.