WES WARRINER
Portfolio
Portfolio Analysis & Optimization
An example of the first steps I might take toward deeper portfolio analysis and optimization, given a predefined set of assets.
Portfolio Optimization
Tech/Growth Portfolio (TXG)
The portfolio optimization process itself, to follow preliminary analysis.
​
A couple of notes:
-
As asset choice is outside the scope of this project, the set of assets used below may be considered arbitrary, although I do discuss some of the relationship between portfolio composition and optimization strategy.
-
In developing the following module I went through many more specification objects and optimization runs than are actually shown as examples below. In addition the 'optimized' portfolios are selected from a randomly generated set of potential portfolios with each run, meaning that the optimization results/visualizations below were produced when I knitted my Markdown to HTML. On the other hand the comments/analysis included within the document are more general, referring to trends I noticed over the course of many optimizations, and as such may not apply specifically to the results you see on this page. In any case, everything here is purely for the sake of example.
Single-Asset Returns Analysis & Forecasting (NDX)
A bit of a branch off the first module, the following is an example of analysis and time series forecasting for a single asset.
This module would also be a part of the preliminary analysis, or better yet an example of how to assess asset behavior and make predictions before deciding whether to include a given asset in the first place.
The NASDAQ-100 index is just that, an index, as opposed to say a single stock or ETF. For the sake of example I wanted to use an asset with at least a decade or more of returns with relatively few missing values, for which this kind of index is ideal, but that doesn't mean the same process isn't replicable on an asset with less history or incomplete data; we would just have to accept a little more uncertainty.
This project will work a bit differently from the others I have up at the moment.
I wanted to include a more comprehensive portfolio analysis/optimization run-through, so I have divided it into 3 parts, or modules.
​
The first module includes preliminary analysis of a set of assets; the foundation of a diversified portfolio.
I focus here on looking at multivariate relationships and analyzing the portfolio as a whole, in addition to exemplifying some efficient frontier visualization.
The second module is the meat of this project, running and comparing several portfolio optimizations of a selection of primarily IT sector & growth stocks.
We look at different strategies for choosing constraints and objectives, and consider what actually turns out to be most important when assigning parameters to select an 'optimized' portfolio.
​
​
The third and (for now) final module in this financial analytics series is a branch off the first module, providing an example of single-asset time series analysis and returns forecasting utilizing a couple R packages which support ARIMA modelling techniques.