We can simply compute a rolling monthly return by subtracting the previous month’s average stock price from the current month and dividing by the previous month’s price.

The return is shown in the following figure,The optimization modelThe return on a stock is an uncertain quantity.

We can model it as a random vector.

The portfolio can also be modeled as a vector.

Therefore, the return on a certain portfolio is given by an inner product of these vectors and it is a random variable.

The million-dollar question is:How can we compare random variables (corresponding to different portfolios) to select a “best” portfolio?Following the Markowitz model, we can formulate our problem as,Given a fixed quantity of money (say $1000), how much should we invest in each of the three stocks so as to (a) have a one month expected return of at least a given threshold, and (b) minimize the risk (variance) of the portfolio return.

We cannot invest a negative quantity.

This is the non-negativity constraint,Assuming no transaction cost, the total investment is restricted by the fund at hand,Return on the investment,But this is a random variable.

So, we have to work with the expected quantities,Supposed we want a minimum expected return.

Therefore,Now, to model the risk we have to compute the variance,Putting together, the final optimization model is,Next, we show how easy it is to formulate and solve this problem using a popular Python library.

Using Python to solve the optimization: CVXPYThe library we are going to use for this problem is called CVXPY.

It is a Python-embedded modeling language for convex optimization problems.

It allows you to express your problem in a natural way that follows the mathematical model, rather than in the restrictive standard form required by solvers.

The entire code is given in this Jupyter notebook.

Here, I just show the core code snippets.

To set up the necessary data, the key is to compute the return matrix from the data-table of the monthly price.

The code is given below,Now, if you view the original data table and the return table side by side, it looks like following,Next, we simply compute the mean (expected) return and the covariance matrix from this return matrix,After that, CVXPY allows setting up the problem simply following the mathematical model we constructed above,Note the use of extremely useful classes like quad_form() and Problem() from the CVXPY framework.

Voila!We can write a simple code to solve the Problem and show the optimal investment quantities which ensure a minimum return of 2% while also keeping the risk at a minimum.

The final result is given by,Extending the problemNeedless to say that the setup and simplifying assumptions of our model can make this problem sound simpler than what it is.

But once you understand the basic logic and the mechanics of solving such an optimization problem, you can extend it to multiple scenarios,Hundreds of stocks, longer time horizon dataMultiple risk/return ratio and thresholdMinimize risk or maximize return (or both)Investing in a group of companies togetherEither/or scenario — invest either in Cococola or in Pepsi but not in bothYou have to construct more complicated matrices and a longer list of constraints, use indicator variables to turn this into a mixed-integer problem – but all of these are inherently supported by packages like CVXPY.

Look at the examples page of the CVXPY package to know about the breadth of optimization problems that can be solved using the framework.

SummaryIn this article, we discussed how the key concepts from a seminal economic theory can be used to formulate a simple optimization problem for stock market investment.

For illustration, we took a sample dataset of three companies’ average monthly stock price and showed how a linear programming model can be set up in no time using basic Python data science libraries such as NumPy, Pandas, and an optimization framework called CVXPY.

Having a working knowledge of such flexible and powerful packages adds immense value to the skillset of upcoming data scientists because the need for solving optimization problems arise in all facets of science, technology, and business problems.

Readers are encouraged to try more complex versions of this investment problem for fun and learning.

If you have any questions or ideas to share, please contact the author at tirthajyoti[AT]gmail.

com.

Also, you can check the author’s GitHubrepositories for other fun code snippets in Python, R, or MATLAB and machine learning resources.

If you are, like me, passionate about machine learning/data science, please feel free to add me on LinkedIn or follow me on Twitter.

Tirthajyoti Sarkar – Sr.

Principal Engineer – Semiconductor, AI, Machine Learning – ON…Georgia Institute of Technology Master of Science – MS, Analytics This MS program imparts theoretical and practical…www.

linkedin.

com.. More details