March 17, 2019

I am writing a new book. I do not have a title yet. Nonetheless, it is filled with simulations, charts, and graphics. For those that have read Beyond The Efficient Frontier, you should find it fascinating, and a must-have since this is the application of what was presented in that book. For those not having the book, here is a summary.

Beyond The Efficient Frontier used an optimizer library (CVXOPT for Python) to make all its stock trading decisions. The book showed that a simple long-term trend was sufficient to extract long-term profits from the market. It would do this for thousands of portfolios having hundreds of stocks.

The particularity was that all stock price series were randomly generated. Because of this, the expected outcome would be zero since no significant long-term trends could be detected within all that randomness. On a purely randomly generated set of stocks, it was demonstrated that it is exactly what happens. For millions of price series and for hundreds of such simulations, the expected profit for all portfolios remained zero.

One could duplicate the same simulations if desired since the cited original program used is available free on the Quantopian website. Quantopian shut down in 2020. So, it is not something thrown on the table without some corroborating evidence. You could do the same tests yourself and obtain the same similar results. I use the word similar because all the performed tests were randomly generated, and no seeds were used. 

Therefore, results would be different for each performed test, no matter how many you did. And there would never be a repeat of a prior or future test either. Everything would be unique each time you ran a new simulation. Nonetheless, each time you would get similar results. In the case of any purely random price series, that expected outcome would be zero.

It was only when you added a long-term drift to the randomly generated price series that the optimizer would wake up and generate profits. The more you raised the long-term drift, the more the optimizer would extract in profits. The sole function of the optimizer was to find an optimal solution as to what would be the best mix going forward based on the past data it was fed.

The long-term drift is equivalent to the secular trend we can find in the market. But, to benefit from it, you have to be in the market for years and years. For instance, the US average long-term upward trend has been close to 10% (reinvested dividends included). This initially translates to about 0.0002 per day of upward trend, or the equivalent of $0.02 cents on a $100 dollar stock. Technically, not even worth the effort on a daily basis due to frictional costs, but over the long term, those pennies do add up and can make all the difference.

What was shown in Beyond The Efficient Frontier was that you could add this small upward drift to the randomly generated price series and design thousands of portfolios that would outperform market averages. All the trading activity was performed by the optimizer, a black box that had for only mission to analyze the data it was submitted and then output its trading decisions. And even if it was randomly generated price data, because of the upward drift, it would generate long-term profits for the thousands of portfolios that would be simulated. When you added some alpha to the mix, you would see profits soar.

Now, this new book goes beyond Beyond The Efficient Frontier. It uses actual market data and shows that you could also control the outcome of the optimizer. The optimizer will do exactly the same job as it did in Beyond The Efficient Frontier. It will issue all trades: going long, short, or do nothing at all. The strategy presented will show outstanding results simply because actual stock prices do have long-term trends built-in. If there were none, the optimizer would return zero. The original program for this is also available on the Quantopian website.

What you will find in this new book is a continuation of my own simulations on Quantopian, where more control is added to the program, and finally, some downside protection is provided. You want outstanding long-term portfolio performance levels, then this will be a tool that can help you do it.


Related article:

Is A Black Box Running The Show

Created...March 17, 2019, © Guy R. Fleury. All rights reserved.