There are so many resources online about machine learning that almost anyone with knowledge of basic math and copy paste capability can write a book on the subject. But in finance results matter only. There are ways of putting ideas to test besides trading with real money.
Three years ago we developed a strategy with DLPAL S for DIA ETF with in-sample data from inception to 12/31/2009 and then tested the out-of-sample performance from 01/04/2010 to 08/16/2016 in Quantopian platform. The related article can be found here. Below is a screenshot of the results from the article.
It may be seen that the strategy outperformed SPY ETF buy and hold return in the out-of-sample by a wide margin.
But how has the strategy done in the last three years given that volatility rose and in 2018 many quant strategies underperformed?
Below is the performance of the strategy from 08/16/2016 to 09/09/2019.
Performance fell but still about 50% of SPY total return in the period. Given that most machine learning models fail immediately after the test sample, we think this is good performance.
Those who thinks a trading strategy can be developed via machine learning they can just go ahead do it and evaluate it in Quantopian platform. Program the strategy in Quantopian and after two to three years you will know whether the math can translate to profits. Until then, books, papers, abstract notions, etc., are all good but are no indication of performance.
If you found this article interesting, I invite you follow this blog via any of the methods below.
If you have any questions or comments, happy to connect on Twitter: @mikeharrisNY
Charting and backtesting program: Amibroker
Technical and quantitative analysis of Dow-30 stocks and 30 popular ETFs is included in our Weekly Premium Report. Market signals for longer-term traders are offered by our premium Market Signals service.