It has been nearly three years since my last update on Midas51, but not for lack of activity. The gulf between updates has been due to project work dominating most of the time that has remained after my day jobs have gotten their due.
The 32% compound annual growth rate (CAGR) and maximum drawdown of -34% that is being yielded in simulation by an ensemble learning model discovered in Midas51’s third iteration warrants a progress update.
I spent much of last summer reading up on data science, machine learning, statistical learning, and predictive modeling. Interestingly those topics turned out to be highly overlapping in the details although they differed in terminology. In turn I will refer to them collectively as data science.
My motivation for getting up to speed on data science came from my experience with the second iteration of the Midas51. Here I was still utilizing well known financial market technical analysis and price chart pattern techniques and some custom techniques that were inspired by the well known techniques. A lot of experiments were done but none had yielded any results that were acceptable to me as the CAGR was to low, the maximum drawdown to deep, and/or the result was most likely overfit and unlikely to be reproducible.
To ensure I wasn’t missing anything, in the spring of last year I read numerous books on financial market technical analysis and price chart patterns. I hadn’t done anything with Elliot Waves (EW) previously, so that was new. After paying to be able to watch pros do EW analysis, it was clear there is a subjective aspect to doing EW analysis that precludes using simple algorithms to automate such analysis. EW practitioners also have fairly inconsistent track records. This makes it unclear if EW analysis can’t work consistently or if the human analysts aren’t applying it consistently enough.
This lead to asking the questions:
- Could the learning algorithms of data science could be applied to do EW analysis in a sufficiently consistent way?
- Or perhaps learning algorithms could do price chart pattern identification and then pattern recognition?
And thus began my journey into data science land. After getting a solid introduction to data science and some inspiration from Mandelbrot’s The Misbehavior of Market, I had some newly formed hypothesis that I believe to be worthy of making a substantial investment in exploring.
So, with this combination of new knowledge and inspiration I left my position at Tableau Software last September and fully immersed myself into the third iteration of Midas51.
And now back to work!