COMPLEX METHODS EASY SOLUTIONS
About ML Perspective
Hi! My name is Alejandro Romero, I am an Industrial Engineer and Master in Finance & Investments by profession. Since I was a child, I got passionated by finance and statistics.
Never believed on the time-series approach for stock prices as the latter ones are purely random, I personally demonstrated that the probability of a stock price ending up above or below one week later, compare to its current level, is 50% at a 1% confidence level for all stocks in the S&P 500.
Therefore, what to do to "somehow" predict a stock quote? Simple, don't try to predict the price but the probability of it ending above or below its present level, based on its ongoing "conditions", some time later.
Thus, what do I mean when I say "conditions"? What people "think" where the stock quote came and where it is aiming at. Behavioral finance makes fundamentals and technicals to work; consequently, by using Machine Learning to represent stock behaviors, this probability can be estimated.
I decided to change the way we, professionals, analyze data. We constrain ourselves by the traditional approach of time-series, 2D graphs, among others. Stock prices, as many variables on earth, are chaotic; hence, the best approach is to predict a range where it will end up at a determined certainty. Let's see what is hidden beyond stochastic processes together with a different approach!
Data Leverage + Different Approach = Missed Insights Coming to Light
Unfortunately, I have seen many cases of data analyzed on the traditional way or as "it has always be done".
These approaches, as expected, will take the analysts involved to the same results over and over again.
I am a person that is not afraid of making mistakes and takes calculated risks all the time, I always propose new features that reveal hidden information of the variable being evaluated.
Everything is Governed by Numbers
Stochastic processes can be seen as random when they are observed within the framework of our 2D and 3D concepts. The universe, however, is believed to be made up of up to 11 dimensions, however, as 3D living beings, we constrain ourselves to this one or lower dimensions. Just by including time we already jump to a 4D scheme. Then, if the right features are used, some behaviors of the analyzed data come to light.
Random points scattered on a positive second derivative surface compounded by complex planes, dark spots mean imaginary positions.
After 20 seconds of moving between the complex planes randomness disappears and points group up on defined pattern.