The markets are the places where opinions are transformed in money. Opinions are represented by numbers. In we consider the S&P 500, it’s numbers represent the greed and fear oscillation, under a global scale, as it is the largest stock index traded. The market is an ever-changing-chaotic-bipolar environment, made up by thousands of individuals, everyone with a different view in mind.

The markets sort out of noise. When you look at a chart of a financial instrument, you see a noise reduction artifact, that makes much more sense to the eye, but is actually a convention for segmenting a free flux into manageable items. The fundamental noise gives shape to the chart. The tenths of thousands of actors that interact with the S&P 500 are the propelling noise that fuels the market.
Before the electronic era, the price was unique for the day, being the equiibrium value for all partecipants. Today, the price is the close of the time frame and that’s not a small difference.

Using technical analysis (ta) tools, I’ve always seen many attempts (indicators) to “clean” the data, to reduce or  to wipe out noise.A clear overfitting attitude. Splash ahead. When you take a position, it is under some expectations and sentiments are involved. Sentiments conflict with opinions. They struggle by nature.

Using neural networks (ai), I’ve discovered a tool that loves noise and is able to manage noisy correlations with a high degree of abstraction, or intuition, as you may call it.

While opinions still run wild in the territory of ta, they drastically reduce if you use have a robo-advisory ai that crunches numbers in the background.  And the robo-advisory is where the knowledge is transferred, where a learning process is applied to give to the ai the experience it needs.

In r.Virgeels’ case, the training is by necessity influenced by my attitude towards the market, I’m aware. I revised the training so many times, that I would say it is scientifically correct, but inside I know that another person should do some different evaluations and should correct some items.  Consider that training is 100% connected to output quality. If you set up a fantastic neural model and feed it with rubbish, you just  get rubbish out, no way.

The neural networks need to have all their data and training correct and well fitted, then the magic may happen. And then, you no more need opinions, just a few about the framework, about the global picture, for being in sinchro. r.Virgeel knows the market better than me and you and probably any human on the planet, for the simple reasons that it can correlate dozens and dozens of inputs, and we cannot.