Predict or Follow?

Like I mentioned earlier, no one way is superior to another. Both are aimed at making money. It all depends upon your inclination & expertise, data points at your disposal, and their suitability to the method you apply.

Just make sure for prediction purposes you rely on, leading indicators into your analysis. They could be technical, news, alternative data etc. However, sourcing of these leading indicators and synthesizing them for use in your algorithms (especially news and alternative data) can be tricky both for cost and their efficiency. Also, be aware of stickiness of predictability of these data points. The value of few data points may evaporate rather quickly in minutes, some in hours and others last longer.

When you choose to follow the market, get ready to believe in the trend. There is a popular adage “market trends for about 20% of the time and consolidates for the remaining period”. So focus more on reducing the traps/false trades.

In early days, when I googled for help, I was surprised to see so many articles indicating how to use different ML models to predict the market. Initially, I told myself, when a properly trained algorithm can classify an image it never saw before, ML must do wonders in predicting the market. With great excitement I started exploring different models only to realize that the objective of these articles is to introduce ML for a given use case. They don’t aim at providing trading algorithm. ML definitely has its place in AlgoTrading, but be prepared for long haul than looking for quick answers. Also make sure that you break your hypothesis into smaller puzzles and try to use optimal approach for each of them. Don’t make the mistake of trying to solve all with one tool.

Please get this, Market is like a live organism responding continuously to new information. The validity of your hypothesis could be well founded at this moment, but that’s not guaranteed for the next.