Imagine millions of alphas as particles in a chaotic environment. Moving up and down, their correlation windows as well as their respective coefficients are ever changing. Sometimes there are patterns sometimes is just noise.
This reminds me of the second law of thermodynamics, which says that the total amount of entropy (or disorder) in the universe can only increase with time.
Some would argue to focus on the ones that make sense at an economic/market structure level. These are very well-known and well-studied in academia. If they aren’t working for the time being, or performance looks random lately, at least you can say they make sense economically/structurally hence worth keeping, just solely because they make sense.
This is a very valid approach for trading. At worst you will have a few alphas that work sometimes and when they do or don’t you can explain why.
What about the ones that humans can’t explain? Call it black box alphas, or programs that are not directly readable. You just know what comes in and what comes out. Whatever happens in the middle God knows.
It’s like deep diving into the ocean, you find these weird-looking monsters. These can behave wildly, just because they haven’t been studied before nor have years of academic research behind them.
These monsters very interesting, maybe just because they are so different and difficult to find. They are unique, dark and hidden, in the sense that people are unaware of the existence of a particular monster just because how can someone going to Google something they can’t explain nor know it exists. You don’t know if they are random at first just because you don’t know how their gears really work, can’t even read them.
Curiosity is a strong motivator taking you to do many deep dives, each time you encounter a new monster. After many sessions, you find yourself with an army of these wild monsters. At first, looking chaotic but starting to study them you discover some of them perform badly. Others behave superior. Some move in groups and inside these groups there are leaders and laggards.
There are so many of them you have to put some order in place to fight an increasing entropy. Ranking and removing noise basically, can call it meta-alphas or the art of combining and filtering many sources of information. Then you start to notice that indeed alphas are just chaotic bodies with temporal patterns, some of them with decent statistical properties for some time, others are utterly nonsense. All they really need is a director to set the tune. And indeed entropy increases with time, just because with time new bodies come out. And the only way to fight chaos is orchestration.
Now that you might have the curiosity about these deep sea monsters may ask me:
“Where can I find them?”
Sadly that’s probably the only thing I can’t help you with, the search for unique and superior information. The purpose of this post is to spark your curiosity.
If you want to learn how to deep dive, you’ve got to jump in.