Eclipse Technologies

Have you ever watched flowing water follow a specific route it seems to “create” on its own, as if it “knows” the path of least resistance? Or an ecosystem, growing with all its species and subspecies and various systems all working together to fill niches where other organisms don’t yet exist? Where does this phenomenon come from? What can we learn from it? How can we use these lessons in financial markets?

I’d like to change the way you view markets, or at the very least show you a new perspective that you can use to better understand them.

I view financial markets as organisms rather than a set of human created systems. To view them as a simple manmade system would be inconsiderate of the incredible emergent properties and trends they give rise to every day, which are difficult if not impossible to predict in many circumstances. Put simply, modern markets are a completely different beast than they once were.
Markets facilitate the movement of money between their participants (Individuals, investment firms/funds, governments, companies, etc) In this sense, money is fluid, like water. Money flows between markets, and between all the participants in those markets in a seemingly random, unpredictable manner. Trends in prices emerge, disappear, reappear, fluctuate over time. This is why we’ve identified markets as chaotic dynamical systems- and the reason why there is a huge number of professionals hell-bent on “solving the market” to generate vast amounts of wealth. Part of what makes this problem so alluring is its difficulty.
I’ve always viewed the problem of forecasting financial trends (over any timeframe) as similar to fluid dynamics in physics. One of the greatest unsolved mathematical problems is the search for solutions to the Navier-Stokes Equations, a set of equations that describes the flow of incompressible fluids. Funnily enough, you also get a large financial reward ($1m) for submitting a genuine solution to this millennium problem.
So, viewing money as an incompressible fluid that “flows” between assets can help you better understand how markets behave, and why they act the way they do. A classic example is when interest rates go down, money flows from less risky assets into riskier assets, causing stock prices to rise, and vice-versa.
Now, this philosophy is only helpful over the long term if you understand what causes money to move. To do this, you must have a good understanding of the goals, power, and information advantages of the systems involved. Is everyone trying to maximize reward? (Not always) Is some money being used for hedging? (Sometimes) What if a player is just trying to diversify (getting the same risk/return profile in a different asset)? These are all considerations that have to be made, and their answers are uncertain, which messes with our model of money flow and makes our lives a lot more difficult.
For a moment, let’s discard the idea of long-term forecasting with its various difficulties and focus on shorter term forecasting in single markets. The shorter the timeframe, the less fundamentals tend to matter.
Now, let’s say our only goal will be to accurately predict if stock XYZ will have more buyers or sellers over the next 10 minutes. If XYZ has more buyers, it will rise, and vice-versa. Simple enough.
I have found through my own research that there is an optimal time horizon for short term forecasting that most market participants operate on. This is due to the average size of firms participating in markets- most cannot get in and out of meaningful positions in milliseconds, and the majority do not take a long time (weeks to months) to accumulate or shed positions unless they are very large firms.
Once we’ve established the timeframe we are using to trade, we can begin building a model to predict the flows of money throughout various markets, asset classes, and specific assets. The more specific our model, the more profit we can potentially extract, but it also becomes increasingly difficult to make accurate predictions due to the higher complexity of the systems involved and the rising number of interactions between them.

This leads to a very interesting finding- there is an “optimal” level of complexity for our model at which it no longer makes sense to increase the number of inputs to try to improve our model’s predicting power/ accuracy, because it will no longer result in worthwhile increases in profits. This is the point at which we cannot, no matter how good our model is, predict more accurately using the information available to us. This is due to constraints imposed by the laws of physics on the amount of information that can be encoded within variables (Shannon Entropy).