Market Insights
By Xiaoxi Wang • Aug 01, 2025
From International Economics to Stock Picking: KnotCord Gravity Models Explained
#international-economics
#gravity-model
#stock-picking
#trading-partners
#feature-engineering
When gravity model is mentioned in talks, people think of physics. It is cold info to the public that gravity model is the power horse model in international economics. It builds a framework to explain the mechanism of international trade, namely the micro side of international economics. To put it in the most raw and basic sense, a pair of countries trade volume with each other is dependent on the two countries' GDP, divided by their distance. We are not jumping into the technical details of the function here. You should head up to pursue the field of international trade in a PhD program in economics, if that really interests you. I bring up this little background here is to explain what inspire me to create the adapted version of the gravity model used in international economics and apply in the field of finance.
The logic is quite straightforward. Stocks are affecting each other. No single stock is running independent of the whole capital market. It is always a group behavior that ignorance to other stocks, or the general market situation lead to memorable life lessons. You know I am telling the truth if you have enough investment experience. If we agree with this first assumption, we are comfortable to assume further that there exist forces between stocks. There are tons of ways to describe a stock's 'growth features' and 'distance' between stocks. That is the secret recipe of each international economist. I'm not handing out like free food sample in Costco. It is called field expertise in the language of AI training!
Let me guess the next questions in your mind.
Q: Why not use correlation coefficient instead?
A: Check out the function first. Take the most commonly used one. It is relationship between two variables only. It missed a lot of information, not to mention that it missed to capture variation across time. The gravity model KnotCord is offering is using multi-dimensional data.
Q: Is this simply pair-trading?
A: It is pair-trading in fancier way. When an economist build up an economic model in theory, it always start from building a two-agent case, say two-country, two-player, two-product, etc. If it makes sense, move forward to three-agent case. To continue, it will be n-agent case. Get back to our question here. It is pair-trading in n-stock case.
Q: Why not use heatmap? It watches the market situation in a bird view sense as well.
A: The heatmap you usually see in other financial analysis platform only tracks the price percentage change. Heatmap is a 2-D way to display data matrix. Plus, it does not entail the force concept at all. It considers all those stocks behaving independently with each other. KnotCord Gravity is 3-D display of multi-dimensional data. What dimension data do we use? To say in a sentence, technical indicators, fundamentals, and sentiments. It is called feature engineering in machine learning language.
Q: Is this some kind of index?
A: You may say that. Usually speaking, index is a 1-D number to capture group behavior. In the KnotCord Gravity view, you see variables such as force, distance, rank.
Q: How do I use it exactly?
A: Choose a date, a central ticker of your stock, and number of connections you are considering, then hit the button to generate a network. You will see the most important stocks around your central stock. This is single-network mode. You may also choose duo-network mode, to enable comparison between networks. This comparison gives you a bird view to monitor market situation in an innovative way. If this is no innovation, I don't know what is. Don't forget the effect when candle chart came out for the first time. If it still doesn't intrigue you to give it a try, you probably misunderstood how innovation in reality looks like. When you look close enough to details and history (literature review in academia), you will agree that innovation is always marginal. The uncertainty is that whether the marginal innovation is going to be the right direction in history.
Q: Is it good or bad to see some unexpected stocks around my central stock in the generated network?
A: That is up to you to decide. Just like how you are going to move after read the candle charts, and all those technical indicators that might be contradicting or not. A polished tool like KnotCord Gravity Model is here to assist you in investment analysis. You are the one to make the final judgmental call.