Graph Applications in Finance: Correlations, Arbitrage, and Risk Networks
Article Quant Q&A · Author: DAGman
Summary
The discussion surveys several ways graph concepts can be applied in finance. Correlation networks can represent relationships among assets and help visualize clusters or tightly connected groups. For foreign exchange, exchange rates can be represented as weighted links, allowing graph algorithms to search for low-cost conversion paths or cycles that may reveal arbitrage opportunities. The responses also point to network models of systemic risk, where connections between institutions help frame contagion questions.
Other examples include dependency graphs in derivatives valuation systems, which track which instruments need repricing after market inputs change, and knowledge graphs as generalized databases. The material is a collection of research directions and references rather than a unified method or tested trading strategy. One respondent cautions that some market graph studies mainly visualize correlation data, while computational dependency graphs may fall more under pricing-system design than quantitative trading. Correlation patterns alone do not demonstrate causal links or profitable signals.
Key ideas
- Correlation matrices can be represented as networks to reveal clusters and highly connected groups.
- Currency exchange rates can be modeled as weighted edges to search for conversion cycles that imply arbitrage.
- Financial networks can be used to study how shocks may spread among connected institutions.
- Dependency graphs can limit derivative repricing to instruments affected by changed market inputs.
- Network visualizations describe relationships but do not by themselves establish causality or trading value.
Tags
Full text
# Graph Theory in Finance? # Graph Theory in Finance? I am an undergrad CS/Finance student, recently been given the opportunity with a team to work on graph theory related problems. I am planning on proposing a study under a CS and a math professor on the applications of Graph theory in the markets. I have not taken too many courses in finance yet, but am keen on learning about the markets, I also do not want to miss the opportunity to work on something like this. Could someone guide me as to what I should be looking at and reading in order to propose the research idea, maybe some topics we could start off with, some papers, so that I can propose the idea? Would this be a valid proposal? Thank You ## Answer by Dimitri Vulis (score 6) https://quant.stackexchange.com/a/61189 Not a lot can be done with graph theory / network analysis, but here is a little note http://jonathankinlay.com/2019/09/applications-graph-theory-finance/ (2019) showing how graphs can be used to visualize some data (a kind of a heatmap of a correlation matrix, looking for components and cliques). Here is an older paper (2005), very similar ideas http://sympa.litislab.fr/sympa/arc/graphstream-users/2011-06/msg00004/Mining_market_data%2C_A_network_approach.pdf A 2009 paper doing the same thing https://www.sciencedirect.com/science/article/abs/pii/S0378437109002519 A 2012 note https://blog.wolfram.com/2012/06/01/graph-theory-and-finance-in-mathematica/ on how to do this in Mathematica. A 2017 paper https://evoq-eval.siam.org/Portals/0/Publications/SIURO/Volume%2010/Analysis_Equity_Markets_A_Graph_Theory_Approach.pdf doing the same thing. A 2019 paper https://arxiv.org/abs/1902.00786 by a high school student (2 of my children went to the same HS) doing the same thing. I don't think the other papers do a lot that isn't in Kinlay's short note cited on top. Some people also tried similar visualization of correlation of currency (inluding crypto) exchange rates and of commodity prices. Edit: as Jan Stuller and Will point out, given some currency exchange rates, graph algorithms can be used to find the cheapest path from one currency to another, or to look for arbitrage opportunities. For example, Thomas H. Cormen, Charles E. Leiserson, Ronald Rivest, Clifford Stein. Introduction to Algorithms, problem 24-3 says: > 24-3 Arbitrage > Arbitrage is the use of discrepancies in currency exchange rates to transform one unit of a currency into more than one unit of the same currency. For example, suppose that 1 U.S. dollar buys 49 Indian rupees, 1 Indian rupee buys 2 Japanese yen, and 1 Japanese yen buys 0.0107 U.S. dollars. Then, by converting currencies, a trader can start with 1 U.S. dollar and buy 49 $\times$ 0.0107 = 1.0486 U.S. dollars, thus turning a profit of 4.86 percent. More details about this example are in answers to this question: When looking for arbitrage among a LARGE amount of assets, is there an optimal way? Edit: Visicalc spreadsheet (and its clones like Lotus 123 and MS Excel) use dependency graphs a lot to optimize/minimaze recalculations. For example, suppose you have a USD IR swap, and the EUR swap curve moves. You don't need to reprice the USD swap. But if the USD swap curve moves, then you reprice the USD trade. A more sophisticated dependency (which is pretty hard with multicurves) is to note that if the 10Y swap rate moves, then you need to reprice 10Y and longer tenor swaps, but not shorter tenors. GS SecDb (and its clones like Beacon), BS Proteus, and similar pricing systems make heavy use of dependency graphs. However ine could argue that this is "computational finance", rather than "quantitative finance", and uses little "graph theory". A good overview is Dependency Graphs: A Derivatives Valuation Perspective by Cetin Karakus (BP), if you can find it. Here is a very nice video https://www.youtube.com/watch?v=dXL2bNNQCSk by Mark Higgins from Beacon. Edit: Knowledge graphs (https://www.youtube.com/watch?v=Txb4azvZZrQ, https://www.youtube.com/watch?v=VnovZk4FFys, https://www.youtube.com/watch?v=-Erg6qkdi3M ) are a kind of generalized databases. They look useful, but I don't see a lot of graph theory there. Edit: some more good examples are in answers to this question: Examples of discrete math and graph theory within quantitative finance ## Answer by Jose Pedro Melo (score 1) https://quant.stackexchange.com/a/61317 Graphs have a similar application to ML, but in finance. They might be used to construct more flexible and efficient pricing frameworks. A good example is A. Savine book about scripting languages. ## Answer by SysRisk (score 1) https://quant.stackexchange.com/a/64043 Man the FX arbitrage is so cool, didn't even think about using graphs to finding least expensive path. There is a lot of research on systemic risk and network theory, from theoretic agent based modeling to actual transaction data based models Georg (2013) article has the code available on github: https://www.sciencedirect.com/science/article/abs/pii/S0378426613001295 https://github.com/cogeorg/InterbankNetworkContagion Craig & von Peter (2014) on real world data: sciencedirect.com/science/article/abs/pii/S1042957314000126 to find related papers, use graph theory based application: connectedpapers.com just add the name of the article There is also youtube seminar from Jackson: https://youtu.be/OX3bD0rkUcY Jackson has made a good book on economic networks "Social and Economic Networks", but its not really related on systemic risk, although network formation and contagion process has similar logic behind it
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.