Minimizing-Slippage-Loss-on-DEXs
An optimization algorithm for liquidity allocation on DEXs
The algorithm has two major components:
-
- A discrete optimization algorithm for choosing which liquidity pools should exist (a modified version of Simulated Annealing, https://en.wikipedia.org/wiki/Simulated_annealing)
-
- A kinda continous optimization algorithm for choosing how much liquidity should be allocated given the set of liquidity pools (an Evolutionary Algorithm)
The loss function that the algorithm trains on captures the minimum cost that traders would have to pay given an order book with 24h of transactions and a fully allocated decentralized exchange (DEX). To do so we created a modified dijkstra's algorithm (https://en.wikipedia.org/wiki/Dijkstra%27s_algorithm), which recursively finds the minimum cost for how much a trader pays in gas, transaction fees and slippage loss for a single transaction.
To improve run time, the optimization algorithm was entirely written in C++, and its results are stored in respective CSV files which can be visualized using the python visualizations. The necessary packages to run all of the python files are:
- Networkx
- Pandas
- Imageio
- Matplotlib
- Numpy
If you are serious about helping contribute to and organizing the repository, reach out on (https://twitter.com/eljhfx)