xujiahuayz / defi-econ Goto Github PK

code base for chow test

• Config
• Data fetching cleaning up

1. Uniswap V2
2. Uniswap V3
3. Lending (compound only, no aave for now)
4. Coinmarketcap / coingecko / cryptocompare

by 24 Dec 2022

• Data processing

1. Code refactoring

• refactor fetching and processing code such that data can be added incrementally (not always from the start), and plots can be updated accordingly
• dictionary for stable coin / governance token (add coin type, see #3), deprecate token_library.csv

2. Variable construction

• Eigenvector centrality
• Betweenness centrality (see #12)
• Borrow rate: instead of utization ratio (borrow of toke XYZ / supply of token XYZ), we use relative borrow (borrow of token XYZ / total borrow in the market)
• Other variables (add coin type, see #3)

3. Summary statistics
4. Visualization

• #14

1. Regression

• Structural break / regime switch - Luna/Terra cash, c.a. May 9 / 10 2022. Chow test / F-test
• Coefficients comparison
• Re-do tables (old version /Dropbox/dominant_defi/data/4_tables/Defi_data_22-10-04)

2. Pareto distribution test -- for centrality measurements

Update README.md as you go -- e.g. which file to run to fetch more data, which file to run to re-plot etc.

• Revise betweenness centrality pseudocode on overleaf
• Design betweenness centrality calculation python function
• Plot time-series (same style as eigenvector centrality)

anology between defi & tradfi not obvious to those lack of defi knowledge

• highlight the analogy: e.g. USD-pegged stablecoins have been used as store of value / to pay salary in crypto, ETH similar to working capital currency as it is used to pay costs on-chain

panel data not sufficiently taken advantage of

• consider using random effect (instead of fixed effect, since we have token-specifc variables); consider hausman taylor estimator

defi data not sufficiently taken advantage of

• develop some testable hypotheses around different defi token types (governance token, utility token, payment token etc.)

consider other centrality measurement

• PageRank centrality & EigenCentrality, Social network analysis 101: centrality measures explained

• Unix freq
• Aggregate portfolio
• Stable and nonstable

1. Could probably also write some mini function for HHI, subscript, superscript naming in constants.py.

• https://anonymous.4open.science/r/unmask-mastermind-3FE1/

• Eigenvector centrality for sorting to forecast return. Lending APY to calculate underlying returns. Price to calculate underlying dollar-denominated returns.
• Lending APY changed every day. Annual return to daily (1 + APY)^(1 / ) 100 dollar -> Aave 2Aave -> dollar Compound calculation.
• Single sort: Eigen + APY + return.
• Momentum and Dominance.

• Eigenvector centrality.
• #5
• Utilization Ratio ——> Relative borrow = Token borrow / Total market borrow; Relative supply = Token lending / Total market lending.
• harmonize variables in text, equations, figures, tables.
• Update data.
• re-generate results.
• Update the reference.
• fact-checking.