https://github.com/mtefagh/fee
Designed an attack for Vitalik's EIP 1559 and proposed an alternative transaction fee pricing protocol based on the Almgren-Chriss framework and median price auctions
https://github.com/mtefagh/fee
cryptocurrency ethereum-blockchain game-theory mechanism-design simulation transaction-fee
Last synced: 3 months ago
JSON representation
Designed an attack for Vitalik's EIP 1559 and proposed an alternative transaction fee pricing protocol based on the Almgren-Chriss framework and median price auctions
- Host: GitHub
- URL: https://github.com/mtefagh/fee
- Owner: mtefagh
- License: gpl-3.0
- Created: 2019-04-09T22:32:04.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2021-08-08T14:22:45.000Z (almost 4 years ago)
- Last Synced: 2024-10-19T05:22:37.952Z (8 months ago)
- Topics: cryptocurrency, ethereum-blockchain, game-theory, mechanism-design, simulation, transaction-fee
- Language: Jupyter Notebook
- Homepage: https://mtefagh.github.io/fee/
- Size: 656 KB
- Stars: 6
- Watchers: 0
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.Rmd
- License: LICENSE
Awesome Lists containing this project
README
---
title: "Median Price Auction"
author: "Mojtaba Tefagh"
date: "4/9/2019"
output:
md_document:
variant: markdown_github
---
We first download the data from the Ethereum blockchain.
```{python}
import pandas as pd
import numpy as np
from web3 import Web3, HTTPProvider
web3 = Web3(HTTPProvider('http://localhost:8545'))
class CleanTx():
"""transaction object / methods for pandas"""
def __init__(self, tx_obj):
self.hash = tx_obj.hash
self.block_mined = tx_obj.blockNumber
self.gas_price = tx_obj['gasPrice']
self.round_gp_10gwei()
def to_dataframe(self):
data = {self.hash: {'block_mined':self.block_mined, 'gas_price':self.gas_price, 'round_gp_10gwei':self.gp_10gwei}}
return pd.DataFrame.from_dict(data, orient='index')
def round_gp_10gwei(self):
"""Rounds the gas price to gwei"""
gp = self.gas_price/1e8
if gp >= 1 and gp < 10:
gp = np.floor(gp)
elif gp >= 10:
gp = gp/10
gp = np.floor(gp)
gp = gp*10
else:
gp = 0
self.gp_10gwei = gp
block_df = pd.DataFrame()
for block in range(5000000, 5000100, 1):
block_obj = web3.eth.getBlock(block, True)
for transaction in block_obj.transactions:
clean_tx = CleanTx(transaction)
block_df = block_df.append(clean_tx.to_dataframe(), ignore_index = False)
block_df.to_csv('tx.csv', sep='\t', index=False)
```
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo=FALSE, warning=FALSE, message=FALSE)
library(tidyverse)
```
```{r}
tx.raw <- as.tbl(read.csv("tx.csv", sep = "\t"))
```
Before we begin, some plots from the raw data (gas price will be normalized later):
```{r}
ggplot(tx.raw,aes(x=block_mined, y=round_gp_10gwei))+geom_point()+labs(title="Gas Price")
ggplot(tx.raw,aes(x=block_mined, y=gas_price))+geom_point()
```
Now, we throw out the `round_gp_10gpwei` column and divide the `gas_price` by $10^8$. Then we group our gas prices data by the blocks and we compute a summary (`min,median,mean,max`). The blocks are consecutive and their numbers are made to start from zero in order to have a bit more visually appealing plots!
```{r}
tx.summary <- tx.raw %>% select(-round_gp_10gwei) %>%
mutate(gas_price=gas_price/10^8,block_mined=block_mined-min(block_mined)) %>%
group_by(block_mined) %>%
summarise_at(.vars=vars(gas_price),.funs=funs(min(.),median(.),mean(.),max(.)))
```
Here are some plots. `geom_smooth()` uses by default the Local Regression (`loess` for short):
*Loess Regression is the most common method used to smoothen a volatile time series. It is a non-parametric methods where least squares regression is performed in localized subsets, which makes it a suitable candidate for smoothing any numerical vector.*
To show the effectiveness/stability of median over other methods, we plot the data points along with the prediction curve of each (`min,median,mean,max`). Notice the scale of `max` is quite different therefore, although it seems stable its prediction curve has much higher errors than median. See the last two plots to compare the scale of their fluctutations.
```{r}
ggplot(tx.summary,aes(x=block_mined,y=min))+geom_smooth()+geom_point()+labs(title="min of gas prices")
```
```{r}
ggplot(tx.summary,aes(x=block_mined,y=mean))+geom_smooth()+geom_point()+labs(title="mean of gas prices")
```
```{r}
ggplot(tx.summary,aes(x=block_mined,y=max))+geom_smooth()+geom_point()+labs(title="max of gas prices")
```
```{r}
ggplot(tx.summary,aes(x=block_mined,y=median))+geom_smooth()+geom_point()+labs(title="median of gas prices")
```
Here is the max, median, and min statistics summary plot:
```{r}
ggplot(data = tx.raw %>% group_by(block_mined)) +
stat_summary(
mapping = aes(x = block_mined, y = gas_price),
fun.ymin = min,
fun.ymax = max,
fun.y = median
)+labs(title="Summary of gas_prices")
```
Here is how the mean and median and minimum curves compare:
```{r}
ggplot(tx.summary,aes(x=block_mined))+geom_smooth(aes(y=min,colour="min"))+
geom_smooth(aes(y=median,colour="median"))+geom_smooth(aes(y=mean,colour="mean"))+labs(title="min_median_mean")+ylab("min_median_mean")+scale_colour_manual(name="legend", values=c("blue", "red","green"))
```
Here is how all curves compare:
```{r}
ggplot(tx.summary,aes(x=block_mined))+geom_smooth(aes(y=mean,colour="mean"))+geom_smooth(aes(y=median,colour="median"))+geom_smooth(aes(y=min,colour="min"))+geom_smooth(aes(y=max,colour="max"))+
labs(title="Summary gas prices")+ylab("min_median_mean_max")+scale_colour_manual(name="legend", values=c("blue", "red","green","black"))
```