Onther
Published in

Onther

Ethereum Gas Price Analysis

This Analysis examines The Time Series of Average Gas Price with considering Transactions, As the Ethereum Network traffic.

The Average Gas Price data provided by Etherscan.io and On-chain data from Geth full node, operated by Onther.Inc.

At the end of the article, I will describe the equity of the Ethereum network users and how the gas price can be seen as a measure of network demand.

1. Daily Analysis

1.1 Average Gas Price

Let’s look at the data provided by the Etherscan [Average GasPrice Chart] (“https://etherscan.io/chart/gasprice”).

[csv file download link]

The distribution of Average Gas Price is assumed that normal distribution.

By plotting 30-day moving average of gas price and with it added the ±2 * Standard Deviation(Sigma) of the 30-day period,

, We cannot ignore the highest average gas price in December 2016. It was so high GAS price used in, quite interesting peak, so let’s take a look. (Can look weired on red line, below zero in graph. It simply calculated -2Sigma with MA. so ignore it)

2770908, Most highest fee in blocks

Block 2,770,908, Block Reward was calculated almost 766.6 Eth, this block mined by DwarfPool_1 pool.
There was only one transaction in this block.
0xeda3bf566ae21fbc7bbfcb805d73181b0037036d700d57695e28f1e9bfe0d69c In block details, Gas Price was 36,267,774.588438875 Gwei. howdy

Not sure, Tx Sender’s Fat Finger Error. but DwarfPool_1 returned the fee, 761.6 Ether to Tx sender 0xf0f3589042f8e74b1c060fe9aa8e320c8d907708.
DwarfPool_1 returned the applicable fees

1.2. Basic Statistic Result

Let’s take a look for a while what it means for ±2sigma in the normal distribution.

If a data sample is Normal Distribution, mean added ±2Sigma points are the likelihood of occurrence is 2.5%. That means 95% of the total data sample between +2Sigma and -2Sigma.

Normal Distribution [Wikipedia]

The Tx Fee 761 Eth in 1.1, It is obviously belongs to more than +3Sigma of the total Gas Price distribution.

Going back to the beginning of this article, look at the days that were above / below the + -2Sigma of the 30-day moving average of the Average Gas Price on the whole time frame.

Days Over/Under +-2Sigma

The total analysis period is 1,135 days (7/30/2015 ~ 9/6/2018), of which 73 days, 6.4% of the total, exceed the + 2sigma point. And under -2sigma periods are 15days, 1.3% of the total.

To put it simply, above “high” gas price event occurs 6 days out of 100 days.

2. Tx Base Analysis

2.1. Under / Over Gas Price on 30 Days MA

We have to figure out the reason why Gas Price goes up or down suddenly through the Tx in blocks.

Look at transactions more details when the average gas price suddenly rises & falls.

The graph below, Mean of Gas Price ±2Simga is represented in red lines and the 30-day moving average of the time series in the orange one.

Tx data was extracted from Etherscan.io between April 2017 to September 2018.

In December 2017, Excessive transactions were pushed into CryptoKitty contract, and gas price also increased sharply. In the meantime A vote on the increase of the block gas limit was conducted in an Ethereum miner group. As a result, The Block Gas Limit increased to almost 8 MM, like now.

Here is a link to the current poll status for Block Gas Limits: https://www.etherchain.org/tools/gasLimitVoting

Block Gas Limit Dramatically Increased Around DEC 2017

In this analysis, Two cases (Over Gas Price, Under Gas Price) were selected within the same block gas limit.

The sample interval was selected from the period when it was out of the range of ±2sigma on the 30-day MA.

The sample period of “over gas price” is set as January 5–6, 2018,
The sample period of “under gas price” is set as March 10 to 11, 2018.

Looking at the Txs in both cases,

During the “Over Gas Price” period, A total of 2,497,210 transactions were occurred. The average gas price per Tx was 90.12 Gwei in this period.

Another “Under Gas Price” period, A total of 1,199,883 transactions were occurred. The average gas price per Tx was 14.30 Gwei in this period.

In “Over Gas Price” period, The gas price was five to six times higher than in “Under Gas Price”. And There are almost double amount of transactions included blocks in “Over Gas Price” than “Under Gas Price”.
This implies that the proportion of 21000 GAS Tx, which is the minimum GAS on Tx, was high in “Over Gas Price”.

2.2. 21K Tx & None-21K Tx

21000 GAS(hereinafter referred to as 21K) is the minimum amount of GAS used to transfer Ether. And The fee is calculated as GAS * Gas Price. Tx using Smart Contract needs greater than 21K GAS.

In the “Over Gas Price” section, As think of CryptoKitties, We can expect that the percentage of None-21K Tx, using Smart Contract, would be relatively high compared to the “Under Gas Price” section, and the 21K Tx ratio would be relatively low.

Let’s see how the 21K Gas Limit Tx has portion in each of the two sections, “Over Gas Price” and “Under Gas Price”.

Between January 5 and 6, 2018, “Over Gas Price”, 21K Tx was 1,013,869, accounted for 40.6% of 2,497,210 of total Tx.
Between March 10 and 11 2018, “Under Gas Price”, 21K Tx was 230,366, accounted for 19.2% of the total Tx 1,199,883.

In “Over Gas Price”, the number of 21K Tx is more than twice the “Under Gas Price” section. If there was numerous of None-21K Tx due to usage of Smart Contract, the 21K Tx weight would have been rather lower, but It is not.

If so, let’s see where the 21K Tx and None-21K Tx are headed. (I took some known names of contract address from Etherscan. Raw addresses are unkown)

Most received addresses in the “Over Gas Price” section

Most received addresses in “Under Gas Price” section

There is a little bit tweaky point here: Tx to the exchange like Binance_1 at the destination address occurs with None 21K Gas. [“Etherscan Binance_1 Sample Tx”] (https://etherscan.io/tx/0x8bc5f4bfbffb52e0db1fa01586b486384509c6be6be81ed81b9983096d527409)
You can see that the Tx Gas Limit was set to 42K but the Gas Used only 21K.

We can see that None 21K Txs are mostly used to DEX, ICO, and Dapps such as CryptoKitty.

2.3. Gas price Time Series

Let’s compare the gas prices of the major addresses of None 21K Tx and the gas prices 21K Tx in early January 2018, the gas price rose sharply period. Draw the graph between None 21K Gas price Txs used, headed, in EtherDelta, CryptoKitty and EOSTokenContract and the gas price of 21K Tx.
(21K Txs are first used Account, excluded twice.)

On Jan 2 2018, None 21K Tx used 12.9 Gwei as Gas Price. And then The gas price of None 21K Tx increased faster than 21K Tx’s Gas Price as contract usage increased after January 3. On the other hand, The Gas price of 21K Tx was much higher than after Jan 5.
It might be Tx has occurred on wallet-by-wallet rather than smart contract usage because The price of Ethereum and Volume of transactions were increasing on Jan 5 in crypto market.

It should be focus to see that January 4–5, when the rise in gas prices, which began with the increase in Dapp / ICO / Dex usage. It led to a rise the gas price of 21K Tx, simple transaction.

3. Conclusion

The statistical approach to the Average Gas Price reveals exactly the period of sudden volatility. In both “Over Gas Price” and “Under Gas Price” periods, We found that Tx concentrated to few contract’s accounts, as much as a large part of total block usage.

In the 30-day moving average of Gas Price, a time series that exceeds + 2Sigma during the period of high volatility. 2.3 Gas price Through Time Series, the gas price that we pay in the Ethereum chain is dominated by major DAPP / ICO / DEX I have learned about the possibility that In addition, we saw that the Ethereum network was burdened with the explosion of the price and volume of Ethereum, another factor that increased gas prices from December 2017 to January 2018.

In “Over Gas Price” term, multiple factors effected to raise gas price. Dapp / ICO / DEX’s None-21K Tx has led to the burden of 21K Tx users having to set a higher gas price. We can think about the ideas that DApp / ICO / DEX have to take responsibility.

We can also be seen Ethereum network demand through the Tx-Pool’s Tx waiting to be put into the block. But there is nowhere to get historical Tx-Pool’s status with pending Tx details. so we can only measure the demands for ethereum networks through the gas prices of the Txs in the blocks, and it seems reasonable.

If there is anything you think would be particularly interesting to explore in-depth, or if you have any thoughts or criticisms, please do leave a comment with your feedback.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store