I'm too lazy to write a traditional formatted blog post, so I'll just post our discussion on the subject here. You can read the content directly…
This is an interesting, completely third-party-free price oracle (UTXOracle.py). The idea is to statistically analyze the transactions in each block of Bitcoin (eg. Users usually send BTC transfers totaling a rounded number like $100), and predict the BTC price for that day.
What’s the accuracy? It may be easy to manipulate, but it's really fun.
The creator said “it's very accurate. It's not clear which exchange price is the "right" price. But it's right in the middle of all of them. Feel free to check yourself”.. Limitations: 1. you have to set a min and max first, so you can't predict prices until 2020; 2. you need a large sample of tx, so currently it can only generate daily prices.
This could be a price feed, which would need to be calculated for the cost of manipulation though.
In fact, on Ethereum, using Uniswap price directly is much better... But this is for a pure Bitcoin node, so this is why they use this approach.
UTXOracle.py: https://twitter.com/SteveSimple/status/1704864674431332503
GitHub Repo: https://github.com/Unbesteveable/UTXOracle
Replicated Heatmap for Ether: https://twitter.com/bantg/status/1705260495920341052
Other Discussions: