[Idea] Introspection of the accuracy of historical data

[kevkevinpal]

Idea

Basically, now we are able to fetch the historical data, but that is kind of useless unless we can evaluate how accurate that data was. I have not read through all of the Augur code but I want to take the value we estimated and then get the number of blocks after that estimate ie (3, 6, 9 ..., 144), then take the lowest value tx (if a parent child relationship, in the case of cpfp take the fee it would have been) and then get the % error.

This way, we can evaluate how good the estimations are in the historical endpoints.

I'll need to formalize this idea before trying to continue with any code.

[laurenshareshian]

We do something like that retroactively in our benchmarking tool (albeit the 5th percentile, not the lowest value):
https://github.com/block/bitcoin-augur-benchmarking