pull down to refresh
I understand, it took me a little while to grasp it at first since I'm not a veteran at this and I'm still learning some things thanks to all of you. / Thanks to what you mentioned, I think I understood this site https://whatthefee.io/
Basically the idea is to keep a record of, say, the last 3 blocks over a given time period, look at the historical percentiles of those 3 blocks based on each sats/vB band in absolute terms, and from there build the percentiles the way you propose using 100% as the range to see the spread of the distribution and thanks to that get a solid idea of what the fee will likely be in the future. Great idea @zeke
As a small token of appreciation, I'm adding you to the supporters section.
https://github.com/Satoshi-Dashboard/project-supporters/blob/main/README.md
(P.S. In case I misunderstood the concept, my apologies in advance if you have more info it would be helpful. But if I got it right, then I've got it down.)
Most underrated metric: the spread between fee percentiles, not the absolute sats/vB.
When mempool.space shows 10th/50th/90th percentiles converging (say 5/6/7 sat/vB), mempool is clearing and fees will drop within 2-3 blocks. When they're spread wide (5/25/80), you have layered congestion — pay the 90th if you're in a hurry, wait 20+ blocks if you're not.
Absolute feerate tells you what happened last block. The percentile spread tells you what the next 3 blocks will cost.