Hash Ribbons is an indicator by Charles Edwards based on the relative positions of simple moving averages of estimated hashrate. Here we use daily windows of 30 and 60 days. Being below the longer average, a rising shorter average, and its crossing above the longer one are different situations; any price confirmation is a further, separate condition.
Charles Edwards introduced Hash Ribbons as a comparison of shorter and longer simple averages of network hashrate. Glassnode describes windows of 30 and 60 days, which we also use here. Both curves have units of computing power. They are not averages of bitcoin price, dollar revenue or mining difficulty. [1] [2]
Coin Metrics derives hashrate from difficulty and the rate at which blocks arrive. Bitcoin Core also labels the output of getnetworkhashps an estimate over a chosen number of blocks. This limits our interpretation: the series is not a direct reading of every running machine. The provider, estimation-window length and sampling affect the input to both averages. [3] [4]
Glassnode associates the region where the shorter average is below the longer one with a capitulation-phase interpretation. Under our precise rule, a downward crossing occurs only when a previous value at least equal to the long average becomes lower. Merely remaining below it is not another new crossing. Equality is the boundary, not a state strictly above or below. [2]
The author’s description distinguishes recovering hashrate from the end of the marked capitulation region. In our example, an upward crossing occurs when the short average is now higher and was below or equal to the long average on the previous day. The short average can therefore rise for several days while remaining below the long one. This mechanical event alone does not identify which miner switched equipment back on. [6]
Cambridge described a sharp hashrate decline and subsequent recovery following China’s mining ban and the relocation of operations. Our interpretation therefore does not automatically treat a decline in this series as evidence of insolvency or forced selling by a particular company. Further data are needed to explain a change in computing power; the pair of averages alone contains no balance sheet, costs or transactions of individual miners. [5]
Edwards’ article adds positive price momentum to hashrate recovery using price averages over 10 and 20 days. This is a different input series from hashrate. Our chart shows only the hashrate component and implements neither a complete trading strategy nor the author’s state annotations. Reproduction requires a documented version, ordering of conditions and trade-execution rule. [1] [6]
The author’s release notes describe a fix for premature alerts on unclosed data and a change of data source. Glassnode Point-in-Time explains preserving values known in the past. For our test we therefore retain the version, day boundary and input-availability time; we do not automatically replace a missing day with zero. The historical scope of Point-in-Time must be checked for the specific series. [6] [7]
The original article publishes results from a historical selection while also warning that past performance does not determine future results. Our interpretation adopts neither its returns as a promise nor the crossing as a certain buying moment. The pair of averages alone determines neither future price, decline duration nor the probability of profit; those conclusions require a separate test with costs and data available at the relevant time. [1]
Rising, equality, and only then a crossing
For the clearest picture, read this entry together with Hashrate, On-chain Analysis, Moving average, Hashprice, Look-ahead Bias. The reverse links also lead from Miner Capitulation.
01Does a rising short average mean that an upward crossing has already occurred?+
No. The short average can rise while remaining below the long one. Under our rule it must now be higher, while on the previous day it was below or equal. Equality alone is not yet a state above the long average.
02Does a marked capitulation phase prove that miners are going bankrupt or selling?+
No. It is an interpretation of the relationship between two averages of estimated computing power. The series alone contains no liabilities, costs or sales of particular miners. The cause of a change and any price confirmation require further data.