The formula you posted is correct for the most basic estimate of the next difficulty. Its disadvantage is that the estimate can fluctuate considerably just after a difficulty change and it will only become more accurate as the next difficulty change approaches. Another approach is to always look at the time spent on the last N blocks, where a larger N gives a more stable prediction, but less accurate in times of quickly changing hashrate.
An even more advanced technique would be to assume that the network hashrate is, at least locally, increasing exponentially. You can then take several datapoints with the time spent on solving the last N blocks, the time spent on solving the last 2N to N blocks, 3N to 2N, etc... and fit an exponential curve to this data.
An even more advanced technique would be to assume that the network hashrate is, at least locally, increasing exponentially. You can then take several datapoints with the time spent on solving the last N blocks, the time spent on solving the last 2N to N blocks, 3N to 2N, etc... and fit an exponential curve to this data.
ahh ok .. great thanks for the info.. I need think If I want to go that route.. cheers!
