Kelly Criterion: Likely How Mining Pools Determine How Much Hashrate to Deploy
This article explains the Kelly criterion: https://en.wikipedia.org/wiki/Kelly_criterion
In a nutshell, its a probabilistic theory for determining the optimal "theoretical size for a bet" when expected returns are known.
This model is very likely employed by mining entities (at least at a macro level; perhaps micro as well, but haven't explored that yet). This is based on observations I've made re: hash rate fluctuations & hash rate % changes of various firms while controlling for (a) changes in price (b) changes in hash rate by competing mining pools (c) changes in demand for hash rate [i.e., via mining equip. sales] and (d) changes in the # of pending transactions/average fee rate per transaction.
Known Expected Return
We know what the reward is for winning a BTC block (6.25 $BTC). Based on one's hash rate % out of the network's total, mining entities can derive a very realistic & reliable expectation for how much $BTC they will obtain in a given day (i.e., 144 blocks/day x 6.25 BTC/block = 900 $BTC; mining pool w 20% hashrate can estimate bringing in 180 BTC that day). [ignoring $BTC earned from added fees to transactions, which are non-trivial but nowhere near as influential on decisions miners make relative to block reward].
This article explains the Kelly criterion: https://en.wikipedia.org/wiki/Kelly_criterion
In a nutshell, its a probabilistic theory for determining the optimal "theoretical size for a bet" when expected returns are known.
This model is very likely employed by mining entities (at least at a macro level; perhaps micro as well, but haven't explored that yet). This is based on observations I've made re: hash rate fluctuations & hash rate % changes of various firms while controlling for (a) changes in price (b) changes in hash rate by competing mining pools (c) changes in demand for hash rate [i.e., via mining equip. sales] and (d) changes in the # of pending transactions/average fee rate per transaction.
Known Expected Return
We know what the reward is for winning a BTC block (6.25 $BTC). Based on one's hash rate % out of the network's total, mining entities can derive a very realistic & reliable expectation for how much $BTC they will obtain in a given day (i.e., 144 blocks/day x 6.25 BTC/block = 900 $BTC; mining pool w 20% hashrate can estimate bringing in 180 BTC that day). [ignoring $BTC earned from added fees to transactions, which are non-trivial but nowhere near as influential on decisions miners make relative to block reward].