Main Point of Observing the 'Kelly Criterion'
Point out that mining pools never deploy their full hashrate capacity.
Bitcoin's current estimated network hashrate = 325 Eh/s. Two days ago, the network notched a new record at 350+ Eh/s.
Those totals are likely far below what the network hashrate would be if all major mining pools (share >1%) deployed their maximum hashrate. Kelly Criterion is a 'proof' for why this holds true (assuming mining pools are rational, financially motivated actors operating with selfish intent).
Example Proving Kelly
1. Suppose someone gave you $25
2. That money is to be used for gambling on the outcome of a coin flip (heads / tails). This special coin lands on heads 60% of the time, not the normal 50. This is a fact that's known to you beforehand.
3. Each bet is "even-money". So that means whatever you put up is what you get back (i.e., bet $5 and win, then you get your $5 back + another $5, leaving you with $30 total).
4. You have to bet at least $1 each round and there are 300 rounds max (assuming you don't go bankrupt beforehand).
Based on the above, most of you reading have likely already concluded that there exists some "optimal" total you should bet each round, contingent on however much you have to start the round.
The formula used to derive that 'optimal bet' is the "Kelly Criterion".
According to that Wikipedia article I linked above, "The right approach would be to bet 20% of one's bankroll on each toss of the coin, which works out to a 2.034% average gain each round." (that % is based on the geometric mean ROI per round after 300 trials have been completed)
^^ The % of the total available capital you should deploy to optimize returns remains the same. However, the literal dollar amount obviously fluctuates contingents on wins or losses.
Assuming your payouts were always honored (i.e., dealer/house has unlimited money to pay out earnings regardless of bet size), then the "theoretical expected wealth after 300 rounds works out to $10,505".
Note that the 'Kelly Criterion' is backed by a mathematical proof.
Point out that mining pools never deploy their full hashrate capacity.
Bitcoin's current estimated network hashrate = 325 Eh/s. Two days ago, the network notched a new record at 350+ Eh/s.
Those totals are likely far below what the network hashrate would be if all major mining pools (share >1%) deployed their maximum hashrate. Kelly Criterion is a 'proof' for why this holds true (assuming mining pools are rational, financially motivated actors operating with selfish intent).
Example Proving Kelly
1. Suppose someone gave you $25
2. That money is to be used for gambling on the outcome of a coin flip (heads / tails). This special coin lands on heads 60% of the time, not the normal 50. This is a fact that's known to you beforehand.
3. Each bet is "even-money". So that means whatever you put up is what you get back (i.e., bet $5 and win, then you get your $5 back + another $5, leaving you with $30 total).
4. You have to bet at least $1 each round and there are 300 rounds max (assuming you don't go bankrupt beforehand).
Based on the above, most of you reading have likely already concluded that there exists some "optimal" total you should bet each round, contingent on however much you have to start the round.
The formula used to derive that 'optimal bet' is the "Kelly Criterion".
According to that Wikipedia article I linked above, "The right approach would be to bet 20% of one's bankroll on each toss of the coin, which works out to a 2.034% average gain each round." (that % is based on the geometric mean ROI per round after 300 trials have been completed)
^^ The % of the total available capital you should deploy to optimize returns remains the same. However, the literal dollar amount obviously fluctuates contingents on wins or losses.
Assuming your payouts were always honored (i.e., dealer/house has unlimited money to pay out earnings regardless of bet size), then the "theoretical expected wealth after 300 rounds works out to $10,505".
Note that the 'Kelly Criterion' is backed by a mathematical proof.