A game you can calculate
A coin lands heads 60% of the time. You choose what share of your bankroll (all the money you have for betting) to bet on each flip. Heads, you win what you bet; tails, you lose it. That's even money.
The EV is positive: per $1 bet, 0.6 × $1 + 0.4 × −$1 = +$0.20. But remember the loss math from Module 2: win 20% then lose 20% and you are at 1.2 × 0.8 = 0.96, down 4%.
The typical growth per flip
Over many flips, what matters is the typical growth per flip (the geometric average), not the EV of one flip.
In words: multiply (1 + bet) by itself for the share of flips you win, and (1 − bet) for the share you lose, then subtract 1.
With numbers, betting 20%:
(1.20)^0.6 × (0.80)^0.4 − 1 = 1.1156 × 0.9146 − 1 = +2.03% per flip
In a sheet: bet size in B1, then =(1+B1)^0.6*(1-B1)^0.4-1.
The Kelly fraction
The Kelly fraction is the bet size, as a share of your bankroll, that gives the highest typical growth, if you know the odds exactly. John Kelly published it in 1956; Edward Thorp later used it in blackjack and in markets.
For an even-money bet: Kelly fraction = win chance − lose chance.
0.60 − 0.40 = 0.20 → bet 20% of the bankroll
For other payoffs: Kelly fraction = win chance − (lose chance ÷ net odds), where the net odds are what you win per $1 bet.