Bill Gross started from a $200 blackjack stake, carried the discipline to the bond desk and grew PIMCO to ~$2T — his '2% rule' is the essence of Kelly.
A bond's default probability is low (p≈99.5%) but its tail loss is large (recovery ~half). Full Kelly computes a very high position; practice must scale it down with fractional Kelly to 'no single position risks more than 2%' — a margin of safety under low-probability estimation error.
Continuous-time Kellytypescript
// f* = (μ − r) / σ²
export function continuousKelly(mu: number, sigma: number, r: number): number {
return (mu - r) / (sigma * sigma);
}
// low default / high tail → full Kelly overstated → fractional caps at 2% per position