Financial Modeling

Kaplan-Knowles Kappa-3 Ratio & Third-Order Lower Partial Moment (LPM3)

Penalize severe left-tail crashes more strictly than Sortino’s quadratic LPM2 by dividing excess return over MAR by the cube root of third-order Lower Partial Moment LPM_3.

How Kappa_n unifies Omega (n=1), Sortino (n=2), and crash-sensitive Kappa-3 (n=3)

Paul Kaplan and James Knowles (2004) showed that Omega and Sortino belong to a single generalized downside-risk family Kappa_n(tau) = (mu - tau) / (LPM_n(tau))^(1 / n), where LPM_n(tau) = (1 / T) * SUM(MAX(0, tau - r_t)^n) relative to Minimum Acceptable Return tau (MAR). Setting n = 2 recovers Sortino’s downside deviation, whereas setting n = 3 cubes every below-threshold loss—making a single -7% crash below a +1% MAR (shortfall 0.08) weigh 8x more heavily per unit of shortfall than four mild -1% dips!

Look at an 8-quarter series averaging +3.00% against MAR = +1.00% with one -7.00% tail quarter (shortfall = 0.08): LPM_3 = 0.08^3 / 8 = 0.000512 / 8 = 0.000064, whose exact cube root is 0.0400, yielding Kappa_3 = (0.03 - 0.01) / 0.04 = 0.5000—materially more conservative than quadratic Sortino Kappa_2 = 0.7071 on the identical return path.

Kappa Order n	Downside Denominator (LPM_n)^(1/n)	Ratio Value (Excess = 0.02)
n = 2 (Sortino / Kappa-2)	SQRT(0.08^2 / 8) = 0.0283	0.7071
n = 3 (Kaplan-Knowles Kappa-3)	(0.08^3 / 8)^(1/3) = 0.0400	0.5000 (Stricter tail penalty)
Zero Below-MAR Quarters	0.0000	"NO_DOWNSIDE_TAIL"

Per-Period Cubed Shortfall Helper Formula

Raise MAX(0, $F$2 - B2) to the 3rd power in column C (and 2nd power in column D for side-by-side Sortino comparison).

=MAX(0, $F$2 - B2) ^ 3

Kaplan-Knowles Kappa-3 Formula with Zero-Shortfall Guard

Average the cubed shortfalls AVERAGE(C2:C9), take the cube root ^(1 / 3), and guard zero-shortfall periods cleanly.

=IF(AVERAGE(C2:C9) = 0, "NO_DOWNSIDE_TAIL", ROUND((AVERAGE(B2:B9) - F2) / (AVERAGE(C2:C9) ^ (1 / 3)), 4))

Try the example

Use the original values or change them to test your understanding.