The Gamma and Beta Functions

Factorials like $5! = 5\times4\times3\times2\times1$ only make sense for whole numbers — there's no obvious way to compute "3.5 factorial" by repeated multiplication. The gamma function, $\Gamma(n) = \int_0^\infty x^{n-1} e^{-x}\, dx$, fixes this by defining a smooth curve that agrees with $(n-1)!$ at every whole number ($\Gamma(n) = (n-1)!$ for integers) but is also defined everywhere in between and for most other real and complex numbers. In plain English: it's the natural, unique way to "connect the dots" between factorial values with a smooth curve, and it shows up automatically in the formulas for the chi-squared, t, gamma, and Dirichlet distributions precisely because those distributions are built from sums and ratios of things that behave like factorials.

The beta function, $B(a,b) = \int_0^1 x^{a-1}(1-x)^{b-1}\, dx$, is a close cousin that measures area under a flexible bump shape confined to the interval $[0,1]$, and it relates back to gamma through $B(a,b) = \frac{\Gamma(a)\Gamma(b)}{\Gamma(a+b)}$. In plain English: any time you're working with a quantity naturally bounded between 0 and 1 — a probability, a proportion, a win rate — the beta function is what normalizes the shape of the beta distribution used to model it, which is why it's the standard Bayesian prior for a probability parameter.

A quant rarely computes either integral by hand; the value comes from recognizing them by sight inside a distribution's formula, since both are the normalizing constants that make a probability density integrate to exactly one.