Skip to main content

monteCarloTrialsForHoeffdingBound

Hoeffding trial count for uniform MC proportion error bound (any underlying p).

monteCarloTrialsForHoeffdingBound(epsilon: number, delta: number): number

Import​

const poker = require('poker-calculations');

ESM (Node):

import { createRequire } from 'module';
const require = createRequire(import.meta.url);
const poker = require('poker-calculations');

When to use​

Plan Monte Carlo sample size when you need a distribution-free guarantee: with probability at least 1−δ, the estimate differs from the true proportion by at most ε, regardless of the unknown p. More conservative than monteCarloTrialsForStandardErrorBound, which assumes a planning value for p̂.

How to use​

const poker = require('poker-calculations');

const epsilon = 0.01; // ±1 percentage point
const delta = 0.05; // 5% failure probability

const nHoeffding = poker.monteCarloTrialsForHoeffdingBound(epsilon, delta);
const nSe = poker.monteCarloTrialsForStandardErrorBound(0.5, epsilon);

console.log({ nHoeffding, nSeAtPHalf: nSe });

Returns​

number — smallest integer n with n ≥ ln(2/δ) / (2ε²) (Hoeffding bound for a proportion in [0, 1]).