A coupling argument compares two random quantities by constructing them on one probability space, each with its correct law, so that an inequality between them holds outcome by outcome; the inequality then holds for their probabilities or expectations.
Voorbeelden
Example 11.10 (A better coin wins more often)
Draw uniform on and let toss be a success for coin if . With and every success of the second coin is a success of the first, so “at least seven successes” for the second implies it for the first, and without computing either (they are 0.382 and 0.172). The same construction proves that the ruin probability of Proposition 11.6 falls as rises.