In game theory and economics maximax is a strategy in which an agent aims for the best possible outcome, without any regard for the risk.
This is in comparison to maximin, the strategy of taking the course of action that minimizes the possible loss for a worst case outcome (maximizing minimal loss; hence maximin).
The maximax strategy is much less subtle; it mandates taking the course of action with the best (maximum) gain, ignoring the worst possible outcome altogether.
The maximax strategy is not generally (if ever) recommended, but rather used as an end point of the continuum, used for examples and as a baseline to modify from; for example, the Hurwicz criterion is a model in which the maximax strategy is weighted by a 'coefficient of optimism', providing a balance between the maximax and maximin strategies.