A zero variance path-stretching biasing scheme proposed for a special case by Dwivedi is derived in full generality. The procedure turns out to be the generalization of the exponential transform. It is shown that the biased game can be interpreted as an analog simulation procedure, thus saving some computational effort in comparison with the corresponding nonanalog game.