To clarify the quick & dirty estimates .... Each voter indicates which alternatives she considers to be viable, in other words to have more than an infinitesimal chance of actually winning the election. If n alternatives are indicated as viable on ballot B, then ballot B contributes 1/n to the tally for each alternatve indicated. The estimates of the respective winning probabilities are proportional to these tallies. So just normalize the talley vector by dividing it by the sum of the raw tallies.