Some points to consider:
1) It is not unusual for there to be a trade off in evolutionary terms. Speed versus endurance for example. Or the enormous energy cost of the human brain. So long as sexual selection confers a sufficiently strong advantage to offset the disadvantage of increased vulnerability it's a win.
2) In birds it is quite common for males to have colourful plumage while the females are much duller. The males are usually less vulnerable (since they do not sit on nests) - and in some species are relatively expendable (if they are not required to help raise the young).
3) We must also not forget that any variation in male appearance that was really problematic would likely be culled from the population before reaching breeding age, so sexual selection doesn't have that great a slope to be climbed.
4) The question would be the spread of the trait rather than the initial acquisition. A preference in itself does not have an evolutionary disadvantage. A drabber mate will do if that is all that is available. An initial variation in the male population is also to be expected. At this stage the trait could spread even by drift. There is no significant disadvantage to the female, and we may assume that the normal variation in male appearance is not greatly problematic either. If the female preference conveyed an evolutionary advantage as well (e.g. brighter appearance correlated with health) then it would have an even stronger tendency to spread.