Agreement between raters who record the frequency at which they would choose options

Say I have 5 raters and 6 choices. For each choice, each rater records a frequency from 0 to 100: the percentage of time they would choose that option. Is there any kind of agreement statistic that can accurately assess the concordance for this situation?

The main problem, I feel, is that each rater’s responses for all 6 choices must sum to 100 and thus each item is no longer independent. If no applicable statistic (or modification thereof) exists, perhaps there is an alternative way to look at this problem?