<result> <column_name>v1st%</column_name> <family>uncertain</family> <subtype>none</subtype> <confidence>0.35</confidence> <requires_user_confirmation>true</requires_user_confirmation> <sum_meaningful>false</sum_meaningful> <signed>false</signed> <reducers> <primary_reducer>mean</primary_reducer> <valid_reducers>mean, median, min, max</valid_reducers> <excluded_reducers>sum (percent shares do not accumulate across volumes), product (0.0% values would zero out the result and there is no evidence of compounding)</excluded_reducers> <average_semantics>arithmetic_mean</average_semantics> </reducers> <inner_outer_measure> <inner_measure_reducer>n/a</inner_measure_reducer> <outer_measure_reducer>n/a</outer_measure_reducer> </inner_outer_measure> <multiplicative_partner_derivable>false</multiplicative_partner_derivable> <reasoning>The table is PATH with a Date sequence axis, so the gate allows multiplicative. The "%" in the name and the percent-formatted strings point to a ratio or share with an implicit denominator. However, the column is VARCHAR, so all numeric statistics (center, spread, log symmetry) are null, and no reconstruction signal is available to confirm compounding. The values look like percentage-of-total shares (0.0%, 1.9%, 10.0%, plus "-" and blank placeholders) centered near 0, not factors around 1.0. A 0.0% value would also break a cumulative product. The naming signal and the data do not support a ratio_multiplicative call, so I abstain and flag for review. The column needs parsing (strip "%", map "-" and blank to null) and re-profiling.</reasoning> <table_type_conflict>false</table_type_conflict> </result>