as far as I know spool does not carry any statistics.However, there is some concept called derived statistics, which Iam not sure whether applicable to spool table.Refer some documentation on this...
Confidence levels are determined based on availability of statistics on the columns used by a query.Low confidence indicates that one of the tables are fields involved in the join does not have statistics.Teradata documentation gives more information on confidence levels
Spools are created internally by teradata to hold intermediate step result sets for eg:- redistributions,aggregations,sorts etc.,. As fas as,Confidence levels are concerned we do not always have high confidence steps across the full explain plan. Once the tables rows are redistributed or aggregated into a spool, we do not have high confidence steps anymore.
This means no sooner the spool files are created to redistribution etc. it looses the confidence , then how joins snd further processing happens , is there anything still can be done to improve performance