100 backtests, is it that there are so many historical records with no transactions exceeding 100? Is this quantitative model a bit obscure?