I applyed stats.boxbox to my data and the returned values are all the same, which seems really unreasonable! it returned this same result in scipy=0.18.1 and scipy=0.17.1. the optimal lambda in my case is -5.501196436791543.
your can download my data(data.txt).
I also tried the boxcox function in R and it returned reasonable result. So i think the boxcox_normmax function should do more test and should be used carefully.