Hi there- I’m not sure that your conclusion is correct that there is no difference? It seems highly unlikely that there is no difference given the two distributions- shouldn’t you be comparing the difference of means against a normal distribution centred around 0 with something like a t-test? The comparison you have (null_val != means) is almost always going to be TRUE hence you get an average value of 1.0.

Seems like you want something like :

null_val = np.random.normal(0, np.std(means), 10000)

scipy.stats.ttest_ind(null_val, means)