So each player makes around 70 decisions per game. Averages around 5.9 VPs per game.
Performance Bits
- Creating an array with `np.array` is much faster than `tf.convert_to_tensor` and
`pd.DataFrame` or `pd.Series`.
In [1]: timeit.timeit("np.array(array)", setup="import numpy as np; array = list([i for i in range(1000)])", number=1000)