“FIN Status Report 2017-3-23”版本间的差异

train the model with different stock set, which consist of 10 stocks belonging to 10 industries.
digitize the asset and set it as reward, which makes the training and the target more mathing
clip the gradients, which stabilizes the training
randomize the goal. Its impact remains to be seen

do more test

Try to use Pyalgotrading to test the algorithm.
Start to develop algorithm for in-day market timing.
Did more experiments of ddpg.

Test the ddpg algorithm with stock selection algorithm.
Develop algorithm for in-day market timing.