Game creation to reinforce time series data and increase assets

example : [login to view URL] 1. Use the time series data and play the game. 2. The robot can only have 2 options. 1. Buy 2. Sell 3. Purchase, sale (short sale) The size of btc is fixed to "1". 4. Reward increased assets 5. Initialize after losing 5 or more consecutive times. 6. Initialize if the lost amount exceeds 1000 yen for 10 minutes. 7. You can short sale. 8. Once selected, I can not do anything for 60 seconds. 9. You may give a penalty if you act within 30 seconds instead of No.7. 10. The environment information shown to the robot is only price, size, Buy or Sell in the past 10 minutes from now 11. You can use any library of GYM or reinforcement learning libraries. 12. I have about 300 environmental information for 1 minute. 13. There are over 50 million environmental information as follows. Time price size Buy 2, Sell 1 2018-12-09 18: 21: 10.07 384198 0.01 2 2018-12-09 18: 21: 10.25 384121 0.24397853 1 2018-12-09 18: 21: 10.32 384168 0.01 2 2018-12-09 18: 21: 10.62 384097 0.0616793 1 2018-12-09 18: 21: 10.73 384201 0.09 1 2018-12-09 18: 21: 10.80 384097 0.85 1 · · ·