Course Name: Deep Reinforcement Learning in Trading, Section No: 22, Unit No: 4, Unit type: Capstone project. I am running the solution download with the following parameter in my loca machine. It has been running for over 12 hours and has only completed six trade outputs. Config setting are : ', disc_RATE': 0.9,'LKBK': 30,'BH_SIZE': 30,'MX_MEM': 600, EP: 0.0001, EPS_MIN': 0.001,'START_IDX': 5000,'TEST_MODE': False '
Hi Vinay,
For a deep reinforcement learning model, the backtest can take several hours and in certain cases, it might go into days, depending on the computer hardware.
Can I ask what your computer specification is? In reference to the processor, RAM, 32 or 64-bit.
Another point to note is that while there are only 6 trades, it might be that the period of the trade is large. Can you share the trade details as well?
Hope this helps.
Hi Rekhit,
I have a 32 GB RAM. I was just running the capstone project data set and was at least expeting the test mode to be done in a few hours. Instead it took 14 hours.
Trade logs are in the link below
trade_logs_fx_pair_a1.zip - Google Drive
Hi Vinay,
There are certain parameters which can increase the computation time, one of which is the batch size. You can set it to a lower value initially to speed up the processing. Try setting it to 1 as follows, 'BATCH_SIZE': 1 in 'rl_config' and see if there is any difference.
Hope this helps.