/
configuration.py
24 lines (21 loc) · 1.41 KB
/
configuration.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
# Model parameters
model_type = 'multiphase' # Current custom network
mem_capacity = 1000 # Size of the memory replay buffer
# Learning parameters
GAMMA = 0.999 # Scaling significance of reward from the next step
EPS_START = 0.9 # This model uses an epsilon greedy strategy, this sets the initial value of epsilon
EPS_END = 0.001 # Lower bound of epsilon
EPS_DECAY = 10000 # Steps between upper and lower bounds on epsilon
TARGET_UPDATE = 900 # Examples between target network updates (want to keep this ~10x policy_update)
POLICY_UPDATE = 90 # Examples between policy updates
BATCH_SIZE = 2048 # Batch size
# Data/Env parameters
data_path = '../data/crypto/ETC_1s.csv'
asset = 'ETC' # Target asset
minutes_back = 10 # Assumes a 1s sampling rate in raw data, how far to look into the past
period = 15 # Assumes a 1s sampling rate in raw data, how many seconds between market queries
params = ['ask'] # Parameters to consider - currently, we are only setup to look at ask price
report_freq = 10 # Number of profitable trades to accumulate before reporting
# Financial parameters
fee = 0.0 # 0.00075 for Binance - 0% currently because Robinhood is on the scene!
investment_scale = 1000 # How much of the asset we purchase with each trade. Value of 1000 = $1,000 invested with each trade