β° Stock Trader
Hallym Univ. Reinforcement Project
- korea stock market : KOSPI200
- reinforcement learning
π Data
- KRX νκ΅ κ±°λμμμ μμ§
Download
python download --start_date [DATE] --end_date [DATE]
π³ Day Bot
KOSPI200μμ ν¬μν νμ¬λ₯Ό μ νν΄μ£Όλ Bot
Train
python train.py
Test
python test.py --load_path [./checkpoint/YOUR_MODEL]
π© Env
Reward
- ν νμ¬μ νλ₯ κ°
reward = change(CC) * action(one-hot encoding vector)
State
π€ Agent
Model
# lib/agent/agents.py
model = tf.keras.Sequential()
model.add(Conv2D(128, kernel_size=(1, 3), strides=1, activation="relu", input_shape=input_shape))
model.add(MaxPool2D(pool_size=(1, 2)))
model.add(Conv2D(64, kernel_size=(1, 4), strides=1, activation="relu"))
model.add(Conv2D(1, kernel_size=1, activation="sigmoid"))
model.add(Flatten())
Optimizer = Adam