class TestEnv(TestCase):
def test_sell(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
sell_mount = self.env.step(action='sell', mount=1)
assert sell_mount == {'fail': 0}
self.env.step(action='buy', mount=1)
sell_mount = self.env.step(action='sell', mount=1)
assert sell_mount == {'success': 249999.98500000002}
assert self.env.stock_balance == 0
assert self.env.balance == 249999.98500000002
def test_buy(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
buy_mount = self.env.step(action='buy', mount=1)
assert buy_mount == {'success': 249999.987}
assert self.env.stock_balance == 1
assert self.env.balance == 249888.668
def test_stay(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
stay_mount = self.env.step(action='stay', mount=1)
assert stay_mount == {'success': 250000.0}
assert self.env.stock_balance == 0
assert self.env.balance == 250000
def test_reset(self):
self.env = Env(
balance=250000,
FX_DATA_FILE='../data/raw/FX_Demo/sample_USD_JPY_S5.pickle')
self.env.step(action='buy', mount=1)
self.env.step(action='sell', mount=1)
self.env.reset()
assert self.env.stock_balance == 0
assert self.env.balance == 250000
if action == 0:
real_action = 0
elif action == 1:
real_action = 3
elif action == 2:
real_action = 4
elif action == 3:
real_action = 7
elif action == 4:
real_action = 8
else:
real_action = 11
# 선택한 행동으로 환경에서 한 타임스텝 진행
observe, reward, done, clear, max_x, timeout, now_x = \
env.step(real_action)
if now_x >= 8776:
reward += 200
done = True
if done and now_x < 8776:
reward = -30
reward /= 100
# reward = np.clip(reward, -1., 1.)
print(now_x)
next_state = pre_processing(observe)
next_state = np.reshape([next_state], (1, 84, 84, 1))
next_history = np.append(next_state, history[:, :, :, :3], axis=3)