threads = []
for worker in workers: # worker threads
t = threading.Thread(target=worker.work, args=())
t.start() # training
threads.append(t)
# add a PPO updating thread
threads.append(threading.Thread(target=GLOBAL_PPO.update, ))
threads[-1].start()
COORD.join(threads)
# plot reward change and test
plt.plot(np.arange(len(GLOBAL_RUNNING_R)), GLOBAL_RUNNING_R)
plt.xlabel('Episode')
plt.ylabel('Moving reward')
plt.ion()
plt.show()
#env = gym.make('Pendulum-v0')
trading_fee = .007
time_fee = .00724
history_length = 1
generator = get_CSV_data(filename="./test_6.csv")
env = SpreadTrading(spread_coefficients=[1],
data_generator=generator,
trading_fee=trading_fee,
time_fee=time_fee,
history_length=history_length)
while True:
s = env.reset()
for t in range(3455):
env.render()
s = env.step(GLOBAL_PPO.choose_action(s))[0]
from tgym.core import DataGenerator
from tgym.envs import SpreadTrading
from tgym.gens.deterministic import WavySignal
generator = WavySignal(period_1=25, period_2=50, epsilon=-0.5)
game_length = 200
trading_fee = 0.2
time_fee = 0
# history_length number of historical states in the observation vector.
history_length = 2
environment = SpreadTrading(spread_coefficients=[1],
data_generator=generator,
trading_fee=trading_fee,
time_fee=time_fee,
history_length=history_length,
game_length=game_length)
environment.render()
while True:
action = raw_input("Action: Buy (b) / Sell (s) / Hold (enter): ")
if action == 'b':
action = [0, 1, 0]
elif action == 's':
action = [0, 0, 1]
else:
action = [1, 0, 0]
environment.step(action)
environment.render()
# In[104]:
# Running the agent
done = False
state = environment.reset()
while not done:
action = agent.act(state)
for position in environment._positions:
if all(environment._position == environment._positions[position]):
position_name = position
for _action in environment._actions:
if all(action == environment._actions[_action]):
action_name = _action
state, _, done, info = environment.step(action)
for position in environment._positions:
if all(environment._position == environment._positions[position]):
next_position_name = position
print position_name, action_name, next_position_name
if 'status' in info and info['status'] == 'Closed plot':
done = True
else:
environment.render(savefig=True)
# In[ ]: