基于https://github.com/mgroncki/DataScienceNotebooks/blob/master/DeepHedging/DeepHedging_Part1.ipynb和他的RNN网络的工作，我正在尝试扩展此价差期权模型。价差期权的收益基本上在时间T处为max（Underlying1-Underlying2-Strike; 0）。
在扩展和更改此代码时，我确实遇到了一些整形问题。
我记下了每个张量，变量等的所有形状，但我不明白这一点。另外，我尝试将其数量减少，以逐步观察整个过程，而没有任何积极的回报。
这是基于Matthias Github代码的LSTM / RNN代码：
class LSTM(object):
def __init__(self, time_steps, batch_size, features, nodes = [62,46,46,2], name='model'):
tf.reset_default_graph()
self.batch_size = batch_size
self.S_t_input = tf.placeholder(tf.float32, [time_steps, batch_size, features])
self.K = tf.placeholder(tf.float32, batch_size)
self.alpha = tf.placeholder(tf.float32)
S_T_0 = self.S_t_input[-1,:,0]
S_T_1 = self.S_t_input[-1,:,1]
# Returns from Underlying 1 and Underlying 2
dS_0 = self.S_t_input[1:, :, 0] - self.S_t_input[0:-1, :, 0]
dS_1 = self.S_t_input[1:, :, 1] - self.S_t_input[0:-1, :, 1]
#dS = tf.reshape(dS, (time_steps, batch_size))
#Prepare S_t for the use in the RNN remove the last time step (at T the portfolio is zero)
#S_t = tf.unstack(self.S_t_input[:-1, :,:], axis=0)
# S_t wird zu einer Liste bestehend aus Tensoren mit Shape
# Build the LSTM (Multi Zellen) !! Output in nodes = 1 !!!
lstm = tf.contrib.rnn.MultiRNNCell([tf.contrib.rnn.LSTMCell(n) for n in nodes])
# Output == Stragegy
self.strategy, state = tf.nn.dynamic_rnn(lstm, self.S_t_input[:-1,:,:], dtype=tf.float32)
#self.strategy = tf.reshape(self.strategy, (time_steps-1, batch_size, features))
# Optionvalue of the Spread option
self.option = tf.maximum(S_T_0-S_T_1-self.K, 0)
# Maybe Wrong but lets focus on the Shapes
self.Hedging_PnL = - self.option + tf.reduce_sum(dS_0*self.strategy[:,:,0], axis=0) + tf.reduce_sum(dS_1*self.strategy[:,:,1], axis=0)
print(tf.shape(self.option))
print(tf.shape(self.Hedging_PnL))
self.Hedging_PnL_Paths = - self.option + dS_0*self.strategy[:,:,0] + dS_1*self.strategy[:,:,1]
# Calculate the CVaR for a given confidence level alpha
# Take the 1-alpha largest losses (top 1-alpha negative PnLs) and calculate the mean
CVaR, idx = tf.nn.top_k(-self.Hedging_PnL, tf.cast((1-self.alpha)*batch_size, tf.int32))
#CVaR = tf.reduce_mean(CVaR)
CVaR = tf.math.reduce_sum(CVaR)
self.train = tf.train.AdamOptimizer().minimize(CVaR)
self.saver = tf.train.Saver()
self.modelname = name
print('--- DONE ---')
Python将建立模型，但是当我运行trainroutine时，我从Python中得到了Errors。训练程序与Matthias完全相同
def _execute_graph_batchwise(self, paths, strikes, riskaversion, sess, epochs=1, train_flag=False):
sample_size = paths.shape[1]
batch_size=self.batch_size
idx = np.arange(sample_size)
start = dt.datetime.now()
for epoch in range(epochs):
# Save the hedging Pnl for each batch
pnls = []
strategies = []
if train_flag:
np.random.shuffle(idx)
for i in range(int(sample_size/batch_size)):
indices = idx[i*batch_size : (i+1)*batch_size]
batch = paths[:,indices,:]
if train_flag:
_, pnl, strategy = sess.run([self.train, self.Hedging_PnL, self.strategy], {self.S_t_input: batch,
self.K : strikes[indices],
self.alpha: riskaversion})
else:
pnl, strategy = sess.run([self.Hedging_PnL, self.strategy], {self.S_t_input: batch,
self.K : strikes[indices],
self.alpha: riskaversion})
pnls.append(pnl)
strategies.append(strategy)
#Calculate the option prive given the risk aversion level alpha
CVaR = np.mean(-np.sort(np.concatenate(pnls))[:int((1-riskaversion)*sample_size)])
if train_flag:
if epoch % 10 == 0:
print('Time elapsed:', dt.datetime.now()-start)
print('Epoch', epoch, 'CVaR', CVaR)
self.saver.save(sess, r"/Users/floriankonig/LSTM/%s/model.ckpt" % self.modelname)
self.saver.save(sess, r"/Users/floriankonig/LSTM/%s/model.ckpt" % self.modelname)
return CVaR, np.concatenate(pnls), np.concatenate(strategies,axis=1)
def training(self, paths, strikes, riskaversion, epochs, session, init=True):
if init:
session.run(tf.global_variables_initializer())
self._execute_graph_batchwise(paths, strikes, riskaversion, session, epochs, train_flag=True)
输入数据是通过蒙特卡罗模拟生成的，并且具有形状（时间步长，批处理大小，特征）。
我想计算CVaR，但是出现以下错误。
除非所有指定的输入大小都不为零，否则Reshape无法推断出空张量的缺失输入大小