To implement deep learning algorithm in option pricing and compare the result with Black Scholes formula for calculating options.
- Options are high risky derivatives that give option buyers the right to buy/sell a security at price on or before maturity.
- Price of options is also called premiums which consist of sum of its intrinsic and extrinsic value.
- Intrinsic value is the amount of money received immediately if an option were exercised less than strike price.
- Extrinsic value is remaining value which exceeds the intrinsic value in premiums
- Creating Dataset - We randomly generated data of 300,000 option calls.
- Training Set - Comprises of 2,40,000 call prices.
- Validation set - 60,000 prices.
- Pricing function is linearly normalized i.e,C(S, K)/K = C(S/K, 1)
- Finally the normalized data is fit into deep net.
-
The size of input is 6 parameters
-
Passed with 4 hidden layers consisting of 100 neurons each.
-
At each layer we have used different Activation function according to the inputs that are
- ELU
- ReLU
- LeakyReLU
-
The loss function used is MSE
-
Epoch = 30
-
Batch size = 100