neuralhedge.data package

Submodules

neuralhedge.data.base module

class neuralhedge.data.base.HedgerDataset(prices: Tensor, info: Tensor, payoff: Tensor)

Bases: Dataset

Dataset contains data for hedging

Parameters:

  • prices (torch.Tensor)

  • information (torch.Tensor)

  • payoff (torch.Tensor)

Shape:

  • prices: (n_samples, n_steps+1, n_assets)

  • information: (n_samples, >= n_steps , n_features)

  • payoff: (n_sample,)

data

(prices, information, payoff)

Type:

tuple

class neuralhedge.data.base.ManagerDataset(prices: Tensor, info: Tensor)

Bases: Dataset

Dataset contains data for management

Parameters:

  • prices (torch.Tensor)

  • information (torch.Tensor)

Shape:

  • prices: (n_samples, n_steps+1, n_assets)

  • information: (n_samples, >= n_steps , n_features)

data

(prices, information)

Type:

tuple

neuralhedge.data.market module

class neuralhedge.data.market.BS_Market(n_sample=10000, n_timestep=30, dt=0.03333333333333333, mu=0.1, sigma=0.2, r=0.0, init_price=100.0)

Bases: object

Data of BS stock + Bond + European call option

Parameters:

  • n_sample – number of samples

  • n_timestep – number of timestep

  • dt – \(dt\)

  • mu – drift

  • sigma – volatility

  • r – risk-free rate

  • init_price – initial prices

bs
Type:

neuralhedge.data.stochastic.BlackScholes

contigent
Type:

neuralhedge.nn.contigent.EuropeanVanilla

bs_pricer
Type:

neuralhedge.nn.blackschole.BlackScholesPrice

bs_delta
Type:

neuralhedge.nn.blackschole.BlackScholesDelta

bs_price

theoretical Black-Scholes price

Type:

float

For portfolio management, the option part e.g. payoff is redundant.

get_hedge_ds()

Get dataset for hedging

Returns:

hedge_ds (neuralhedge.nn.HedgerDataset)

get_manage_ds()

Get dataset for managing

Returns:

manage_ds (neuralhedge.nn.ManagerDataset)

get_price_delta()

Get dataset for managing

Returns:

bs_price, bs_delta (float, BlackScholesDelta)

neuralhedge.data.stochastic module

class neuralhedge.data.stochastic.BlackScholes(n_sample, n_timestep, dt, mu, sigma)

Bases: object

A class with BlackScholes parameters to generate BlackScholes prices

get_prices() → Tensor
Return type:

self.prices (torch.Tensor)

Shapes:

self.prices: (n_sample, n_timestep+1,1)

neuralhedge.data.stochastic.simulate_BM(n_sample, dt, n_timestep) → Tensor
Return type:

BM_paths (torch.Tensor)

Shapes:

BM_paths: (n_sample, n_timestep+1,1)

neuralhedge.data.stochastic.simulate_BS(n_sample, dt, n_timestep, mu, sigma) → Tensor
Return type:

BS_paths (torch.Tensor)

Shapes:

BS_paths: (n_sample, n_timestep+1,1)

neuralhedge.data.stochastic.simulate_time(n_sample, dt, n_timestep, reverse=False) → Tensor
Return type:

time_paths (torch.Tensor)

Shapes:

time_paths: (n_sample, n_timestep+1,1)

Module contents