fincore.simulation

MonteCarlo

fincore.simulation.monte_carlo.MonteCarlo(returns)

Monte Carlo simulation engine for financial risk analysis.

Provides methods for: - Path simulation using Geometric Brownian Motion - Risk metric calculation (VaR, CVaR) - Stress testing scenarios

参数：

名称	类型	描述	默认
returns	Series or ndarray	Historical returns for parameter estimation.	必需

simulate(n_paths=1000, horizon=252, *, drift=None, volatility=None, antithetic=False, seed=None)

Simulate future return paths using Geometric Brownian Motion.

参数：

名称	类型	描述	默认
n_paths	int	Number of paths to simulate.	1000
horizon	int	Number of time steps to simulate.	252
drift	float	Annualized drift rate. If None, estimated from returns.	None
volatility	float	Annualized volatility. If None, estimated from returns.	None
antithetic	bool	Use antithetic variates for variance reduction.	False
seed	int	Random seed for reproducibility.	None

返回：

类型	描述
SimResult	Object containing simulated paths and statistics.

示例：

>>> import numpy as np
>>> from fincore.simulation import MonteCarlo
>>> returns = np.random.normal(0.001, 0.02, 252)
>>> mc = MonteCarlo(returns)
>>> result = mc.simulate(n_paths=10000, horizon=252)
>>> print(f"95% VaR: {result.var(0.05):.2%}")

var(alpha=0.05, n_paths=10000, horizon=252, seed=None)

Calculate Value at Risk using Monte Carlo simulation.

参数：

名称	类型	描述	默认
alpha	float	Significance level (0.05 = 95% VaR).	0.05
n_paths	int	Number of simulation paths.	10000
horizon	int	Simulation horizon.	252

返回：

类型	描述
float	VaR at the specified significance level.

cvar(alpha=0.05, n_paths=10000, horizon=252, seed=None)

Calculate Conditional Value at Risk (Expected Shortfall).

参数：

名称	类型	描述	默认
alpha	float	Significance level (0.05 = 95% CVaR).	0.05
n_paths	int	Number of simulation paths.	10000
horizon	int	Simulation horizon.	252

返回：

类型	描述
float	CVaR at the specified significance level.

price_paths(S0, n_paths=1000, horizon=252, mu=None, sigma=None, seed=None)

Simulate price paths starting from given initial price.

参数：

名称	类型	描述	默认
S0	float	Initial asset price.	必需
n_paths	int	Number of paths to simulate.	1000
horizon	int	Number of trading days to simulate.	252
mu	float	Annual drift. Estimated from returns if None.	None
sigma	float	Annual volatility. Estimated from returns if None.	None
seed	int	Random seed.	None

返回：

类型	描述
SimResult	Simulated price paths.

stress(scenarios=None)

Perform stress testing on historical returns.

参数：

名称	类型	描述	默认
scenarios	list of str	Scenarios to apply. Options: 'crash', 'spike', 'vol_crush', 'vol_spike'. If None, applies all.	None

返回：

类型	描述
dict	Stress test results for each scenario.

stress_table()

Generate a summary table of stress test results.

返回：

类型	描述
DataFrame	Formatted stress test results.

from_parameters(mu, sigma, S0=1.0, n_paths=1000, horizon=252, seed=None) staticmethod

Create Monte Carlo simulation from known parameters.

参数：

名称	类型	描述	默认
mu	float	Annual drift rate.	必需
sigma	float	Annual volatility.	必需
S0	float	Initial price/return level.	1.0
n_paths	int	Number of paths.	1000
horizon	int	Simulation horizon in days.	252
seed	int	Random seed.	None

返回：

类型	描述
SimResult	Simulation result.

Bootstrap

fincore.simulation.bootstrap

Bootstrap methods for statistical inference.

Provides non-parametric statistical inference using resampling techniques.

bootstrap(returns, n_samples=10000, statistic='mean', seed=None)

Bootstrap resampling for statistical inference.

Resamples the returns with replacement and computes a statistic on each resample to build its distribution.

参数：

名称	类型	描述	默认
returns	Series or ndarray	Input returns data.	必需
n_samples	int	Number of bootstrap resamples.	10000
statistic	str or callable	Statistic to compute. Can be: - String: 'mean', 'std', 'sharpe', 'median' - Callable: Custom function taking array and returning scalar	"mean"
seed	int	Random seed for reproducibility.	None

返回：

类型	描述
(ndarray, shape(n_samples))	Bootstrap distribution of the statistic.

示例：

>>> import numpy as np
>>> from fincore.simulation import bootstrap
>>> returns = np.random.normal(0.001, 0.02, 252)
>>> boot_mean = bootstrap(returns, n_samples=10000, statistic="mean")
>>> np.percentile(boot_mean, [2.5, 97.5])  # 95% CI

bootstrap_ci(returns, n_samples=10000, alpha=0.05, statistic='mean', method='percentile', seed=None)

Calculate bootstrap confidence interval.

参数：

名称	类型	描述	默认
returns	Series or ndarray	Input returns data.	必需
n_samples	int	Number of bootstrap resamples.	10000
alpha	float	Significance level (0.05 = 95% confidence interval).	0.05
statistic	str or callable	Statistic to compute.	"mean"
method	str	Method for CI calculation: - 'percentile': Simple percentile method - 'bc': Bias-corrected (not implemented) - 'bca': BCa method (not implemented)	"percentile"
seed	int	Random seed for reproducibility.	None

返回：

类型	描述
tuple(lower, upper)	Lower and upper bounds of the confidence interval.

示例：

>>> from fincore.simulation import bootstrap_ci
>>> import numpy as np
>>> returns = np.random.normal(0.001, 0.02, 252)
>>> ci = bootstrap_ci(returns, alpha=0.05)
>>> print(f"95% CI: [{ci[0]:.4f}, {ci[1]:.4f}]")

bootstrap_summary(returns, n_samples=10000, alpha=0.05, seed=None)

Compute comprehensive bootstrap statistics summary.

参数：

名称	类型	描述	默认
returns	Series or ndarray	Input returns data.	必需
n_samples	int	Number of bootstrap resamples.	10000
alpha	float	Significance level for confidence intervals.	0.05
seed	int	Random seed for reproducibility.	None

返回：

类型	描述
dict	Dictionary containing bootstrap statistics for multiple metrics.

Paths

fincore.simulation.paths

Path generation methods for Monte Carlo simulation.

Implements various stochastic processes for generating price/return paths: - Geometric Brownian Motion (GBM) - Jump Diffusion - Heston Stochastic Volatility (future)

geometric_brownian_motion(S0, mu, sigma, T, dt, n_paths, rng=None, seed=None)

Generate price paths using Geometric Brownian Motion.

The GBM model is: dS = mu * S * dt + sigma * S * dW

Discrete form: S(t+dt) = S(t) * exp((mu - 0.5sigma^2)dt + sigmasqrt(dt)Z)

where Z ~ N(0,1).

参数：

名称	类型	描述	默认
S0	float	Initial asset price.	必需
mu	float	Annualized drift rate.	必需
sigma	float	Annualized volatility.	必需
T	float	Total time in years.	必需
dt	float	Time step size in years.	必需
n_paths	int	Number of paths to simulate.	必需
rng	Generator	Random number generator.	None
seed	int	Random seed for reproducibility.	None

返回：

类型	描述
(ndarray, shape(n_paths, n_steps))	Simulated price paths.

gbm_from_returns(returns, horizon=252, n_paths=1000, frequency=252, rng=None, seed=None)

Generate GBM paths estimated from historical returns.

参数：

名称	类型	描述	默认
returns	Series or ndarray	Historical returns to estimate parameters from.	必需
horizon	int	Number of time steps to simulate.	252
n_paths	int	Number of paths to generate.	1000
frequency	int	Number of periods per year (for parameter estimation).	252
rng	Generator	Random number generator.	None
seed	int	Random seed for reproducibility.	None

返回：

类型	描述
(ndarray, shape(n_paths, horizon))	Simulated cumulative return paths.

antithetic_variates(paths)

Generate antithetic variates for variance reduction.

Creates mirror paths using -Z instead of Z, which reduces Monte Carlo variance by approximately half.

参数：

名称	类型	描述	默认
paths	ndarray	Original paths generated with random variates Z.	必需

返回：

类型	描述
ndarray	Original paths concatenated with antithetic paths.

latin_hypercube_sampling(n_samples, n_dimensions, rng=None, seed=None)

Generate Latin Hypercube samples for quasi-Monte Carlo.

LHS provides better coverage of the sample space than pure random sampling, often reducing required samples.

参数：

名称	类型	描述	默认
n_samples	int	Number of samples to generate.	必需
n_dimensions	int	Number of dimensions (time steps).	必需
rng	Generator	Random number generator.	None
seed	int	Random seed for reproducibility.	None

返回：

类型	描述
(ndarray, shape(n_samples, n_dimensions))	LHS samples in [0, 1].

Scenarios

fincore.simulation.scenarios

Stress testing scenarios for portfolio analysis.

Generates extreme market scenarios for stress testing portfolios.

stress_test(returns, scenarios=None, custom_scenarios=None)

Perform stress testing on returns under various scenarios.

参数：

名称	类型	描述	默认
returns	Series or ndarray	Historical returns to stress test.	必需
scenarios	list of str	Predefined scenarios to apply. Options: - 'crash': Sudden market crash (-20% or more) - 'spike': Sudden market jump (+10% or more) - 'vol_crush': Volatility crush (vol drops 50%) - 'vol_spike': Volatility spike (vol doubles) - 'correlation_breakdown': Correlation goes to 1 If None, applies all predefined scenarios.	None
custom_scenarios	dict	Custom scenarios as {name: {params}}. Each params dict can contain: - 'return_shift': Additive return adjustment - 'return_mult': Multiplicative return adjustment - 'vol_mult': Volatility multiplier	None

返回：

类型	描述
dict	Dictionary with scenario names as keys and results as values. Each result contains: - 'stressed_returns': The stressed return series - 'cumulative': Cumulative stressed return - 'max_drawdown': Maximum drawdown under stress - 'volatility': Volatility under stress

generate_correlation_breakdown(n_assets, target_correlation=1.0)

Generate correlation matrix for correlation breakdown stress test.

In correlation breakdown, all assets become perfectly correlated, which eliminates diversification benefits.

参数：

名称	类型	描述	默认
n_assets	int	Number of assets.	必需
target_correlation	float	Target correlation between all assets.	1.0

返回：

类型	描述
(ndarray, shape(n_assets, n_assets))	Correlation matrix with uniform correlation.

scenario_table(stress_results)

Format stress test results as a table.

参数：

名称	类型	描述	默认
stress_results	dict	Output from stress_test() function.	必需

返回：

类型	描述
DataFrame	Summary table of stress test results.