蒙特卡洛模拟
Monte Carlo simulation
题目详情
A. 如何用蒙特卡洛(Monte Carlo)定价欧式看涨?
B. 如果只能生成 ,如何生成 ?
C. 有哪些常见的方差缩减(variance reduction)技巧?
D. 没有闭式解时,如何估计 Delta / Gamma?
E. 如何用蒙特卡洛估计 ?
A. Price a European call with Monte Carlo?
B. Generate if only uniform(0,1) is available?
C. Variance reduction techniques?
D. No closed form => how to estimate delta/gamma?
E. Estimate with Monte Carlo?
解析
A. 在风险中性测度下模拟 条路径得到 ,则看涨期权价格为
B.
- 先用 Box–Muller、接受-拒绝等方法生成 ;
- 输出 。
C. 常见方法:
- 对偶变量(antithetic):用 和 。
- 控制变量(control variate):选一个相关且有已知价格的量来校正估计。
- 重要性采样(importance sampling):在对 payoff 贡献大的区域多采样。
- 矩匹配(moment matching):强制样本均值/方差贴近理论值。
- 低差异序列(low-discrepancy):准蒙特卡洛(quasi-MC)。
D. 用同一组随机数种子做有限差分:
E. 在单位正方形 上均匀采样 ,落在四分之一圆 内的概率为 ,所以用命中比例乘以 4 即可估计 。
Original Explanation
AnswerA: Simulate paths under GBM in risk-neutral measure. The call price is
AnswerB:
- Generate using (Box-Muller, acceptance-rejection, etc.).
- Output .
AnswerC:
- Antithetic: use and .
- Control variate: find a related derivative w/ known price, correct the estimate.
- Importance sampling: sample from a distribution that places more weight in payoffs’ significant region.
- Moment matching: force sample means, variances to match the theoretical distribution.
- Low-discrepancy: quasi-Monte Carlo sequences.
AnswerD:
Finite difference around . Use the same random seeds.
AnswerE:
Sample uniform in . Probability inside quarter circle is . So multiply ratio by 4.