楼主: 万帅
3681 11

[金融经济学] 一道随机微分方程的题目,求论坛里的牛人解答啊 [推广有奖]

  • 1关注
  • 1粉丝

本科生

37%

还不是VIP/贵宾

-

威望
0
论坛币
701 个
通用积分
1.2000
学术水平
0 点
热心指数
0 点
信用等级
0 点
经验
1176 点
帖子
40
精华
0
在线时间
127 小时
注册时间
2011-9-21
最后登录
2024-10-10

楼主
万帅 在职认证  发表于 2014-5-29 12:26:11 |AI写论文

+2 论坛币
k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群
赵安豆老师微信:zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

感谢您参与论坛问题回答

经管之家送您两个论坛币!

+2 论坛币
无标题.png
上面的那道题怎么做呢?或者说思路是什么?最近感觉学习SDE都要被虐哭了
二维码

扫码加我 拉你入群

请注明:姓名-公司-职位

以便审核进群资格,未注明则拒绝

关键词:随机微分方程 微分方程 随机微分 牛人解 SDE

回帖推荐

Chemist_MZ 发表于4楼  查看完整内容

is it E(Y|X) or E(Y/X)? if it is E(Y/X) just applying Ito lemma to Z=Y/X and put dY and dX into it. You will get the process that dZ should follow and the expectation of Z is easily obtained. intuitively, rho stands for the projection of X on Y. The higher the rho, the larger Y's mean is offset, thus smaller E(Y/X).

Chemist_MZ 发表于8楼  查看完整内容

You have to include the second order terms. d(y/x)=1/x*dy-y/x^2*dx+y/x^3 (dx)^2-1/x^2(dxdy) (I may write it wrong, you need to check yourself) Z will be a lognormal, you can express Z as exp(a) where "a" is a normal variable N(mu,sigma^2). so E(Z) will be exp(mu+0.5*sigma^2)

本帖被以下文库推荐

沙发
guowy67 发表于 2014-5-29 13:47:36
代入ITo公式,可以吧

藤椅
万帅 在职认证  发表于 2014-5-29 13:52:40
guowy67 发表于 2014-5-29 13:47
代入ITo公式,可以吧
额,能说详细一点吗?

板凳
Chemist_MZ 在职认证  发表于 2014-5-29 14:39:40
万帅 发表于 2014-5-29 13:52
额,能说详细一点吗?
is it E(Y|X) or E(Y/X)?

if it is E(Y/X) just applying Ito lemma to Z=Y/X and put dY and dX into it. You will get the process that dZ should follow and the expectation of Z is easily obtained.

intuitively, rho stands for the projection of X on Y. The higher the rho, the larger Y's mean is offset, thus smaller E(Y/X).
已有 1 人评分经验 论坛币 收起 理由
见路不走 + 5 + 5 精彩帖子

总评分: 经验 + 5  论坛币 + 5   查看全部评分

扫头像关注公众号“二点三西格玛”衍生品定价与风险管理

报纸
万帅 在职认证  发表于 2014-5-29 19:07:32
Chemist_MZ 发表于 2014-5-29 14:39
is it E(Y|X) or E(Y/X)?

if it is E(Y/X) just applying Ito lemma to Z=Y/X and put dY and dX into ...
谢谢你啊,但是Z是一个二元函数,是用偏导数求dz吗?

地板
Chemist_MZ 在职认证  发表于 2014-5-29 22:38:42
万帅 发表于 2014-5-29 19:07
谢谢你啊,但是Z是一个二元函数,是用偏导数求dz吗?
Yes, the same as you learned in the calculus class.
扫头像关注公众号“二点三西格玛”衍生品定价与风险管理

7
万帅 在职认证  发表于 2014-5-30 10:46:34
Chemist_MZ 发表于 2014-5-29 22:38
Yes, the same as you learned in the calculus class.
IMG_0130.JPG 恩,能不能帮我看一下上面我做的对不对呢?如果对的话,接下来怎么能通过dz求Ez?

8
Chemist_MZ 在职认证  发表于 2014-5-30 10:55:24
万帅 发表于 2014-5-30 10:46
恩,能不能帮我看一下上面我做的对不对呢?如果对的话,接下来怎么能通过dz求Ez?
You have to include the second order terms.

d(y/x)=1/x*dy-y/x^2*dx+y/x^3 (dx)^2-1/x^2(dxdy) (I may write it wrong, you need to check yourself)

Z will be a lognormal, you can express Z as exp(a) where "a" is a normal variable N(mu,sigma^2). so E(Z) will be exp(mu+0.5*sigma^2)
已有 1 人评分经验 论坛币 收起 理由
见路不走 + 5 + 5 精彩帖子

总评分: 经验 + 5  论坛币 + 5   查看全部评分

扫头像关注公众号“二点三西格玛”衍生品定价与风险管理

9
万帅 在职认证  发表于 2014-5-30 12:12:59
Chemist_MZ 发表于 2014-5-30 10:55
You have to include the second order terms.

d(y/x)=1/x*dy-y/x^2*dx+y/x^3 (dx)^2-1/x^2(dxdy) (I  ...
恩,我已经知道怎么做了,实在是太感谢你了!!!!!
但是最后一个问题,为啥要求二次倒数呢?

10
Chemist_MZ 在职认证  发表于 2014-5-30 12:41:46
万帅 发表于 2014-5-30 12:12
恩,我已经知道怎么做了,实在是太感谢你了!!!!!
但是最后一个问题,为啥要求二次倒数呢?
because the quadratic variation of a brownian motion W(t) is t, in a more understandable form (dw)^2=dt.

and you know we write the formula in the differential form but actually it is in the integral form (can be viewed as an abbreviation) . Thus, all the integral with repeat to (dwdt) or (dt)^2 are zero (this can be proved). So they can be discarded. But only (dw)^2=dt is retained,because it is not zero so that this term can not be ignored.

Thus we need to include the second order terms to include the (dw)^2 terms.
扫头像关注公众号“二点三西格玛”衍生品定价与风险管理

您需要登录后才可以回帖 登录 | 我要注册

本版微信群
加好友,备注jr
拉您进交流群
GMT+8, 2026-1-29 10:22