L*******t
发帖数: 2385 | 1 啊，我傻了，不是Markovian的话，总有办法加State Variable搞成Markovian的，比如
你的例子，Price就不是markovian的但是整个系统是马尔科夫的。
然后我当时想到的是，Gyongi Theorem，这个大家都知道，我一下子没转过弯来 |
|
Q*******1
发帖数: 91 | 2 怎么确定是否被学校坑了。给你看一个学校的课程表，你看看会不会被坑呢。
%%%%%%%%%%%%%%%这个是bgsu的课程设置
Fall
MSA 5020 Regression Analysis
MSA 5400 Database Management
MSA 5470 Exploratory Data Analysis
MSA 6010 Decision Optimization
MSA 6701 Analytics Project I
Spring
MSA 5160 Time-Series Analysis and Forecasting
MSA 5600 Business Intelligence
MSA 6440 Data Mining
MSA 6500 Big Data Analytics
MSA 6702 Analytics Project II
Summer
MSA 6450 Advanced Data Analytics
MSA 6600 Project Management
MSA 6703 Analytics Project III
%%%%%%%%%%%%%这个是S... 阅读全帖 |
|
Q*******1
发帖数: 91 | 3 怎么确定是否被学校坑了。给你看一个学校的课程表，你看看会不会被坑呢。
%%%%%%%%%%%%%%%这个是bgsu的课程设置
Fall
MSA 5020 Regression Analysis
MSA 5400 Database Management
MSA 5470 Exploratory Data Analysis
MSA 6010 Decision Optimization
MSA 6701 Analytics Project I
Spring
MSA 5160 Time-Series Analysis and Forecasting
MSA 5600 Business Intelligence
MSA 6440 Data Mining
MSA 6500 Big Data Analytics
MSA 6702 Analytics Project II
Summer
MSA 6450 Advanced Data Analytics
MSA 6600 Project Management
MSA 6703 Analytics Project III
%%%%%%%%%%%%%这个是S... 阅读全帖 |
|
w*********i
发帖数: 77 | 4 X_{t} = \int_{0}^{t}Y_{s}dW_{s}
W_{t} is the B.M. And Y_{t} is adapted.
X_{t} is a martingale due to the martingale representation theorem.
But X_{t} is not Markovian if Y_{t} is chosen to be non-Markovian. |
|
l******i
发帖数: 1404 | 5 Set W(t) to be standard Wiener process.
If X(t) follows that dX(t) = \mu dt + \sigma dW(t),
then X(t) is Markovian but not martingale.
If X(t) follows that dX(t) = (\int_0^t X(s) ds) dW(t),
then X(t) is martingale but not Markovian. |
|
L*******t
发帖数: 2385 | 6 有一个gyongy Markovian projection theorem说任何一个Ito过程都有一个Markovian
的过程使得两个过程的Marginal Distribution是一样的
所以price Plain Vanilla European call put，用啥都一样的。
好的模型需要更多的技术去hack。这个避免不了啊 |
|
r**a
发帖数: 536 | 7 我猜他想说的是：price process is not markovian
e.g.
dS_t=(int_0^t\sigma(v, x_v, S_v)dv)dW_t + drift
dx_v= ...
Markovian |
|
r**a
发帖数: 536 | 8 what r u talking about? ou process is markovian. See below from wiki
From wiki
"In mathematics, the Ornstein–Uhlenbeck process (named after Leonard
Ornstein and George Eugene Uhlenbeck), is a stochastic process that, roughly
speaking, describes the velocity of a massive Brownian particle under the
influence of friction. The process is stationary, Gaussian, and Markovian,..
." |
|
G******i
发帖数: 5226 | 9 ☆─────────────────────────────────────☆
MadMoney88 (MadMoney88) 于 (Sat Jun 2 12:23:14 2012, 美东) 提到:
一般不怎么来这个版，没想到今天来看到一群码工自我感觉过于良好。实在受不了一些
人一叶障目，不清楚真相就跟着颠倒事实。就是因为这些井底之蛙的言论，国内很多不
明状况的人才一直有错觉，似乎毕业后去大公司的才是最好的出路。首先，拿钱作为比
较工作的唯一标准就很偏激。其次，你们有多少人自己有过AP的经历，或是家人朋友是
AP，你们所述的AP信息有多少又是道听途说？
发这个帖子没别的意思（绝非BSO），就是想跟大家介绍一下一些相对少数的AP情况（
绝非那些混在3流teaching school的）。本人今年30，刚phd毕业没多久，目前在美东
一私立名校（top5）做AP，专业不是商学院中最热门的finance或accounting，起薪是
20W+每年，外加充足研究经费（平均每年美国内会议旅行5次左右，全球1-2次），无项
目申请压力，招phd学生全由学院出资fellowship。暑假... 阅读全帖 |
|
t******e
发帖数: 673 | 10 The index price does not follow a mean reverting Ornstein-Uhlenbeck process.
Contrary to Naive experience.
SPX, NDX are not stationary, nor Markovian.
However the index log return are approximately stationary and Gaussian.
If we assume the index log return follows Ornstein-Uhlenbeck process. The
return will catch up to historical means and likely overshoot.
Given the low return 2014-2016, 2000-2016. This is a volatility breakout and
we must catch the train. |
|
|
s****d
发帖数: 83 | 12 。。。。吾记得左。。。。
吾好意思。。哈哈哈哈哈。。
mm1 and mg1都是数学模型，
X/Y/N/s/q
X---type of arrival process
Y type of server process
N number of servers
s number of places in queue
q number of sources..
e.g. MM1
既是 both arrival and server process r Markovian
, 1 servers.
好多公式。慢慢查拉 |
|
q********n
发帖数: 355 | 13 本人不是统计专业，用multiple linear regression分析了一些观测结果。改文章时，
遇到一个问题。
1.有一个数据，随时间变化的曲线不好，无法用线性或者其它curve来fit。主要原因是
，数据的变化比较乱，得不到逐渐减小的趋势。我在文章中说no smooth decreasing
curve can be fitted using X.Reviewer认为这种说法太vague，请问要如何讲比较好。
2.我在introduciton里面的原话是“Regression models and neural networks are
deterministic while Markovian models are probabilistic. Deterministic
methods use models from which performance is predicted as a precise value by
mathematical deterioration functions, whereas probabilistic models utilize
a... 阅读全帖 |
|
j******a
发帖数: 1599 | 14 我们学校的数据库没有老的paper。那位能帮个忙，发到j******[email protected]
包子答谢了。
International Journal of Control, Volume 48, Issue 2 August 1988 , pages 481
- 498
Controllability, observability and discrete-time markovian jump linear
quadratic control
Authors: Yuandong Ji a; Howard J. Chizeck a |
|
i******e
发帖数: 171 | 15 I have a 100,000 x 100,000 matrix M which is a nonsymmetric markovian
matrix. I want to get the exp(A). I've got stuck with this problem for a
long time. Now I have to go back to it and still have no good solution.
Is there anybody dealing with this kind of problem as well? Thanks. |
|
m******t
发帖数: 4077 | 16 no, I think american call no advantage in early exercise is due to there is
no arbitrage opportunity, the examples in Hull's book is clear, it has no
relationship to markovian property of the underline equity.
But for put, I don't really know. |
|
a*z
发帖数: 294 | 17 Can anybody solve it in Markovian chain? |
|
D*****a
发帖数: 2847 | 18 没啥意义啊
基本是Markovian
filter？ |
|
J**i
发帖数: 166 | 19 the property is called Markovian, haha, that is a lame question. |
|
l*******l
发帖数: 248 | 20 有平方了，就不是Markovian了，用定义证，开方之后有个正负号问题。
n |
|
x********o
发帖数: 519 | 21 just finished a phone interview 10 minutes ago.
and could not figure out the answer for the following question:
find a process that is martingale but not markovian.
anyone know the answer? |
|
x********o
发帖数: 519 | 22 I said something similar, but clearly the interviewer was not satisfied.
if we choose Y_t=max W_t, then it is non-markovian, right? |
|
k**u
发帖数: 60 | 23 The second example is still a Markov Process.
Solve this SDE:
X_t= x_0 e^{-0.5t+W_t}, which is still a markov process, the operator EXP,
preserves the markovian property. |
|
c*********n
发帖数: 128 | 24 第二个用Markovian Chain，就是state稍微多了点，有八个：
0 - A:T, B:T
1 - A:T, B:H
2 - A:T, B:HH
3 - B:HHH
4 - A:H, B:T
5 - A:H, B:H
6 - A:H, B:HH
7 - A:HH
其中3和7是absorbing state，各个point之间的conversion prob都可以写出来，然后
列个8元一次方程组计算最后收敛到7的概率就是X>Y的概率。
P(
another
of |
|
|
s*******h
发帖数: 1361 | 26 Steve Zhu's paper is actually the easy part. The complexity can explode once
you try to implement it in real world. You are about to solve one of the
hardest yet mandatory cross-asset-class problems that each bank faces.
Smaller firms may not be able to set up the problem even if they want to.
- How can you generate scenarios across different asset classes and
instruments in each netting set and across all netting sets? Are simple
difussion processes enough to model individual assets, and are si... 阅读全帖 |
|
c****u
发帖数: 584 | 27
In W.Harrison's book Applied Quantum Mechanis. Also Mode Coupling Theory
by Gotze shows 1/f noise at higher order singularities.
Nonliner feedback+Non Markovian process+perturbative expansion
Hu |
|
q********n
发帖数: 355 | 28 本人不是统计专业，用multiple linear regression分析了一些观测结果。
我在introduciton里面的原话是“Regression models and neural networks are
deterministic while Markovian models are probabilistic. Deterministic
methods use models from which performance is predicted as a precise value by
mathematical deterioration functions, whereas probabilistic models utilize
a transition probability matrix to predict future performance."
有个reviewer提意见说This statement is not correct. Some advanced regression
models can deal with probabilistic... 阅读全帖 |
|
N******n
发帖数: 3003 | 29
by
utilize
regression
基本上是对的， 主要是regression很多时候和Markovian model 结合起来，就很难说
了。 |
|
g*******i
发帖数: 258 | 30 MCMC is usually under Bayesian framework. In MCMC, the first MC refers to
Markov chain, meaning that the samples have the Markovian property.
Specifically, the result in iteration n+1 is fully depent of the result in
iteration n. That is why we need to check convergence of the sample chain
over the iterations.
In ususl Monte Carlo simution, replications are independent. Typically the
summary is to report the mean and standard error. |
|
|
