p******t
发帖数: 228 | 1 如果X是一个Ornstein-Uhlenbeck过程，那么X^2是啥啊？还是一个马可夫过程么？那分
布函数会变成啥样子？
谢谢，做了一个月也没做出来。。。 |
|
m******2
发帖数: 564 | 2 Not exactly
Ornstein-Uhlenbeck has an exact solution |
|
h***s
发帖数: 35 | 3 Hello Boddies,
we use ornstein uhlenbeck as FX LOG return process and used as calibration.
As known issue of least square, reversion speed and mean level are not
accurate, but volatility is good when sample size is not big enough.
Simulation with the calibrated parameters, FX forward rates from market even
are out of 99% scope. Traders think 50% or expection of simulation FX spot
rate curve should be very close to forward rate curve.
In theory, least square method is correct for calibration reve... 阅读全帖 |
|
P*********0
发帖数: 4321 | 4 By NPR
Congressional scholars Thomas Mann and Norman Ornstein are no strangers to D
.C. politics. The two of them have been in Washington for more than 40 years.
They came together in 2006 to write a book about dysfunction in Congress,
called The Broken Branch. But their assessment of Congress today is even
more dire — so dire, they've called their new book It's Even Worse Than It
Looks.
Mann, a senior fellow of the Brookings Institution, and Ornstein, a resident
scholar at the American Enterpri... 阅读全帖 |
|
l****z
发帖数: 29846 | 5 May 23, 2012 by Warner Todd Huston
The Washington Post published a long Op Ed by a pair of think tankers
pretending at both being “centrists” and offering an unbiased analysis of
why politics has gotten so “partisan” these days. The pair also claim they
know how to end this messy partisanship. But what they wrote is a perfect
example of why things have become so polarized, not an example of how to fix
anything.
The authors, Thomas E. Mann and Norman J. Ornstein, published theirs
headlined, “Wa... 阅读全帖 |
|
t******e
发帖数: 673 | 6 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. |
|
r*k
发帖数: 915 | 7 打超级ornstein不是活在裆下么....我都是绕着他的两条腿滚来滚去....看他要泰山压
顶了滚开。
有时候觉得smough比他还难打些，因为smough的泰山压顶和推土机的攻击范围都超大，
经常搞错距离被一下秒掉....ornstein那个雷插插中之后动画炫的不得了我经常以为是
一招秒了，结果摔下来也不过大半血... |
|
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,..
." |
|
t*******a
发帖数: 4055 | 9 《大西洋月刊》| 美国：一个恶人当政的国家
来源：法意读书公众号 作者：诺姆·奥恩斯坦 时间：2017-11-30
81 0 美国 字号:A-A+
分享到：
0 收藏 打印
本文由《大西洋月刊》特约作者诺姆·奥恩斯坦（Norm Ornstein）撰写。他在文中提
到了政府内阁官员滥用税金、特朗普违反总统道德规范、国会漠视程序规则等问题，并
细数了特朗普及其团队种种荒唐的所作所为。他认为，美国如今正由最无能无德的一届
政府所统治。以下是正文部分。
“恶人政治（Kakistocracy）”这个词语首次出现在十七世纪，它起源于一个希腊词汇
，字面意义是指由最坏、最不择手段之人领导的政府。更广泛地说，它可以指最无能的
、最令人生厌的政府。过去的一个多世纪里我们已经不再提到这个词汇，因此很少有人
听说过它。
我和迪翁、汤姆·曼恩一起撰写的新书《特朗普统治下的美国（One Nation Under
Trump）》中大量使用了这个词语。我们正在美国直观地感受恶人政治。最近几周，一
些肆无忌惮的行为已经备受关注。特朗普内阁成员和白宫工作人员花了大量纳税人的钱
包机，有些目的地通过便宜的商业交通就能到达... 阅读全帖 |
|
h*****h
发帖数: 264 | 10 https://bernstein.harvard.edu/pages/PastMembers.html
Former Members
Graduate Students
Srinjan Basu
2006-2012
Cambridge University
Paul Blainey 2001-2007 Assistant Professor, Department of Biological
Engineering, MIT, Broad Institute
Long Cai 2001-2006 Assistant Professor of Chemistry, CalTech
Alec Chapman 2009-2018 Harvard Univerisity, Cambridge, Massachusetts
Chongyi Chen 2009-2014 Harvard Univerisity, Cambridge, Massachusetts
Huiyi Chen 2006-2011 Nanyang Technolog... 阅读全帖 |
|
t**x
发帖数: 20965 | 11 张益唐就是看鹿的经验， 看看人家谢晓亮实验室的人
Graduate Students
Srinjan Basu
2006-2012
Cambridge University
Paul Blainey 2001-2007 Assistant Professor, Department of Biological
Engineering, MIT, Broad Institute
Long Cai 2001-2006 Assistant Professor of Chemistry, CalTech
Alec Chapman 2009-2018 Harvard Univerisity, Cambridge, Massachusetts
Chongyi Chen 2009-2014 Harvard Univerisity, Cambridge, Massachusetts
Huiyi Chen 2006-2011 Nanyang Technological Univerisity, Lee Kon... 阅读全帖 |
|
y***r
发帖数: 16594 | 12 最近国会太差了。经常为了反对而反对，无所作为。给你看一组数据。
A recent study showed that post-2007, with Republicans in the minority,
threatened or actual filibusters have affected 70 percent of major
legislation. In the 1980s, that number was 27 percent. In the 1960s, it was
8 percent. “This level of obstruction is extremely unusual,” says Norman
Ornstein, a congressional scholar at the conservative American Enterprise
Institute. “And the core of the problem is the GOP.” |
|
z****0
发帖数: 3942 | 13 As Thomas Mann and Norman Ornstein put it last year in their book, “It’s
Even Worse Than It Looks,” the G.O.P. has become “an insurgent outlier —
ideologically extreme; contemptuous of the inherited social and economic
policy regime; scornful of compromise; unpersuaded by conventional
understanding of facts, evidence and science; and dismissive of the
legitimacy of its political opposition.”
It's so sad that we have such a small insane and incompetent group of
politicians who are driving this co... 阅读全帖 |
|
L********g
发帖数: 4204 | 14 Edit Page Knight Laurtrec of Carim
Found upstairs in the Undead Parish, past an obscured door you must break
through. Very similar to Yurt, The Silent Chief, found in 3-2(?) in Demons
Souls. Unlike Yurt, however, he kills only one inportant NPC, and you can
undo his evil by completing a side quest.
If you don't free him, he'll free himself at some point after defeating the
gargoyles. Either way you'll find him in the Fireling Shrine waiting in
front of Anastacia. If you did free and talk to him,... 阅读全帖 |
|
L********g
发帖数: 4204 | 15 lordvessel 是在anor londo 拿的，要打完胖瘦boss（Executioner Smough&Dragon
Slayer Ornstein).我现在好后悔当时没有先杀胖子再杀瘦子，瘦子那一身盔甲真tm帅
阿。 |
|
r*k
发帖数: 915 | 16 单纯从伤害来说，数值相同的情况下闪电枪这种一半物理一半闪电的伤害就是最差的..
..因为这游戏你的伤害数字比防御高越多增加幅度就越大，400物理武器打出的伤害基
本都比200物理武器的两倍还多，然后不同属性的伤害对应不同的防御分别计算。所以
200物理200电的伤害肯定不如400纯物理来得疼，防御越高的差别越大。月光大剑和月
光蝶角牛逼的地方也在这里，他们是纯魔法伤害，所以即使数值不高但是打起来比那些
数字高一两百的武器还要疼。
闪电在原版除了ornstein和gwyn之外没有电抗高的，所以衰减的不是太厉害，但是DLC
那些怪一个个电抗都爆高，打上去就知道区别了...
5 |
|
t******g
发帖数: 17520 | 17 本来就说要常换武器，砍人下药，
闪电在原版除了ornstein和gwyn之外没有电抗高的，所以衰减的不是太厉害
你原来说的伤害不够大， 有点矛盾吧
关于elemental 伤害的必要性 我了解的不多， 我知道的是有些地方有elemental确实
比没有好打， 即使另一个武器纯物理伤害一样
我没有打DLC， 所以也不知道电抗增高了
用矛的哪位没有在打DLC， 我只是想说 LAVA 和那位， 还有我， 还有我看的youtube
， 有2个用电矛的，在某一时期， 还是挺好用的， 你就要强调， 这个武器比起其他
就是渣， 然后我们就对于是不是渣的问题展开了讨论， 其实很没有必要， 爽就好了
， 打不动了自然就换了， 对吧？
..
DLC |
|
g****g
发帖数: 1828 | 18 In probability theory, the normal (or Gaussian) distribution, is a
continuous probability distribution that is often used as a first
approximation to describe real-valued random variables that tend to cluster
around a single mean value. The graph of the associated probability density
function is “bell”-shaped, and is known as the Gaussian function or bell
curve:[nb 1]
f(x) = \tfrac{1}{\sqrt{2\pi\sigma^2}}\; e^{ -\frac{(x-\mu)^2}{2\sigma^2}
},
where parameter μ is the mean (location of the pe... 阅读全帖 |
|
f*******y
发帖数: 2368 | 19 官网的确是这么说。“Albert Einstein received his Nobel Prize one year later,
in 1922. During the selection process in 1921, the Nobel Committee for
Physics decided that none of the year's nominations met the criteria as
outlined in the will of Alfred Nobel. According to the Nobel Foundation's
statutes, the Nobel Prize can in such a case be reserved until the following
year, and this statute was then applied. Albert Einstein therefore received
his Nobel Prize for 1921 one year later, in 1922.”
但官网显然是掩耳盗铃，不... 阅读全帖 |
|
t****g
发帖数: 715 | 20 I know this is a long shot: is anyone here who does some work on roots near
unity ?
For example, in Stock's 1991 JME paper "Confidence intervals for the largest
autoregressive root in U.S. macoreconomic time series", the distribution of
the ADF statistic follows a non standard distribution dependent on c, the
local to unity parameter.
I want to replicate tables offered in the paper, which requires simulation
based on the non standard distribution, involoving the Ornstein-Uhlenbeck
process. Stock |
|
|
c*****e
发帖数: 238 | 22 如果没有critical phenomena,那么一般体系中的density correlation都满足这个关系.
相当于brownian particle记忆效应逐渐消失,呈指数衰减.
如果有critical phenomena,那么这就变成power law了. |
|
z**h
发帖数: 224 | 23 How about composition flucutation? around crutucal point, the correlation
becomes infinite, how comes power law?
. |
|
c*****e
发帖数: 238 | 24 composition fluctuation应该差不多吧.
不过在critical point附近,体系的行为由critical point决定,这样当你做
展开的时候,自然会有power law的结果,至于non-analytical exponent,这就需要比较
多的东西了. |
|
v**i
发帖数: 50 | 25 请问大家：Hermite 函数定义成 d^2(y)/dx^2-2x(dy/dx)+2vy=0 的一个解 （v是正整
数时也叫Hermite 多项式），关于它的性质很多特殊函数书上都有讨论，常见的：比如
把解表示成幂级数的形式，或积分的形式，或写成其他常见的特殊函数的变换。但我想
知道
d^2(y)/dx^2+2x(dy/dx)+2vy=0 的解有没有讨论过？
我也翻过一些特殊函数的书，没摸着什么头绪...请高手指点。
(背景：我在写篇论文，里面出现了Ornstein-Ohlenbeck Process，但是我这个随机过
程不是mean-reverting, 而是 diverging的，离中心越远离心越强，所以才产生了以上
的微分方程问题） |
|
|
|
b***k
发帖数: 2673 | 28 ☆─────────────────────────────────────☆
hahaguy (笑笑生，小声笑) 于 (Fri Aug 8 19:14:43 2008) 提到:
求解
dB = -B / (1-t) *dt + dW (0<= t < 1)
B(0) = 0
这里W是一个wiener process
我的结果是
B = W-t*W
结果答案还含有积分
向各位请就解答
先谢了
☆─────────────────────────────────────☆
bladehaze (^_^) 于 (Fri Aug 8 21:01:00 2008) 提到:
It is a Ornstein - Uhlenbeck process, there is a trick to solve this kind of
process. Check Mark Steele's book, or google.
☆─────────────────────────────────────☆
Taikonaut (遨游太空) 于 (Fri Aug 8 22 |
|
c**********e
发帖数: 2007 | 29 Found they have the same SDE. |
|
c****2
发帖数: 31 | 30 For your info. Vascicek model has exact solution as well. |
|
c**********e
发帖数: 2007 | 31 Agree. As the SDE is the same, so is the solution.
Maybe the name is different only because it is used in different scenarios? |
|
T*******t
发帖数: 9274 | 32 OU可负，V非负...
啥时候这俩SDE一样了...
? |
|
w**********y
发帖数: 1691 | 33 楼上说的是CIR吧?
OU是一个非常宽泛的model.
经典论文是2001年Borndoff的Non-Gaussian Driven OU..可以加stochastic
volatility和levy
所以V应该算OU的一种特例吧. |
|
c****2
发帖数: 31 | 34 Even SDE forms are the same, the solution may not be the same. The solution
depends on boundary conditions as well. If the boundary conditions are
different, the solutions are likely to be different.
Also, Vasciek model does have closed-form solution and CIR model does not
have closed form solution.
Hope this clarifies things a little bit. |
|
|
T*****w
发帖数: 802 | 36 多谢zhucai大牛，
网上看了有一堆papers 讨论 Option Pricing in Stochastic Volatility Models
of the Ornstein-Uhlenbeck type的，
看不明白呢. 是不是no effect on option price at first approximation.
严格解还是有影响的？ 要不这么多papers 讨论
比如这个
http://www.cls.dk/caf/wp/wp-108.pdf |
|
|
G*********o
发帖数: 2045 | 38 search for Ornstein–Uhlenbeck process |
|
d*******k
发帖数: 56 | 39 mean reverting process AKA Ornstein–Uhlenbeck process |
|
o******e
发帖数: 1001 | 40 谢谢moonsspring.
其实这个问题是从Ornstein–Uhlenbeck模型转化而来的。OU模型解出来后，我需要实
际的数据去拟合，但是一般的情况下(t-s)是一样的，所以随机项 c*d可以认为是独立
同分布，从而用least square去拟合。如果(t-s)不一样，那随机项就是独立但是不同
分布，在这种情况下应该如果拟合？
TERM |
|
r********9
发帖数: 22 | 41
Ornstein–Uhlenbeck process
我主要是通过中介申，也有一回是有HR到学校来招。 |
|
k*****n
发帖数: 117 | 42 just to name a few:
Burgess PhD Thesis - A Computational Methodology for Modelling the Dynamics
of Statistical Arbitrage
Avellaneda and Lee 2008, Statistical Arbitrage in the U.S. Equities Market
Meucci, Review of Statistical Arbitrage, Cointegration, and Multivariate
Ornstein-Uhlenbeck |
|
o******e
发帖数: 1001 | 43 fanta,你很内行啊！
这个方程是从stochastic differential equation 过来的。原始的模型是Ornstein–
Uhlenbeck process.通常OU模型假定时间步长一样，所有随机项是IID分布，可以用OLS或MLE解决，我现在时
间步长不一样，所以时间也是一个变量。
其实这就是非线性拟合问题，但是随机项不是IID,是独立但不同分布。
我还不知道MLE能不能处理不同分布的随机项。你有什么idea吗？ |
|
g*****u
发帖数: 14294 | 44 用的是MIPS与S&P500的daily chart.
没有用指标，算了一下return correlation，residual return 用经典的Ornstein-
Uhlenbeck process模型逼近了一下。标准的Mean Reversion模型做法。
弱与强是这两个的相对表现。
还凑合算点干货吧。现在，你能否讲讲为什么MIPS是个好PICK。 |
|
w*******o
发帖数: 6125 | 45 Ornstein-Uhlenbeck process模型是SDE的干活?
FA我就不谈了，不是芯片专家，TA的话,我的方法很野鸡，
有点难登大雅之堂，这个贴子里面其实有点提示，
http://www.mitbbs.com/clubarticle/Stockcafeteria/96333_3.html
如果一定要安个名头的话，就叫筹码分析吧(很郎中的感觉吧?).
如果记性好的同学，还记得去年经典一役PMI的话(现场直播)
PMI还在$5的时候，我说$7.5会是非常强的Resistance
可能冲不过去，结果日线收盘在$7.5上的好像就没有，
我个人的经验是，对于个股的强弱分析,还没有找到什么特别
有效的统计上的分析方法，不知道你这个模型去BackTeset
一下上升过程中AAPL的和BIDU有什么样的结论，随便截一段都可以. |
|
