k*******d
发帖数: 1340 | 1 如果你有1000个stock，要估计一个covariance matrix.要看多久的历史数据来估计是
合理的？如
何保证估计出来的covariance matrix是positive semidefinite的？在实际中如果出现
估算出
来的cov matrix 不是p.d.的怎么办呢？ Jull的书上有简单提了一下，但是太简略了。 | A********a
发帖数: 133 | 2 U need reduce the dimension of the problems, first because stock returns are
likely driven by a few factors, second because there are missing obs for
such data sets. For the first, u need a factor model (Barra, Northfield, APT
, etc), for the latter, u need some robust methods to accommodate missing
observations, e.g., Em algo, Stambaugh's method and others.
To make the covariance matrix spf is easy, do the eigen-decomp, and ... | k*******d
发帖数: 1340 | 3 En... Got some idea.
请问有什么教材或者是资料详细介绍这些的吗？
are
for
APT
missing
【在 A********a 的大作中提到】
: U need reduce the dimension of the problems, first because stock returns are
: likely driven by a few factors, second because there are missing obs for
: such data sets. For the first, u need a factor model (Barra, Northfield, APT
: , etc), for the latter, u need some robust methods to accommodate missing
: observations, e.g., Em algo, Stambaugh's method and others.
: To make the covariance matrix spf is easy, do the eigen-decomp, and ...
| q**j
发帖数: 10612 | 4 sample covariance 就是正定的。如果要fancy，看newey west的经典paper。
【在 k*******d 的大作中提到】
: 如果你有1000个stock，要估计一个covariance matrix.要看多久的历史数据来估计是
: 合理的？如
: 何保证估计出来的covariance matrix是positive semidefinite的？在实际中如果出现
: 估算出
: 来的cov matrix 不是p.d.的怎么办呢？ Jull的书上有简单提了一下，但是太简略了。
| l****o
发帖数: 2909 | 5 try shrinkage estimation. | A********a
发帖数: 133 | 6 sample covariance is too noisy.
Among 1000 stocks, u may have some with less than 1000 obs, better use
factor based method, check out BARRA document, if u want fancy, go RMT. | X*****r
发帖数: 2521 | 7 cov matrix可能不是PSD吗？
【在 k*******d 的大作中提到】
: 如果你有1000个stock，要估计一个covariance matrix.要看多久的历史数据来估计是
: 合理的？如
: 何保证估计出来的covariance matrix是positive semidefinite的？在实际中如果出现
: 估算出
: 来的cov matrix 不是p.d.的怎么办呢？ Jull的书上有简单提了一下，但是太简略了。
| w****i
发帖数: 143 | 8 PCA or factor model to reduce dimension.
Should covariance matrix always be positive semidefinite?
【在 k*******d 的大作中提到】
: 如果你有1000个stock，要估计一个covariance matrix.要看多久的历史数据来估计是
: 合理的？如
: 何保证估计出来的covariance matrix是positive semidefinite的？在实际中如果出现
: 估算出
: 来的cov matrix 不是p.d.的怎么办呢？ Jull的书上有简单提了一下，但是太简略了。
| z****g
发帖数: 1978 | 9 shrinkage method: linear combination of PCA/Factor based cov and sample
based cov | k*******d
发帖数: 1340 | 10 恩，我查到几篇paper讲这个的了
【在 z****g 的大作中提到】
: shrinkage method: linear combination of PCA/Factor based cov and sample
: based cov
| l****o
发帖数: 2909 | 11 The best paper series are respectively by a Deutsche Man called T G Anderson@Duke
regarding realized covariance, a british man called neil sherpard@oxford,
and two swithland man called Olivier Ledoit and Michael Wolf@Zurich. | b***k
发帖数: 2673 | 12 hey, AlphaNBeta,
May I ask what is APT stand for here?
thanks.
are
APT
【在 A********a 的大作中提到】
: U need reduce the dimension of the problems, first because stock returns are
: likely driven by a few factors, second because there are missing obs for
: such data sets. For the first, u need a factor model (Barra, Northfield, APT
: , etc), for the latter, u need some robust methods to accommodate missing
: observations, e.g., Em algo, Stambaugh's method and others.
: To make the covariance matrix spf is easy, do the eigen-decomp, and ...
| A********a
发帖数: 133 | 13 1. APT is a risk/optimization system bu Sunguard http://www.sungard.com/apt/learnmore, people use APT/Northfield/Axioma/Barra to measure and control their factor/risk exposure. All these systems provide risk/covariance estimation.
2. There have been different approaches to estimate covariance/correlation
matrices, factor-models (above), Bayesian shrinkage estimations (Ledoit &
Wolf), high-frequency (realized vol, realized covariance matrix), dynamic
measures (DCC-GARCH, Exponential weighting, etc), implied vol and
correlations from options, robust estimates, etc.
3. Since asset returns (return/risk) may not captured by mean-covariance
paradigm, all these measures need to be taken with a pitch of salt, how much
. It is better taking a holistic approach to return/risk measure depending
on ur situation, e.g., leverage, liquidity, crowdness, tail risk. |
|
