l********a 发帖数: 126 | |
P****d 发帖数: 369 | 2 implied vol 就是市场根据历史对未来做出的市场预测 |
n*********e 发帖数: 9 | 3 weighted monte carlo?
【在 l********a 的大作中提到】 : 用于预测将来的volatility?
|
l********a 发帖数: 126 | 4 但并不是直接根据historical time series来的。而historical vol是直接从
historical time series来的。
【在 P****d 的大作中提到】 : implied vol 就是市场根据历史对未来做出的市场预测
|
P****d 发帖数: 369 | 5 我觉得historical vol是历史信息的子集吧。 |
q**j 发帖数: 10612 | 6 garch算出来的比较好吧。用Implied vol 不害怕Model错了?
【在 P****d 的大作中提到】 : 我觉得historical vol是历史信息的子集吧。
|
c*******e 发帖数: 150 | 7 yep, agree. for options traders, implied vol sometimes is said to be "
forward-looking", whereas historical vol, no matter how fancy your
econometric method is, is always "backward-looking".
For an easy example, suppose there is this low-beta company whose realized
volatility has been hovering in the low 20s for about 2 months. but its
annual earnings announcement is in 2 days. the implied vol (especially for
super-short tenors) will and should take such scheduled events into account.
on the other hand, IV eventually is determined by supply and demand in the
options mkt, and sometimes due to S/D imbalance, it can be overpriced/
underpriced compared to a fair forecast of the realized vol in the next
period. This is the basic rationale for so-called "vol arbitrage", but that'
ll be the start of a totally different story.
for the purpose of building the second-moment in risk-models, a bayesian
blend of the implied-vol from options mkt and the backward-looking vol using
, say GARCH, may help you attain the most favorable performance.
my two cents :)
【在 l********a 的大作中提到】 : 但并不是直接根据historical time series来的。而historical vol是直接从 : historical time series来的。
|
x********9 发帖数: 31 | 8 No.
But actually, VIX future is probably the best forecast for realized vol(what
you called historical vol). If you regress realized vol on VIX, the
coefficient is probably .8 but R^2 is probably .6 |
l********a 发帖数: 126 | 9 这个 bayesian blend 怎么做?
可否给点reference?
谢谢!
account.
【在 c*******e 的大作中提到】 : yep, agree. for options traders, implied vol sometimes is said to be " : forward-looking", whereas historical vol, no matter how fancy your : econometric method is, is always "backward-looking". : For an easy example, suppose there is this low-beta company whose realized : volatility has been hovering in the low 20s for about 2 months. but its : annual earnings announcement is in 2 days. the implied vol (especially for : super-short tenors) will and should take such scheduled events into account. : on the other hand, IV eventually is determined by supply and demand in the : options mkt, and sometimes due to S/D imbalance, it can be overpriced/ : underpriced compared to a fair forecast of the realized vol in the next
|
c*******e 发帖数: 150 | 10 using implied vols from the options-market as the prior distribution, and
using the physical vol from the filtration as the sample. This method is
effective if you have alpha views trading the underliers and only
need the second-moment for your risk-model. Not effective if you are an
options-trader not having views on the underliers (typically delta-neutral)
but want to earn the "dollar theta minuts dollar gamma" basis-trading
everyday, (in other words, an accurate forecast of spot vol is essentially
your alpha source)
realized
for
the
【在 l********a 的大作中提到】 : 这个 bayesian blend 怎么做? : 可否给点reference? : 谢谢! : : account.
|
|
|
l********a 发帖数: 126 | 11 为什么是implied vol是prior,而不是hist vol?
我觉得反过来似乎更make sense一点啊,hist是prior,implied vols作为当前的
observation。
)
【在 c*******e 的大作中提到】 : using implied vols from the options-market as the prior distribution, and : using the physical vol from the filtration as the sample. This method is : effective if you have alpha views trading the underliers and only : need the second-moment for your risk-model. Not effective if you are an : options-trader not having views on the underliers (typically delta-neutral) : but want to earn the "dollar theta minuts dollar gamma" basis-trading : everyday, (in other words, an accurate forecast of spot vol is essentially : your alpha source) : : realized
|
l********a 发帖数: 126 | 12 另外,vol trading(trade option或者variance swap)里用的vol forecast本质上也
是Bayesian inference吧?
因为除了market-implied和historical data似乎没有其他的information source来预
测vol。而这两个information source,据我所知没有什么单一的model能把它们统一到
一起。所以唯一的把两个统一到一起到的方法就是Bayesian inference了(或者更
sophisticated的Machine learning方法)。
我不做vol trading,不知道以上猜测对不对。
)
【在 c*******e 的大作中提到】 : using implied vols from the options-market as the prior distribution, and : using the physical vol from the filtration as the sample. This method is : effective if you have alpha views trading the underliers and only : need the second-moment for your risk-model. Not effective if you are an : options-trader not having views on the underliers (typically delta-neutral) : but want to earn the "dollar theta minuts dollar gamma" basis-trading : everyday, (in other words, an accurate forecast of spot vol is essentially : your alpha source) : : realized
|
P****d 发帖数: 369 | |
l********a 发帖数: 126 | 14 那应该是怎么样的呢?
【在 P****d 的大作中提到】 : 完全不对。。。。。
|
l********a 发帖数: 126 | 15 有没有人指点一下这个?
【在 l********a 的大作中提到】 : 为什么是implied vol是prior,而不是hist vol? : 我觉得反过来似乎更make sense一点啊,hist是prior,implied vols作为当前的 : observation。 : : )
|
w***e 发帖数: 6 | 16 There is no such thing called forward looking vol. VIX was originally
thought of this kind but later people realized it is more proper to be
referred as a fear index.
One should be careful when he/she uses the implied vol as the indicator. |