Hmmmm. For me it returns 758 million hits.
The first link, from gummy stuff, explains it well enough.
Or try googling "square root time" for a higher proportion of relevant links.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
delta gamma neutral calendar spreads
- Thread starter luisHK
- Start date
The first link, from gummy stuff, explains it well enough.
Or try googling "square root time" for a higher proportion of relevant links.
What you need to look up is "root time vega"
Generally you cannot add the vegas of two options with different expiries. One way to do it is to normalise the vega(s) by dividing with the square root of time. Then you can weigh your calendar to be vega neutral.
My turn to ask a question:
This makes some assumptions about volatility. For example that there is no mean reversion. Is that something you need to consider in practice?
Generally you cannot add the vegas of two options with different expiries. One way to do it is to normalise the vega(s) by dividing with the square root of time. Then you can weigh your calendar to be vega neutral.
My turn to ask a question:
This makes some assumptions about volatility. For example that there is no mean reversion. Is that something you need to consider in practice?
Quote from kapw7:
What you need to look up is "root time vega"
Generally you cannot add the vegas of two options with different expiries. One way to do it is to normalise the vega(s) by dividing with the square root of time. Then you can weigh your calendar to be vega neutral.
My turn to ask a question:
This makes some assumptions about volatility. For example that there is no mean reversion. Is that something you need to consider in practice?
Interesting question. Natenberg seems to think that volatility of the underlying is mean reverting.
What I have not quite got is why he compares the HV and IV for the period to expiry against the long term mean.
For example if the options have 40 days to expiry, he will look at the 40 day HV and 40 day IV and compare these to the long term mean. I'm used to looking at moving averages for fixed periods, eg 20 day, 50 day, etc, so the significance of locking in on the exact period to expiry is something that I don't get.
I wonder how many traders look at the IV for past periods, as well as the obvious HV which seems par for the course.
Quote from kapw7:
What you need to look up is "root time vega"
Generally you cannot add the vegas of two options with different expiries. One way to do it is to normalise the vega(s) by dividing with the square root of time. Then you can weigh your calendar to be vega neutral.
My turn to ask a question:
This makes some assumptions about volatility. For example that there is no mean reversion. Is that something you need to consider in practice?
I'm not sure wether you're adressing me, as the discussion about root time vega is as much out of my league as before I opened 2 links on the topic.
I would assume Volatility is mean reverting, and consider it when trading it.
Quote from luisHK:
I'm not sure wether you're adressing me, as the discussion about root time vega is as much out of my league as before I opened 2 links on the topic.
I would assume Volatility is mean reverting, and consider it when trading it.
don't worry its out of no ones league... you can't look at one option for jan expiration and one at feb expiration and account for the difference in time.. so you adjust for time..
thats all multiplying by the square foot of time does.. It normalizes the figure to time... brings it back to a unit that is directly comparable between different expiration..
don't be like me.. i sit there and say i don't understand before i even try because i'm scared i'll feel stupid and not get it after really trying hard.. theres no rush to figure everything out. The joy is in the Journey!
Quote from cdcaveman:
don't worry its out of no ones league... you can't look at one option for jan expiration and one at feb expiration and account for the difference in time.. so you adjust for time..
thats all multiplying by the square foot of time does.. It normalizes the figure to time... brings it back to a unit that is directly comparable between different expiration..
don't be like me.. i sit there and say i don't understand before i even try because i'm scared i'll feel stupid and not get it after really trying hard.. theres no rush to figure everything out. The joy is in the Journey!
Square foot of time? Is that how Home Depot measure tile and carpet jobs?
Yes, square root time scaling can be misleading. Diebold et al. address this in the attached paper.Quote from kapw7:
This makes some assumptions about volatility. For example that there is no mean reversion. Is that something you need to consider in practice?
Attachments
Quote from Kevin Schmit:
Yes, square root time scaling can be misleading. Diebold et al. address this in the attached paper.
i think i get the idea that he is trying to convey.. if your taking a averaged number(standard dev/volatility )...your smoothing
(the square root of... the sum of the differences from the mean squared divided by the number of differences-1)
its basically an exponentially averaged measure of the dispersion of prices around the mean
if you look at a yearly historical volatility chart.. it will have smoothed out what you would see on a monthly chart..
so as you aggregate more and more vol the more you smooth your historical vol line looks.. .. this will make history look more linear .. takes the roughness out and smooths any tail events..
so if you took this aggregated number and scaled it down it will over estimate volatility.. partly because most of the time you have small moves.. as the distribution has most likely a high kurtosis with wide tails.. like the expression.. "would you cross a lake that is on average 4 feet deep"
corrections? i'm not sure i'm right.. but thats the gist of what i got.. the other part is about how garch is better at estimating future vol if you have the weights tweaked right.. but it is widely assumed that the maximum likelyhood method is best..