SV = Statistical Volatility SV = HV IV is usually higher than SV (HV), especially for indexes. You buy and sell IV.
SV = HV.....right. Another question for you.....how does HV come into the equation when dispersion trading ?
How about shorting the component calls and hedging with long index calls? If the weighting is right, the worst case of a strong upward move should cover the losses of the components. At expiration, you should earn the IV differences caused by the different skew. Is there a flaw in this reasoning?
That is exactly the strategy that I'm working on. Since I'd be long correlation (short dispersion) the "flaw" is that correlation of the chosen basket stocks will will go out of synch. I have alot more work to do and need some help with the 2nd & 3rd volatility level coefficient calculations..... Anyone ?
Still working on reverse dispersion even though you've been getting killed by the outliers? This may be the time for the reverse since index IV has now become cheap (see chart) The only thing to do is to model this. HV is for backtesting, but IV is needed also. Yes the skew curves sound enticing, but keep in mind (and this may well be the "flaw" in the reasoning) that the IV levels of the stocks are often twice that of the index, so the playing field is not level. Have you got an example of the 1st calc -- the formula I posted? Anyone wanting to form a working group to model this email me.
The trade(s) would be done as close to ATM where possible, so don't follow your thinking wrt outliers. That has been my recent intuition, so I delved a bit deeper.... Looking at the FTSE100, the WtdCompIV is 21.21% with actual IV 11.2%, all taken ATM and from Friday's close. Implied correlation is therefore 0.53 with 1st level Vol corr 1.89. From what I've read, this isn't a good time to run a dispersion trade, so by inference it's a good time to run a reverse. Comments ?
Those are your own words from near the beginning. Outliers start ATM and disperse. You don't want that in the reverse dispersion. I don't follow you on the "therefore". Did you use correlation in the WtdCompIV formula or just a weighting of the comps IV? How are you calculating implied correlation?
I still donât follow you. Maybe weâre talking cross purposes here, outliers in what context ? WtdCompIV ignores correlation between the components. I summed the IV x stockâs weighting. My understanding is that itâs the index IV / weighted components IV. Where the values are the same the implied correlation would be 1, i.e. perfect correlation. For the FTSE100 the values are 11.20 (index) / 21.21 (Components) = 0.53 First volatility level coefficient is the reverse 21.21 (components) / 11.20 (Index) = 1.89 Agreed ?
Go back and read your post where you said outliers were killing you, or maybe that was IV_Trader. If you are short an ATM component straddle and that stock makes a 2 std. dev. move while the index stays the same, that is a big loser for your reverse dispersion. Yes. Totally agree. I went back to the article and now believe that it introduces confusion needlessly. The term "Implied Index Correlation" is misleading because it is not IMO the same statistical correlation that appears later in the article. The IIC is nothing but the ratio of the index IV to the weighted IV of its components, so it should be called the Implied Index Ratio or something like that. Yes, but I don't see the point of having 2 terms that are nothing but the inverse of each other. Confusing.