ah that makes more sense now. Thanks for correcting. While selling options has a higher probability of winning more of the time than buying, seems unreasonable to apply that across the board for all scenarios. Just IMO. And it seems risky to sell a strangle/straddle right before earnings for certain stocks, such as GOOG.
The key to understanding this lies in noting that index "vol" has two main components: component vols and component pairwise correlations. Vol risk premia boil down essentially to tail risk premia. When, for example, markets react down big to a news shock, component vol increases slightly, but component correlation increases more, so index vol, which incorporates both component vol and correlation, increases more. Also, I believe single name vols do express risk premia, just not as large as index risk premia. The correlation component of index vol is why index vol is always higher than weighted average single-name vol. This makes the extremely low index vol prints we have seen over the past few years even more remarkable, as component correlation (e.g Kritzman's absorption index) has been on an uptrend for the last three decades.
The theoretical reason is that single name do not have as much of a persistent positioning bias as the index does. Supposedly, that leads to fewer hedger/speculator imbalances and less of a persistent risk premium. That thought process extends beyond equity markets to cross asset space.
hmm, I am just reading a paper on the Kritzman's ratio and it is confusing me how this could be rising while the CBOE implied correlation ratio has been falling for the past several years....
Partially an artifact of how CBOE solves for impled correlation. Simpllifying a complicated double summation which requires a good estimate of the instantaneous covar matrix, by replacing rho[i,j] with a single comon rho, saves a headache but biases the result. That is not a big effect though. Mostly it is because the two measure different things -- absorption ratio is first few eigenvalues divided by the trace, and the implied cor index is a solved-for rho in the index vol equation. While proportion of variance accounted for by the first few eigenvalues has been rising the volatility of the associated eigenvectors has been falling while idiosyncratic vol (of smaller e-vectprs, more or less) has been relatively rising (or at least falling less). I have a paper somewhere that discusses these phenomena in relation to random matrix theory, I'll see if I can dig it up and post it.
I think those are different underlying ideas - explanatory value of the first N factors vs ratio of basket vol to vol of basket. They could diverge pretty far, IMHO.
We can all agree that, the SPX IV is systematically overpriced as stated in the comments above. I'm sure we can all agree the SPX has excess Kurtosis. So why does systematically Buying butterfly's on SPX do so poorly? If the answer is drift, would buying just OTM butterflies be profitable?? These butterfly's are 30 Dte and drift is not a dominant factor during a 30 day period
Sells a 1 month rolling put and call ATM and buys delta 5 calls and puts. Holds a money market fund from the credit received Edit** 5% OTM