Check it out -- a new OSF data collection of option-implied variables based on several published/ working papers: implied betas, implied correlations, generalized lower bounds for expected excess simple returns, and model-free implied skewness. Very useful if one wants to replicate papers in questio...

A new paper on recovering subjective physical and risk-neutral bivariate distributions from SP500 and VIX options; the only identifying assumption is the sentiment of an agent (the most probable, according to the agent, region of the future SP500 and VIX) + a good-deal bound restricting the likeliho...

<t>Good time of the day! Appreciate your opinion on the following topic. it is well-known that skewness in return distribution, e.g., risk-neutral distribution implied by the option prices, is the consequence of the leverage effect, or asymmetric volatility, i.e., it follows from the fact that varia...

Paper on identifying implied correlation between VIX and SPX link to SSRN here! -- would appreciate your ideas on the applications! is it good for pricing? risk management? Thanks,G

<t>We just revised our old implied correlations paper from 2005 that looks at the implied correlations for SP/DJ. It looks like the implied correlation is much more stable in predicting market returns than the variance risk premium. Interestingly, both measures (VRP and IC) are not substitutes, i.e....

<t>Thanks, acastaldo, it might be interesting, but I would be afraid of just creating another data-mined instrument as well... VIX and IC are very much related, and while VIX can be used directly as a moment of market return (as well as SKEW, or, rather implied skewness before that CBOE transformati...

Thank you, gardener3, what do you consider an acceptable R2 on the monthly data? BTW, in Bollerslev's paper we talk about in-sample R2... out-of-sample it would be quite a bit lower..

<t>Yes, indeed, I mean the Bollerslev, Todorov and Zhou (Review of Financial Studies, 2009), and the link for SSRN version is here -- however, all the tests are in-sample, and hence do not show much in terms of real predictability.. I just posted a paper ---again on SSRN---, where I try a small out-...

<t>Appreciate your input. There are many option-implied indices nowadays -- VIX, SKEW, IC,.. one can compute implied variance of the VIX itself, and many other quantities. Question: putting away all arbitrary and non-supported statements like "VIX is a fear index.. buy fear, sell greed, etc.." do th...

<t>QuoteOriginally posted by: FermionBTW the 1/N rule is clearly ambiguous, since there is no reason given for not treating any two assets as a single combined 1/2 asset. Hence those two assets will then have a 1/2(N-1) weighting in the overall portfolio. Deciding which asset or combination of asset...

You will not completely sell the winners/ buy underperformers -- just rebalance slightly by reducing the weight in the winners and icreasing it in the underperforming stocks.

<t>Following multiple questions about 1/N and its superior performance, we have just finished a paper Why Does an Equal-Weighted Portfolio Outperform Value- and Price-Weighted Portfolios?, where we try to analyze the sources of 1/N outperformance formally. It turns out that the main reason for high ...

<t>Hi, we have just finished a new paper on using option-implied information in the mean-variance optimization. Could anybody comment how plausible this study is in real life? Turnover of 50-70% per month for a mean-variance portfolio? What would be the transaction costs for, say, SP100 components? ...

<t>Hi, we started a new project recently, trying to understand how options are produced on the market (as initially they are in a zero-net supply), and what are the main determinants for the bid-ask spreads on the market. I would appreciate your feedback from the practical point of view. Let's assum...

<t>Hi, recently calculated model-free (option-implied) variance, skew and kurtosis using OptionMetrics data - first raw data, and then using the surface data. The model-free variance looks kind of the same, but skew and kurtosis for raw data is about twice bigger (in absolute value for skew surely) ...