Have you looked at the Bloomberg BVOL package? I was on help help yesterday and they showed me some nifty things.
No, I have not (though I have heard that their infrastructure has improved significantly). Maybe I will check it out tomorrow. I do have access to BBG options database and their vols/greeks are fairly accurate. S&P index is a market-cap weighted based of stock and those stocks do pay dividends. The S&P futures basis (i.e. the difference between the spot index and the futures price) exists partly because of the expected dividends. Even though the futures and/or options do not pay dividends, index arbitrageurs make sure that the dividend expectations are in-line.
BVOL might be the same thing. For example, they can plot implied vs realized correlation. The numbers seemed low (.2 realized corr) but my mental calibration might be off.
I do NOT have access to BVOL, and am curious of the precise references for implied and realized you mention! Are they using series IV or ATM IV, or something else? .2 - .28 seems to be close if using Series IV, but is tighter correlation if using ATM IV (for interval samples > a few months). Kinda curious what they are measuring.
When you mean "series", do you mean closest-to-atm fixed strike? I.e. approximately index implied over mean single implied or do you mean something else? It's actually a tricky question, as both implied and realized correlations are multiply defined. For example, if you are using ex-post analysis, you have a choice of realized Pearson correlation (which would have little to do with your dispersion PnL), weighted vol/vol relationship (i.e. replication-inspired correlation) or actual dispersion PnL (which will naturally be weighted by gamma).
.. Something else.. Pardon my lack of terminology precision. My reference to "series IV" is a proxy for the single value typically provided for the specific option chain for a specific Expiration. This is typically a value near that of using the CBOE VIX White paper computation (used in derivation of VIX), and each brokerage will provide "their" number with each Expiration. -- This value is always (for SPX much higher than the ATM IV, and seems to correlate to the iiv of a PUT strike between -22 to about -30 Delta. -- I find less and less value of this metric except if one is actually trading VX futures. After your reference to "tricky" ... I lost traction as you are referring to things that may be above my pay-grade! I did some simple studies to merely observe Actual Realized Volatility of SPX and compared with the above referenced "Series IV" to find the rough correlation of between 20-28% over multiple year and multiple month timeframes. No rocket science, but "seat of the pants" observations. Noting the value of the ATM IV being a much closer fit, and a cleaner representation of where the money is pointing (for Volatility) caused me to change my Forward looking Volatility of the underlying metric to simply the ATM IV using the series with the appropriate timeframe.
Oh, I see. We are discussing the correlation between different stocks in the S&P basket as it's implied by the ratio of implied volatilities. In short, you can take the implied volatility of S&P, divide it by the weighted average implied vol of the component stocks which would give you the market's expectation of correlation (*). Then you can compare it with the actual correlation realized in the market. Sounds easy enough, but the devil is in the details. (*) it's beyond this thread, but if you PM me I can walk you through the derivation and the reasons why it's tricky