what are your views on VIX  so that i can start discussion  from there

My views on it are that there is not enough competition out there in terms of alternative models, and so it has not had rigorous theoretical and applied challenge as a model. I'd be interested in hearing more about what people think are the strengths and weaknesses of VIX from an econometric perspective, how/if alternative models handle those weaknesses and what the degree of concordance between VIX and these other measures look like.

I think that it is in need of revision, at least as it appears to me in its 'published' form and also could do with further econometric scrutiny, especially (if I'm right) with the incorporation of the notion of partial Granger-causality, and I think some time spent using the Bradford-Hills criteria for causality should also be considered. 

My view on VIX is it has some merit, of course, but it is not 'a solution' (as nothing is - currently). My interest is in seeing if VIX can more accurately model the volatility of systems because I am very interested in the future of cryptocurrencies and, in particular because of the massive diversity of currencies that have not failed (yet) I'd really like to see how much we can learn by improving volatility indices generally before the cryptocurrency activity stabilizes (as I think it will) with a handful (rather than thousands) of different coins.

In  the paper in attache, an alternative for measuring market stress (model-free measure for implied herd behavior) is presented.

Article The Herd Behavior Index: A New Measure for the Implied Degre...

Jonathan, My view on VIX is it has no merit – it shows the degree of current market heat.

I did some analysis suggesting that the is no predictive power rather looking into the past. And if one valuates options with Black-Scholes and uses VIX as volatility prediction fair values are 30 percent higher than the exercised returns – found by backtesting during 2000 – 2014, daily call options at-the-money. This it would have been pure loss.

Still interested in that subject?

Article Volatility Indexes seem to point to the Past

Sorry, been out of ResearchGate for a while, and when I came back to it I had a million messages, mostly pointless attempts to get 'question ratings' on the ResearchGate system up of no relevance to all. But that's another complaint I have, on the other thread!

I'm certainly interested in the idea that there's potential for improvement, Gerhard, and really interested in your paper. I don't know if no merit (see my comments in my posting) but I certainly think it is dogged by problems and there are issues with all kinds of aspects of the model (except maybe it's tractability) . Again, my question is very much about the shortcomings and a desire to improve what is here. I think the idea is extremely meritworthy, and the need is there, but VIX doesn't do what is needed.

Jan, Yuri again, your papers are very interesting too, and I'd not heard of it. Thank you! Yuri, is this the same thing though?  Jan, this seems to me to address a part of what VIX attempts to, but not the full range of espoused objectives (I think that is the issue, why it clearly fall short)..

Anurag, you didn't follow up with your views! Cheat :D

Folks, I would be happy for a comment whether the idea to by S&P calls at fair value with VIX as volatility and execute them during 15 years = 3750 (bank) days and compare the totals sounds feasible to  verify the B&S formula and the VIX?

In this case the corrective action should be: Better deduct 30 percent from the suggested fair value (calls). 

For other portfolios find out the individual mispricing percentage and proceed accordingly.

I am currently experimenting to sell the calls in between as soon as there is a defined plus (instead of waiting until execution).

I'd be interested in seeing that happen.

If I've understood it's a proposal going forward, rather than a retrospective of existing (past) data already collected?

Not that I have a problem with that, but presumably VIX has been cross-referenced to that already in calibrating and testing the model/tool?

It's retrospective - backtesting, using the daily S&P500 quotations since 31. 12.1999.

I am issuing day by day calls at-the-money and using both, the VIX and the standard deviation during leadtime and bying them. Then 2260 calls are executed (1485 had no returns).

It's just the B&S formula - no calibrating.