NuclearPhynance Archive

Has anybody here ever had experience with the data from Optionmetrics?  What do you think?

i was looking at there stuff a little while back.



their surfaces didn't seem to look that clean, has anyone else noticed this?

i'm not sure what you mean by their surfaces looking clean. but i looked at their stuff a few years ago. i think that there are only two vendors of implied vols for equity options, optionmetrics and ivolatility. ivolatility did not provide prices and did not provide individual strikes; they only provide their standardized surface (i.e., 10% otm, 20% otm, etc) which is interpolated. i did not get an answer regarding what sort of data cleansing or error checking that they did. on the other hand, optionmetrics provides all of the raw data. you can do the interpolation yourself. plus, i think that he does error checking. which is really the biggest problem.

As a preliminary I checked SPX surfaces for 2005 and there were quite a few "blips".



I was assured that all errors had been fixed, but im still concerned that their single stock surfaces would be messier.  Having said that though you are completely correct they do provide the raw data so I could just construct the surfaces myself.

Yes, raw prices will just magically lead to completely smooth surfaces. Especially on the Single Stock side.

I've used ivols surface data in the past.  They have a shorter history, but allow a range from 10 to 90 deltas out to a year.  Another person in the firm used OptionMetrics, I believe the surface was only from 20 to 80 delta.  For what I was using it for, vol index, skew spread and term structure slopes it was fine.  Altho I was not in need of extremely detailed per strike info.  Btw, the data I had were about 400 US single stock names within a convert universe.

For the most part I have been happy with their data.  For some periods - like 2002 summer for example - I found skew moves in the SPX that were simply not possible.  All the calls would have vol go up and all the puts would have vol go down - the next day the reverse would happen...and this nonsense would go on for a few months leading to a very choppy behavior in back tests on collared positions.  Crap data I say.  The prez of option metrics claims that this behavior might be a result of the 4:15 close for futures trading on the SPX and the fact that my data set only gives me cash values of the SPX spot.  I could buy his answer if the moves were small, but the P&L swings were huge using his data - and quite predictable - if skew moves were guaranteed to be huge and alternating in direction on a daily basis, then we would all be rich.



Other than this SPX data and the fact that he does not handle odd splits well, I like the data.

of course, constructing the surface yourself does not provide a smooth surface. but seeing a smooth surface without the raw data does not mean that what is given matches reality. in fact, this relates to what Kurt said regarding the range of strikes/deltas. optionmetrics provides all of the available strikes as given by the exchanges. so if that is only to 20 delta options, what exactly is it that ivolatility is providing? some sort of an extrapolation. personally, i'd rather see a bumpy surface that represents actually mid-market quotes. i don't know how ivol's data cleansing depends on the same assumptions used to construct the surface. i use the ivol web site and often see puts and calls of the same strike at wildly different vols. of course, there seems to be an und price issue or an interest rate issue, but still that should not be a regular "feature" of the data.



btw, omd, unless i misunderstood your problem, it is likely that what you are looking at for the errors that you saw is an incorrect underlying. try using a few atm combos as the underlying price.

I should mention that ivolatility sells both option prices + vols for all individual option contracts on a ticker and implied volatility surfaces. While I can't comment on their implied volatility surfaces, the prices on individual contracts appear reasonable to me on financial sector single stocks and ETFs. I personally would rather use my own inputs for backing out the implied volatility.



This may seem really stupid, but given that these are American options, do the implied vols technically have to be the same on puts and calls? I understand that one might want a model that enforces this, but if the expected exercise time is very different than the vols would be different, no? Also, if you build in some of the frictions this would also push the vols apart.

dgn2 -- not stupid. the model should take care of most of it, except for exercises and borderline exercise. i was referring to near money options with radically different vols. like 20% call and 40% put. i agree frictions and minimum bid-ask width will create small differences, but the way that i remember it, it was not close. perhaps i saw a strange few days or something.