TVIX Explosion

I just came across an excellent paper titled TVIX Explosion Drives Vol-of-Vol Higher [link removed, use google] from DB about TVIX. There are significant structural changes in the market with increasing TVIX AUM, and its impact on all other volatility instruments.

The main takeaway from the article is that leveraged products are short-gamma - requiring ETN provider to buy after positive move in the underlying, and sell after negative move - whether it is bull or bear. Natural hedge for such product, a long gamma ETN, would require leverage between 0 and 1, and does not exist, or likely to ever appear in the market.

Nikkei Volatility Futures - first day of trading

On Monday Osaka Exchange started trading in futures tied to Nikkei Volatility Index. According to bloomberg/businessweek 49 March-expiration contracts traded, but exchange lists monthly expirations until November 2012. Futures are trading under ticker NKVI, and the front two months are at a healthy 2 point premium to the index.

Contract specification for NKVI futures are available here.

VHSI, VXEW, OV First Week Of Trading

Last Monday HKFE started trading in VHSI Futures linked on Hong-Kong volatility index. Three months are listed - Feb, Mar, Apr (the contracts expire toward the end of calendar month), and traded total of 47 contracts.

Things did not go as well for other volatility products débuted at CBOE last week: neither CBOE Brazil ETF Volatility Index Security Futures (VXEW) nor CBOE Oil ETF Volatility Index Security Futures (OV) traded at all.

Time Decay

Many traders are under assumption that they can profit by selling options before weekend (especially 3-day weekend like last week) because market-makers are not properly adjusting options. It is true that in the 80s and 90s options pricing was done using computer-generated sheets updated overnight, and market-makers were quite liberal in terms of updating the "date". Still it probably never provided an opportunity for other traders since bid-ask spreads were significant to prevent any meaningful profit.

In the late '90 and early '00 computers started playing dominant role in options pricing. I have heard the following "rule" from a market-maker: on Friday set the date for Sunday, to prevent Friday to Monday price jumps. 

Currently, options market-making is highly automated. Time to expiration is being adjusted in real-time, millisecond by  millisecond. Not only that, different days of the week have different decay rates: trading days -high, weekends - low. Different times of day decay at different rates as well - for example the first half-hour after the open have higher decay rate than the mid-day. 

Now I will address a theoretical point on options pricing: let's consider 3 calendars and its implication on implied volatility calculation. Link to google docs spreadsheet. The first trader is using regular calendar - 365 days per year, the second trader uses business days calendar with 252 days per year. Days to expiration, starting from 2 weeks before x-date are columns in the spreadsheet. Market, however prices options somewhat in between, counting weekends as half-days of decay, with 280 "days" in the year (1/4 * 365 + 3/4 * 252 ~ 280)

For simplicity consider ATMF straddle that we price as 100 * sqrt(T) , using market 280 day convention (row 6 in the spreadsheet) and implied bias from calculating using 365 (row 8) or 252 (row 9). This biases our volatility estimate from 5% below to 14% above, which is significant. 

Rolling Volatility Futures Indexes

Fellow blogger Vance from Six Figure Investing created a complete suite of volatility indexes based on VIX futures, from their inception in 2004 until present. The indexes he created are total return indexes and therefore suitable for backtesting.

Actually they are even better than "official" indexes like SPVXSTR, because they have longer history, and correct data errors in the original index calculation! The last point is extremely important if you are trying to backtest VIX futures strategy - on some days settlement data from CBOE and Bloomberg did not match, and in other days rollover days were not accounted for in a proper manner. Link to the data, link to very detailed description of the calculations and sources.