142 CHAPTER ◆ 1 4 Gather Historical Data
space, and then converted to dollars, the primary question is the following: what is
the proper way to calculate implied volatility? Should it be done per exchange, per
option and then a blend, or should you blend prices? Because we believe in put/call
parity, call and put volatilities should be equal. Many market participants and data
vendors do not really believe in put/call parity and calculate separate volatilities for
puts and calls. These vendors ’ data shows the put volatility below and call volatility
above the at-the-money volatility.When we convert from put and call option prices to implied volatilities, we try to calcu-
late a single value for implied volatility. This can only be done by incorporating multi-
ple interest rates depending upon the delta of the position, hard to borrow rules, earnings
information, boundary conditions, splits, and other data to properly adjust the call and put
model inputs. As a final step, we smooth the final difference, which is normally less than
half a volatility point, by delta-weighting the values.
The firms that do not follow put/call parity prefer instead to allow call and put vola-
tilities to be different, and simply average the two volatilities to price OTC options.Incorrect pricing models that incorporate incorrect implied volatilities have caused several blow-ups.
The firms used incorrect volatilities and proprietary models to mark-to-model their positions. The symp-
toms are always the same: on paper the system is making a lot of money, but when the positions are
marked-to-market by third-party bids and offers, the system is losing money. The response is always the
same: “ Our theoretical edge is growing so pay us our bonus. ”At the end of the day either your implied volatility calculation is right and your mark-
to-model and mark-to-market values are very close, or you are wrong and go bankrupt.
Financial engineers sometimes forget that you need to sell to make money. The finan-
cial engineers use the same incorrect model and volatility surface in both the front office
and back office at large firms since the desk quants mainly come from the back office
at large firms so they convince the back office quants in their logic. We believe the gold
standard in implied volatility is where put/call parity holds for the entire surface.● Fixed income data. Several firms sell fixed income data, although sometimes it is
not readily apparent whether the prices are observed transactions or theoretical val-
ues. The major data vendors provide both. On days where there is no activity, they
will calculate a theoretical price for the fixed income securities and derivatives. This
theoretical price appears to reduce the volatility of the price as it could be the same
for several days in a row. Of course, no dealer is obligated to buy or sell at a theo-
retical value.
● Futures data. In general, futures data is very clean. Standardized contracts with
well-defined pricing, and generally little cross-listing are the reasons volume is so
large in electronically traded contracts. However, on days where contracts do not
trade, that is, when the contract ’ s lock limit is up or down, exchanges publish no
price data. Regardless, OTC contracts for the underlying may still trade.14.2.2. Valuation Data
Valuation data is different from price data. For bonds, swaps, and all OTC derivatives, no
historical trade price data exists, or if it does, it is a highly guarded secret. For all these,
valuation data is all there is, that is, the “ price ” exists only in theory, and, furthermore,