Real Time Data
Like any processes, pricing and revaluation tools are only as good as their inputs. While it is fairly easy to ensure that correct trade information has been recorded, up-to-date market data is another story.
For accurate pricing, we really need real-time intra-day data. For end of day revaluations, we need reliable data representing the day’s closing prices. For exchange-traded instruments, this is not an issue. By paying a fee we can access accurate tick-by-tick or end of day data from most exchanges. Good data for OTC instrument trading however is more difficult to source.
The main problem stems from whom this data is collected. Rather than being officially and centrally recorded, the data is contributed by market participants to the vendors. An exception to this would be when instruments are traded on the vendor’s system. In this case, the vendor has access to all of the traded prices, however the reliability of this information depends on the share of the market that is traded on the vendor system.
For most OTC instruments, however, the transaction occurs between the two parties in a confidential fashion and the prices are not made public. The lack of price transparency increases with the complexity and customisation of the trades.
Consider a feed such as AFMA data. Contribution-based rates are of course vulnerable to manipulation by the contributors themselves. Because AFMA data suffers from being a contributed and averaged rate feed, from a small collection of contributors, its reliability for data such as cap and swaption volatilities is open to question. This problem is compounded by the fact that AFMA data is in many cases the only feed available. Major vendors such as Bloomberg and Reuters all carry the same AFMA data feed, so for interest rate derivative data, there is little comparative advantage between vendors.
So are we stuck with just plain guessing? Not necessarily. In some cases, however, value-added analytics can help to mitigate the problems of reliable data. This works by implying certain numbers from other markets and other types of market information.
Using Value-Added Analytics to Counter Unreliable Data
Consider swaption volatilities. Generally speaking, the only source of swaption volatilities is ‘broker pages’. These pages run off contributed data from the sell-side only, not from any real trade information. Spreads tend to be very large and there tend to be inconsistencies between various sources.
However, we do have a source of highly liquid and very transparent trade data that we can use – exchange traded options (ETOs). Specifically, we are talking about SFE bill and bond future option contracts. Now obviously, neither a bill or bond futures contract will be a useful approximation for swaption volatility, so this is where our value-added analytics come into play.
Firstly, we need to calculate what the implied volatility of each option contract is. This information is not readily available, but can be implied by using a theoretical option ‘straddle’ strategy. With a straddle, we buy a call and sell a put option at the same strike, giving us a V-shaped payoff diagram that ‘straddles’ the strike price. When the underlying instruments are highly volatile, this V will be wider than when there is low volatility. Remember, to create a straddle we need to buy and sell options, and the bid/offer prices reflect the market volatility. So by creating the straddle structure, we can effectively buy or sell the volatility, not the price, of the underlying asset.
The next step is to calculate the exact volatility implied by the price we pay for the straddle structure. Remember that this price will be the call option premium plus the put option premium, usually represented as basis points of the underlying price. Using the Black-Scholes pricing formula, if we have inputs of time to expiry, spot price, strike price and volatility, we can calculate the option premium. Similarly, if we have the option premium, we can back-out the volatility implied by this premium, by means of an interactive process.
So now we can calculate the implied volatility of each listed exchange traded option. Assume that we do this for the next 8 bill future option contracts, and the next 4 options on each of the three-year and ten-year bond future option contracts. Keep in mind that the next 4 listed three-year bond future option contracts will only cover the next two futures roll dates, as the SFE lists ‘serial options’ on each future with a range of expiry dates.
Now that we have our implied ETO contract volatilities, we can use them to build a volatility surface. Consider the surface as a matrix with the start date of the underlying contract on the X-axis and the option expiries along the Y-axis. With bill future option contracts, the expiry of the option falls 7 days before the underlying futures roll date and the underlying contracts roll quarterly. Because of this, our bill futures create a ‘leading diagonal’ on the surface, because each expiry is very similar to each start date. We do not really care about anything below this leading diagonal, as this would imply an option in which the underlying rolls before the option expires.
While a leading diagonal of quarterly volatilities can be useful for pricing quarterly caps, it is not as useful for swaptions. We need to add some depth, and for this we use our bond futures contracts. By using various techniques and approximations, we can interactively fill out the top section of the leading diagonal, giving us half a matrix full of implied volatilities. We can do this due to the long tenor of the underlying three and ten year bonds (actually the underlying of the underlying futures). As volatilities of short-rate based instruments fluctuate, there should be less impact on the longer-term rates, and hence we can make certain assumptions for options with start dates well after the expiry.
Now we come to reading the volatilities from the surface. All we have at the moment is a quarterly grid of implied volatilities; we do not necessarily reflect the attributes of a particular trade. A three-month year option on a one-year underlying swap will have a different rate exposure profile to that of a three year option on a five year swap. For this reason, we need to go through a process of aggregating and interpolating the raw surface volatilities over the tenor period of the swaption. This process will take as inputs a particular section of the surface and produce a single, implied volatility figure. This is the figure that we can then use as our swaption volatility.
It is true that our implied volatility is in no way a market ‘observed’ volatility, and we should not necessarily expect to be able to trade at a premium implied by this number. For indicative pricing and revaluation purposes, however, it is a far more reliable figure to use than those from the broker pages. The whole process assumes a certain efficiency of the market meaning that there are no arbitrage opportunities between the swaption and exchange traded option markets.
The above example of swaption volatilities is a complex one, but simpler ‘tricks’ can be used in other areas to improve the consistency and reliability of revaluations. The old saying is ‘garbage in, garbage out’.
