Swap Curve construction using Global Optimization

In my previous post, I discussed option Implied Volatility and Binomial Model. In this post, we switch gears and discuss Swap Curve construction. Building a swap curve is so fundamental to interest rate derivatives pricing that it is probably one of most closely guarded proprietary information on the trading desk.

Pricing an interest rate swap involves bootstrapping a blended curve of different instruments based on their maturities and market liquidity. Usually cash deposits, Eurodollar futures are used at the short end and market swap rates are used at the long end. At the long end, however, we have only a subset of market swap rates available and bootstrapping requires all the missing rates to be interpolated from known rates. This makes interpolation methodology a critical part of curve building. Also, since forward rates are a gradient of the discount rates, any misalignment in the latter is magnified in the former.

There is an alternative approach to bootstrapping called global optimization approach, where the short end of the curve is bootstrapped as usual but at the longer end we “guess” the forward rates, compute the par rates of the market swaps and minimize the error between the actual par rates and the computed par rates. We also add a smoothness constraint to the minimization procedure so that overall gradient of the curve is minimized. This approach is illustrated in the excellent book Swaps and Other derivatives 2nd Ed by Richard Flavell

I will use QuantLib to generate swap schedules and to deal with business day conventions. QuantLib can of course generate a fully built swap curve but I will use Scipy’s optimize package for curve building. My objective was to match Spreadsheet 3.9 “Building a blended curve” from the above book. Unfortunately, the spreadsheet does not actually show the equations for Excel’s Solver used for optimization but shows the final result, so that leaves considerable ambiguity in understanding which I hope I will be able to clear.

Note on QuantLib and Python

There are numerous resources online on how to build QuantLib from sources and then build the Python extensions, I would like to point you to the precompiled package for QuantLib-Python maintained by Christoph Gohlke. If you are on windows, you can just install the whl package and get started.

First some common formulae we will be using:

$$Discount Factor : DF_t = \frac{1}{(1 + r_t * d_t)}$$ where $d_t$ is year fraction and $r_t$ is annual rate

$$Forward Rate : F_{t/T} = \frac{[(DF_t/DF_T)- 1]}{(T-t)}$$ where $DF_t$ is disocunt factor to t and $DF_T$ is disocunt factor to T (both from today)

$$Par  Rate  of  Swap: \frac{(1-D_T)}{\sum_{n=1}^n(\Delta_n * D_n)}$$ where $D_T$ is maturity date discount factor, $\Delta_n$ are the time fractions between 2 reset dates and $D_n$ are the various reset date discount factors.

gtol termination condition is satisfied.
Function evaluations: 20, initial cost: 1.7812e-03, final cost 1.5585e-09, first-order optimality 1.04e-08.
### Scipy optimization took:2.5262434 seconds

Option Volatility and Binomial Model

In my previous post Options and Volatility Smile , we used Black-Scholes formula to derive Implied Volatility from given option strike and tenor. Most of the options traded on exchanges are American (with a few index options being European) and can be exercised at any time prior to expiration. Whether it is optimal to exercise an option early depends on whether the stock pays dividend or the level of interest rates and is a very complex subject. What I want to focus on is using Binomial model to price an American option. I summarize below Binomial model theory from the excellent book Derivative Markets 3rd Ed. by Robert McDonald

There are many flavors of the Binomial model but they all have following steps in common:

  • Simulate future prices of underlying stock at various points in time until expiry
  • Calculate the payoff of the option at expiry
  • Discount the payoff back to today to get the option price

1. Simulate future prices of the underlying

Assume the price of a stock is $S_0$ and over a small time period $\Delta t$, the price could either be $S_0u$ or $S_0d$ where u is a factor by which price rises and d is a factor by which price drops. The stock is assumed to follow a random walk, also assume that $p$ is the probability of the stock price rising and $(1-p)$ is the probability of it falling. There are many ways to approach the values of $u$,$d$ and $p$ and the various Binomial models differ in the ways these three parameters are calculated.

In Cox-Ross-Rubinstein (CRR) model, $u = \frac{1}{d}$ is assumed. Since we have 3 unknowns, 2 more equations are needed and those come from risk neutral pricing assumption. Over a small $\Delta t$ the expected return of the stock is

$$pu + (1-p)d = e^{r \Delta t}$$

and the expected variance of the returns is

$$pu^2 + (1-p)d^2 – (e^{r \Delta t})^2 = \sigma ^2 \Delta t$$

Solving for $u$, $d$ and $p$, we get

$$p = \frac{e^{r\Delta t} – d}{u-d}$$

$$u = e^{\sigma \sqrt{\Delta t}}$$

$$d = e^{-\sigma\sqrt{\Delta t}}$$

The CRR model generates a following tree as we simulate multi step stock price movements, this a recombining tree centered around $S_0$.

 


2. Calculating payoffs at expiry

In this step, we calculate the payoff at each node that corresponds to expiry.

For a put option, $payoff = max(K – S_N, 0)$

For a call option, $payoff = max(S_N – K, 0)$

where $N$ is node at expiry with a stock price $S_N$ and $K$ is the strike.

3. Discounting the payoffs

In this step, we discount the payoffs at expiry back to today using backward induction where we start at expiry node and step backwards through time calculating option value at each node of the tree.

For American put, $V_n = max(K – S_n, e^{-r \Delta t} (p V_u + (1-p) V_d))$

For American call, $V_n = max(S_n – K, e^{-r \Delta t} (p V_u + (1-p) V_d))$

$V_n$ is the option value at node n

$S_n$ is the stock price at node n

$r$ is risk free interest rate

$\Delta t$ is time step

$V_u$ is the option value from the upper node at n+1

$V_d$ is the option value from the lower node at n+1

All the variants of Binomial model, including CRR, converge to Black-Scholes in the limit $\Delta t \to 0$ but the rate of convergence is different. The variant of Binomial model that I would like to use is called Leisen-Reimer Model which converges much faster. Please see the original paper for formulas and a C++ implementation at Volopta.com  which I have ported to Python in the next section.

The python code is going to look very similar to Options and Volatility Smile post except that we will swap out Black-Scholes framework with Leisen-Reimer model. We will also use the same option chain data AAPL_BBG_vols.csv

A note on Python code

I usually do not write code like below, I am purposely avoiding using any classes so that the focus remains on the objective which is to understand the technique.