Berms without Calibration
K. E. Feldman · 2025
Read Berms without Calibration by K. E. Feldman (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-15984.txt.
Introduction
K. E. Feldman, writing alone, prices Bermudan swaptions with a new semi-analytical model. The model needs no calibration that is specific to one product. A Bermudan swaption gives its holder the right to enter an interest-rate swap on preset dates. Pricing such options in the usual way means fitting a model to market data for each product in turn. Feldman's model works from swap rate distributions and the correlations between them instead. That lack of a calibration step is the model's chief claimed gain for daily pricing work.
Catalog record
| CREATOR | K. E. Feldman |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 46 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-15984
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Berms without Calibration K. E. Feldman Abstract We derive a new semi-analytical pricing model for Bermudan swaptions based on swap rates distributions and correlations between them. The model does not require product specific calibration. Introduction A Bermudan swaption on an interest rate swap gives the holder the right to enter the swap on some specified exercise dates. This product is considered to be the simplest exotic which requires the use of the term structure models for valuations and risk management. It is customary to use an LGM style model to price Bermudan swaptions. The volatility curve of the underlying Hull-White process is calibrated to the implied volatilities of the coterminal swaptions and the mean reversion curve is often recalibrated to the results of the Totem submissions. It is known that the mean reversion parameter is related to the forward volatility of the swaptions par rates (see for example [ 2 ] ) but calibration techniques with forward volatilities are not widely used, as it is difficult to express the forward volatility in terms of products liquidly traded in the interest rate market. Often midcurve swaptions (swaptions with a significant delay between the swaption expiry and the underlying start date) are considered as products on the forward volatility of the underlying (forward starting) swap rate, though fitting them in the term structure model calibration routine is not an easy task. In this paper we connect valuations of Bermudan swaptions and midcurve swaptions directly. As a connecting bridge we look at a less known modification of the midcurve swaption - the swaption with a relative strike where the strike is fixed sometime before the swaption exercise, but the swaption exercise is standard and within few business days of the underlying start. This swaption contract directly depends on the forward volatility between the strike fixing date and the swaption exercise date. To clarify the midcurve-Bermudan swaptions relationship we study first the Canary swaption [ 9 ] which is the simplest case of a Bermudan swaption - the swaption with only two exercise dates. Representing a Canary swaption as a basket of two option style products on coterminal swap rates we derive an analytic valuation based on minor assumptions on the dynamics of the swap rates between the first and the second expiry. We generalise the valuation to the case of arbitrary Bermudan swaption by developing Hagan LGM [ 7 ] style model. The main result of the present paper is similar in nature to the classical reduction of an European swaption valuation to the product of the swap annuity and a vanilla option price. We show how the numeraire change from the zero coupon bond to the annuity simplifies valuations of Bermudan swaptions by replacing unobservable mean reversion parameters with swap rates correlations. The later can be easily estimated from the historical data, or directly implied from midcurve swaptions or CMS spreads markets. The model developed in the present paper can be used to improve risk management of Bermudans as in addition to vanilla swaptions it allows inclusion of the simplest IR correlation products in to the hedging portfolio. Another advantage of the approach is that it does not require any preliminary product specific calibration. We also provide a toy model under the assumption of deterministic ratios of annuities and perfect correlations of the forward swap rates. In this setting we show how to value a Bermudan swaption as a zero floor on the maximum of correlated Gaussian variables using techniques presented in [ 10 ] . The paper has the following structure. In Section 1 we introduce the notations to be used throughout the paper, explain the pricing mechanism for midcurve swaptions and formulate the modelling assumption on annuity ratios. In Section 2 we introduce a modification to the swaption contract - swaptions with relative strikes, and provide its valuation formula. In Section 3 we present explicit valuation formulas for Canary swaptions under the assumption of deterministic ratios of annuities. In Section 4 we discuss an extension of the Canary swaption pricing approach to the case of a generic Bermudan swaption. We provide a general valuation mechanism and give two practical specialisations - one suitable for the moment matching valuation and another for Hagan lattice valuation. We discuss how stochastic annuity ratios can be handled in Section 5. We summarise our findings in Conclusion. 1 Midcurve Swaptions This section follows closely [ 5 ] and we refer the reader to it for all the main details, notations and proofs. We shall be working in the annuity probability measure {\cal A} where the relevant underlying swap annuity is a martingale. We assume that the corresponding sample space of continuous outcomes has a time dependent filtration \{{\cal F}_{s},s\in[t,T_{e}]\} with {\cal F}_{s_{1}}\subseteq{\cal F}_{s_{2}} , whenever s_{1}\leq s_{2} , t being the time corresponding to the valuation date and T_{e} corresponding to the end date of the underlying swap (see [ 1 ] for details of the use of martingale theory for interest rate modelling). A midcurve swaption is a swaption W(t,T_{ex},S(T_{s},T_{e},K)) on a forward starting swap S(T_{s},T_{e},K) with a significant delay between the swaption expiry T_{ex} and the underlying swap start date T_{s} ( T_{ex}<T_{s} , K is the swaption strike which coincides with the underlying swap fixed rate). The underlying (forward starting) swap S(T_{s},T_{e},K) can be decomposed in to the difference of the long swap S(T_{ex},T_{e},K) from the expiry date T_{ex} to the underlying swap end date T_{e} and the short swap S(T_{ex},T_{s},K) from the expiry date T_{ex} to the underlying swap start date T_{s} . This decomposition allows a midcurve swaption to be priced in the underlying swap annuity measure as an option on a correlated weighted basket of the long swap par rate R(t,T_{ex},T_{e}) and the short swap par rate R(t,T_{ex},T_{s}) with the weights given by the ratios of the respective swap annuities - the long swap annuity A(t,T_{ex},T_{e}) , the short swap annuity A(t,T_{ex},T_{s}) and the underlying swap annuity A(t,T_{s},T_{e}) (see [ 5 ] for more detailed explanation): \displaystyle W(t,T_{ex},S(T_{s},T_{e},K)) \displaystyle= \displaystyle A(t,T_{s},T_{e})\mathbb{E}^{\cal A}\Bigl[\Bigl(\omega(R(T_{ex},T_{s},T_{e})-K)\Bigr)^{+}\Big|{\cal F}_{t}\Bigr] (1) \displaystyle= \displaystyle A(t,T_{s},T_{e})\mathbb{E}^{\cal A}\Bigr[\Bigl(\omega\Bigl(\frac{A(T_{ex},T_{ex},T_{e})}{A(T_{ex},T_{s},T_{e})}R(T_{ex},T_{ex},T_{e})- \displaystyle\frac{A(T_{ex},T_{ex},T_{s})}{A(T_{ex},T_{s},T_{e})}R(T_{ex},T_{ex},T_{s})-K\Bigr)\Bigr)^{+}\Big|{\cal F}_{t}\Bigr]. Here and in what follows \omega=\pm 1 is used to denote whether the option is a payer or a receiver. We shall use the first argument in A(\cdot,T_{s},T_{e}) and R(\cdot,T_{s},T_{e}) to specify the observation (or the fixing) time of the stochastic variable. We reserve the low case t to specify the valuation date so that A(t,T_{s},T_{e}) and R(t,T_{s},T_{e}) are the expected (fair) values of the swap annuity and the swap par rate at the valuation date t . The upper case T will be used to indicate that the variables A(T,T_{s},T_{e}) , R(T,T_{s},T_{e}) are stochastic when seen from the valuation date t . As both A(T,T_{s},T_{e}) , R(T,T_{s},T_{e}) are martingales in the underlying (forward starting) swap annuity measure we can write: \displaystyle A(t,T_{s},T_{e})=\mathbb{E}^{\cal A}[A(T,T_{s},T_{e})], \displaystyle R(t,T_{s},T_{e})=\mathbb{E}^{\cal A}[R(T,T_{s},T_{e})]. (2) More generally we also have \displaystyle A(T_{fix},T_{s},T_{e})=\mathbb{E}^{\cal A}[A(T,T_{s},T_{e})|{\cal F}_{T_{fix}}], \displaystyle R(T_{fix},T_{s},T_{e})=\mathbb{E}^{\cal A}[R(T,T_{s},T_{e})|{\cal F}_{T_{fix}}]. (3) The expectation calculation in the midcurve payoff ( 1 ) can be performed by the copula integration of the payoff along the joint distribution of the long and the short swap rates in the forward starting annuity measure (see [ 5 ] for details). If the annuity