1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385
|
/* -*- mode: c++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
/*
Copyright (C) 2014 Peter Caspers
This file is part of QuantLib, a free-software/open-source library
for financial quantitative analysts and developers - http://quantlib.org/
QuantLib is free software: you can redistribute it and/or modify it
under the terms of the QuantLib license. You should have received a
copy of the license along with this program; if not, please email
<quantlib-dev@lists.sf.net>. The license is also available online at
<https://www.quantlib.org/license.shtml>.
This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the license for more details.
*/
#include <ql/experimental/volatility/zabr.hpp>
#include <ql/termstructures/volatility/sabr.hpp>
#include <ql/errors.hpp>
#include <ql/math/comparison.hpp>
#include <ql/math/distributions/normaldistribution.hpp>
#include <ql/math/ode/adaptiverungekutta.hpp>
#include <ql/methods/finitedifferences/operators/fdmlinearoplayout.hpp>
#include <ql/methods/finitedifferences/meshers/fdm1dmesher.hpp>
#include <ql/methods/finitedifferences/meshers/uniform1dmesher.hpp>
#include <ql/methods/finitedifferences/meshers/concentrating1dmesher.hpp>
#include <ql/experimental/finitedifferences/glued1dmesher.hpp>
#include <ql/methods/finitedifferences/meshers/fdmmeshercomposite.hpp>
#include <ql/methods/finitedifferences/operatortraits.hpp>
#include <ql/methods/finitedifferences/utilities/fdmdirichletboundary.hpp>
#include <ql/methods/finitedifferences/solvers/fdmbackwardsolver.hpp>
#include <ql/experimental/finitedifferences/fdmdupire1dop.hpp>
#include <ql/experimental/finitedifferences/fdmzabrop.hpp>
using std::pow;
namespace QuantLib {
ZabrModel::ZabrModel(const Real expiryTime, const Real forward,
const Real alpha, const Real beta, const Real nu,
const Real rho, const Real gamma)
: expiryTime_(expiryTime), forward_(forward), alpha_(alpha), beta_(beta),
nu_(nu * std::pow(alpha_, 1.0 - gamma)), rho_(rho), gamma_(gamma) {
validateSabrParameters(alpha, beta, nu, rho);
QL_REQUIRE(gamma >= 0.0 /*&& gamma<=1.0*/,
"gamma must be non negative: " << gamma << " not allowed");
QL_REQUIRE(forward >= 0.0,
"forward must be non negative: " << forward << " not allowed");
QL_REQUIRE(expiryTime > 0.0, "expiry time must be positive: "
<< expiryTime << " not allowed");
}
Real ZabrModel::lognormalVolatilityHelper(const Real strike,
const Real x) const {
if (close(strike, forward_))
return std::pow(forward_, beta_ - 1.0) * alpha_;
else
return std::log(forward_ / strike) / x;
}
Real ZabrModel::lognormalVolatility(const Real strike) const {
return lognormalVolatility(std::vector<Real>(1, strike))[0];
}
std::vector<Real> ZabrModel::lognormalVolatility(const std::vector<Real> &strikes) const {
std::vector<Real> x_ = x(strikes);
std::vector<Real> result(strikes.size());
std::transform(strikes.begin(), strikes.end(), x_.begin(), result.begin(),
[&](Real _k, Real _x) { return lognormalVolatilityHelper(_k, _x); });
return result;
}
Real ZabrModel::normalVolatilityHelper(const Real strike, const Real x) const {
if (close(strike, forward_))
return std::pow(forward_, beta_) * alpha_;
else
return (forward_ - strike) / x;
}
Real ZabrModel::normalVolatility(const Real strike) const {
return normalVolatility(std::vector<Real>(1, strike))[0];
}
std::vector<Real> ZabrModel::normalVolatility(const std::vector<Real> &strikes) const {
std::vector<Real> x_ = x(strikes);
std::vector<Real> result(strikes.size());
std::transform(strikes.begin(), strikes.end(), x_.begin(), result.begin(),
[&](Real _k, Real _x) { return normalVolatilityHelper(_k, _x); });
return result;
}
Real ZabrModel::localVolatilityHelper(const Real f, const Real x) const {
return alpha_ * std::pow(std::fabs(f), beta_) /
F(y(f), std::pow(alpha_, gamma_ - 1.0) *
x); // TODO optimize this, y is comoputed together
// with x already
}
Real ZabrModel::localVolatility(const Real f) const {
return localVolatility(std::vector<Real>(1, f))[0];
}
std::vector<Real> ZabrModel::localVolatility(const std::vector<Real> &f) const {
std::vector<Real> x_ = x(f);
std::vector<Real> result(f.size());
std::transform(f.begin(), f.end(), x_.begin(), result.begin(),
[&](Real _f, Real _x) { return localVolatilityHelper(_f, _x); });
return result;
}
Real ZabrModel::fdPrice(const Real strike) const {
return fdPrice(std::vector<Real>(1, strike))[0];
}
std::vector<Real> ZabrModel::fdPrice(const std::vector<Real> &strikes) const {
// TODO check strikes to be increasing
// TODO put these parameters somewhere
const Real start =
std::min(0.00001, strikes.front() * 0.5); // lowest strike for grid
const Real end =
std::max(0.10, strikes.back() * 1.5); // highest strike for grid
const Size size = 500; // grid points
const Real density = 0.1; // density for concentrating mesher
const Size steps =
(Size)std::ceil(expiryTime_ * 24); // number of steps in dimension t
const Size dampingSteps = 5; // thereof damping steps
#if defined(__GNUC__) && (__GNUC__ >= 12)
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Warray-bounds"
#endif
// Layout
std::vector<Size> dim(1, size);
const ext::shared_ptr<FdmLinearOpLayout> layout(
new FdmLinearOpLayout(dim));
#if defined(__GNUC__) && (__GNUC__ >= 12)
#pragma GCC diagnostic pop
#endif
// Mesher
const ext::shared_ptr<Fdm1dMesher> m1(new Concentrating1dMesher(
start, end, size, std::pair<Real, Real>(forward_, density), true));
// const ext::shared_ptr<Fdm1dMesher> m1(new
// Uniform1dMesher(start,end,size));
// const ext::shared_ptr<Fdm1dMesher> m1a(new
// Uniform1dMesher(start,0.03,101));
// const ext::shared_ptr<Fdm1dMesher> m1b(new
// Uniform1dMesher(0.03,end,100));
// const ext::shared_ptr<Fdm1dMesher> m1(new Glued1dMesher(*m1a,*m1b));
const std::vector<ext::shared_ptr<Fdm1dMesher> > meshers(1, m1);
const ext::shared_ptr<FdmMesher> mesher(
new FdmMesherComposite(layout, meshers));
// Boundary conditions
FdmBoundaryConditionSet boundaries;
// initial values
Array rhs(mesher->layout()->size());
for (const auto& iter : *layout) {
Real k = mesher->location(iter, 0);
rhs[iter.index()] = std::max(forward_ - k, 0.0);
}
// local vols (TODO how can we avoid these Array / vector copying?)
Array k = mesher->locations(0);
std::vector<Real> kv(k.size());
std::copy(k.begin(), k.end(), kv.begin());
std::vector<Real> locVolv = localVolatility(kv);
Array locVol(locVolv.size());
std::copy(locVolv.begin(), locVolv.end(), locVol.begin());
// solver
ext::shared_ptr<FdmDupire1dOp> map(new FdmDupire1dOp(mesher, locVol));
FdmBackwardSolver solver(map, boundaries,
ext::shared_ptr<FdmStepConditionComposite>(),
FdmSchemeDesc::Douglas());
solver.rollback(rhs, expiryTime_, 0.0, steps, dampingSteps);
// interpolate solution
ext::shared_ptr<Interpolation> solution(new CubicInterpolation(
k.begin(), k.end(), rhs.begin(), CubicInterpolation::Spline, true,
CubicInterpolation::SecondDerivative, 0.0,
CubicInterpolation::SecondDerivative, 0.0));
// ext::shared_ptr<Interpolation> solution(new
// LinearInterpolation(k.begin(),k.end(),rhs.begin()));
solution->disableExtrapolation();
std::vector<Real> result(strikes.size());
std::transform(strikes.begin(), strikes.end(), result.begin(), *solution);
return result;
}
Real ZabrModel::fullFdPrice(const Real strike) const {
// TODO what are good values here, still experimenting with them
Real eps = 0.01;
Real scaleFactor = 1.5;
Real normInvEps = InverseCumulativeNormal()(1.0 - eps);
Real alphaI = alpha_ * std::pow(forward_, beta_ - 1.0);
// nu is already standardized within this class ...
Real v0 = alpha_ * std::exp(-scaleFactor * normInvEps *
std::sqrt(expiryTime_) * nu_);
Real v1 = alpha_ *
std::exp(scaleFactor * normInvEps * std::sqrt(expiryTime_) * nu_);
Real f0 = forward_ * std::exp(-scaleFactor * normInvEps *
std::sqrt(expiryTime_) * alphaI);
Real f1 = forward_ * std::exp(scaleFactor * normInvEps *
std::sqrt(expiryTime_) * alphaI);
v1 = std::min(v1, 2.0);
f0 = std::min(strike / 2.0, f0);
f1 = std::max(strike * 1.5, std::min(f1, std::max(2.0, strike * 1.5)));
const Size sizef = 100;
const Size sizev = 100;
const Size steps = Size(24 * expiryTime_ + 1);
const Size dampingSteps = 5;
const Real densityf = 0.1;
const Real densityv = 0.1;
QL_REQUIRE(strike >= f0 && strike <= f1,
"strike (" << strike << ") must be inside pde grid [" << f0
<< ";" << f1 << "]");
// Layout
std::vector<Size> dim;
dim.push_back(sizef);
dim.push_back(sizev);
const ext::shared_ptr<FdmLinearOpLayout> layout(
new FdmLinearOpLayout(dim));
// Mesher
// two concentrating mesher around f and k to get the mesher for f
const Real x0 = std::min(forward_, strike);
const Real x1 = std::max(forward_, strike);
const Size sizefa = std::max<Size>(
4, (Size)std::ceil(((x0 + x1) / 2.0 - f0) / (f1 - f0) * (Real)sizef));
const Size sizefb = sizef - sizefa + 1; // common point, so we can spend
// one more here
const ext::shared_ptr<Fdm1dMesher> mfa(
new Concentrating1dMesher(f0, (x0 + x1) / 2.0, sizefa,
std::pair<Real, Real>(x0, densityf), true));
const ext::shared_ptr<Fdm1dMesher> mfb(
new Concentrating1dMesher((x0 + x1) / 2.0, f1, sizefb,
std::pair<Real, Real>(x1, densityf), true));
const ext::shared_ptr<Fdm1dMesher> mf(new Glued1dMesher(*mfa, *mfb));
// concentraing mesher around f to get the forward mesher
// const ext::shared_ptr<Fdm1dMesher> mf(new Concentrating1dMesher(
// f0, f1, sizef, std::pair<Real, Real>(forward_, densityf), true));
// Volatility mesher
const ext::shared_ptr<Fdm1dMesher> mv(new Concentrating1dMesher(
v0, v1, sizev, std::pair<Real, Real>(alpha_, densityv), true));
// uniform meshers
// const ext::shared_ptr<Fdm1dMesher> mf(new
// Uniform1dMesher(f0,f1,sizef));
// const ext::shared_ptr<Fdm1dMesher> mv(new
// Uniform1dMesher(v0,v1,sizev));
std::vector<ext::shared_ptr<Fdm1dMesher> > meshers;
meshers.push_back(mf);
meshers.push_back(mv);
const ext::shared_ptr<FdmMesher> mesher(
new FdmMesherComposite(layout, meshers));
// initial values
Array rhs(mesher->layout()->size());
std::vector<Real> f_;
std::vector<Real> v_;
for (const auto& iter : *layout) {
Real f = mesher->location(iter, 0);
// Real v = mesher->location(iter, 0);
rhs[iter.index()] = std::max(f - strike, 0.0);
if (iter.coordinates()[1] == 0U)
f_.push_back(mesher->location(iter, 0));
if (iter.coordinates()[0] == 0U)
v_.push_back(mesher->location(iter, 1));
}
// Boundary conditions
FdmBoundaryConditionSet boundaries;
ext::shared_ptr<FdmZabrOp> map(
new FdmZabrOp(mesher, beta_, nu_, rho_, gamma_));
FdmBackwardSolver solver(map, boundaries,
ext::shared_ptr<FdmStepConditionComposite>(),
FdmSchemeDesc::/*CraigSneyd()*/ Hundsdorfer());
solver.rollback(rhs, expiryTime_, 0.0, steps, dampingSteps);
// interpolate solution (this is not necessary when using concentrating
// meshers with required point)
Matrix result(f_.size(), v_.size());
for (Size j = 0; j < v_.size(); ++j)
std::copy(rhs.begin() + j * f_.size(),
rhs.begin() + (j + 1) * f_.size(), result.row_begin(j));
ext::shared_ptr<BicubicSpline> interpolation =
ext::make_shared<BicubicSpline>(
f_.begin(), f_.end(), v_.begin(), v_.end(), result);
interpolation->disableExtrapolation();
return (*interpolation)(forward_, alpha_);
}
Real ZabrModel::x(const Real strike) const {
return x(std::vector<Real>(1, strike))[0];
}
std::vector<Real> ZabrModel::x(const std::vector<Real> &strikes) const {
QL_REQUIRE(strikes[0] > 0.0 || beta_ < 1.0,
"strikes must be positive (" << strikes[0] << ") if beta = 1");
for (auto i = strikes.begin() + 1; i != strikes.end(); ++i)
QL_REQUIRE(*i > *(i - 1), "strikes must be strictly ascending ("
<< *(i - 1) << "," << *i << ")");
AdaptiveRungeKutta<Real> rk(1.0E-8, 1.0E-5,
0.0); // TODO move the parameters here as
// parameters with default values to
// the constructor
std::vector<Real> y(strikes.size()), result(strikes.size());
std::transform(strikes.rbegin(), strikes.rend(), y.begin(),
[&](Real _k) { return this->y(_k); });
if (close(gamma_, 1.0)) {
for (Size m = 0; m < y.size(); m++) {
Real J = std::sqrt(1.0 + nu_ * nu_ * y[m] * y[m] -
2.0 * rho_ * nu_ * y[m]);
result[y.size() - 1 - m] =
std::log((J + nu_ * y[m] - rho_) / (1.0 - rho_)) / nu_;
}
} else {
Size ynz = std::upper_bound(y.begin(), y.end(), 0.0) - y.begin();
if (ynz > 0)
if (close(y[ynz - 1], 0.0))
ynz--;
if (ynz == y.size())
ynz--;
for (int dir = 1; dir >= -1; dir -= 2) {
Real y0 = 0.0, u0 = 0.0;
for (int m = ynz + (dir == -1 ? -1 : 0);
dir == -1 ? m >= 0 : m < (int)y.size(); m += dir) {
Real u = rk([&](Real _y, Real _u){ return F(_y, _u); },
u0, y0, y[m]);
result[y.size() - 1 - m] = u * pow(alpha_, 1.0 - gamma_);
u0 = u;
y0 = y[m];
}
}
}
return result;
}
Real ZabrModel::y(const Real strike) const {
if (close(beta_, 1.0)) {
return std::log(forward_ / strike) * std::pow(alpha_, gamma_ - 2.0);
} else {
return (strike < 0.0
? Real(std::pow(forward_, 1.0 - beta_) +
std::pow(-strike, 1.0 - beta_))
: Real(std::pow(forward_, 1.0 - beta_) -
std::pow(strike, 1.0 - beta_))) *
std::pow(alpha_, gamma_ - 2.0) / (1.0 - beta_);
}
}
Real ZabrModel::F(const Real y, const Real u) const {
Real A = 1.0 + (gamma_ - 2.0) * (gamma_ - 2.0) * nu_ * nu_ * y * y +
2.0 * rho_ * (gamma_ - 2.0) * nu_ * y;
Real B = 2.0 * rho_ * (1.0 - gamma_) * nu_ +
2.0 * (1.0 - gamma_) * (gamma_ - 2.0) * nu_ * nu_ * y;
Real C = (1.0 - gamma_) * (1.0 - gamma_) * nu_ * nu_;
return (-B * u + std::sqrt(B * B * u * u - 4.0 * A * (C * u * u - 1.0))) /
(2.0 * A);
}
}
|