0% found this document useful (0 votes)
19 views49 pages

Numerical Methods and Optimization in Finance 2Nd Edition Manfred Gilli - Ebook PDF

The document provides information about various eBooks available for download on ebookluna.com, including 'Numerical Methods and Optimization in Finance' by Manfred Gilli and several other titles related to numerical methods and optimization in engineering and environmental data analysis. It includes links to download these eBooks in different formats such as PDF, ePub, and MOBI. The document also outlines the contents of the 'Numerical Methods and Optimization in Finance' book, detailing its structure and key topics covered.

Uploaded by

manfiosuup14
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
19 views49 pages

Numerical Methods and Optimization in Finance 2Nd Edition Manfred Gilli - Ebook PDF

The document provides information about various eBooks available for download on ebookluna.com, including 'Numerical Methods and Optimization in Finance' by Manfred Gilli and several other titles related to numerical methods and optimization in engineering and environmental data analysis. It includes links to download these eBooks in different formats such as PDF, ePub, and MOBI. The document also outlines the contents of the 'Numerical Methods and Optimization in Finance' book, detailing its structure and key topics covered.

Uploaded by

manfiosuup14
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 49

Get the full ebook with Bonus Features for a Better Reading Experience on ebookluna.

com

Numerical Methods and Optimization in Finance 2nd


edition Manfred Gilli - eBook PDF

https://ebookluna.com/download/numerical-methods-and-
optimization-in-finance-2nd-ed-ebook-pdf/

OR CLICK HERE

DOWLOAD NOW

Download more ebook instantly today at https://ebookluna.com


Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...

Numerical Methods in Engineering with MATLAB 3rd Edition

https://ebookluna.com/product/numerical-methods-in-engineering-with-
matlab-3rd-edition/

ebookluna.com

(eBook PDF) Numerical Methods in Engineering with MATLAB


3rd Edition

https://ebookluna.com/product/ebook-pdf-numerical-methods-in-
engineering-with-matlab-3rd-edition/

ebookluna.com

Numerical Methods in Environmental Data Analysis 1st


Edition- eBook PDF

https://ebookluna.com/download/numerical-methods-in-environmental-
data-analysis-ebook-pdf/

ebookluna.com

(eBook PDF) Optimization in Operations Research 2nd


Edition

https://ebookluna.com/product/ebook-pdf-optimization-in-operations-
research-2nd-edition/

ebookluna.com
Complex Variables and Numerical Methods Gujarat
Technological University 2017 2nd Edition Ravish R Singh -
eBook PDF
https://ebookluna.com/download/complex-variables-and-numerical-
methods-gujarat-technological-university-2017-ebook-pdf/

ebookluna.com

(eBook PDF) Numerical Methods for Engineers 7th Edition

https://ebookluna.com/product/ebook-pdf-numerical-methods-for-
engineers-7th-edition/

ebookluna.com

(eBook PDF) Numerical Methods for Engineers 8th Edition

https://ebookluna.com/product/ebook-pdf-numerical-methods-for-
engineers-8th-edition-2/

ebookluna.com

(eBook PDF) Numerical Methods for Engineers 8th Edition

https://ebookluna.com/product/ebook-pdf-numerical-methods-for-
engineers-8th-edition/

ebookluna.com

(eBook PDF) Numerical Methods for Engineers 8th Edition

https://ebookluna.com/product/ebook-pdf-numerical-methods-for-
engineers-8th-edition-3/

ebookluna.com
Numerical Methods
and Optimization
in Finance
Second Edition

Manfred Gilli
Geneva School of Economics and Management
University of Geneva
Geneva, Switzerland
Swiss Finance Institute
Zürich, Switzerland

Dietmar Maringer
Faculty of Business and Economics
University of Basel
Basel, Switzerland
Geneva School of Economics and Management
University of Geneva
Geneva, Switzerland

Enrico Schumann
Faculty of Business and Economics
University of Basel
Basel, Switzerland
© 2019 Elsevier Inc.

ISBN: 978-0-12-815065-8
Contents

List of figures xi Example of an


List of tables xvii ill-conditioned problem 25
List of algorithms xix 2.5 Condition number of a matrix 26
Acknowledgments xxi Comments and examples 27
Foreword to the second edition xxiii 2.6 A primer on algorithmic and
computational complexity 28
Part I Criteria for comparison 28
Order of complexity and
Fundamentals classification 29
1. Introduction Appendix 2.A Operation count for
basic linear algebra operations 30
1.1 About this book 3
The growth of computing
power 3 3. Linear equations and Least
Computational finance 4 Squares problems
1.2 Principles 6
3.1 Direct methods 32
1.3 On software 7
3.1.1 Triangular systems 32
1.4 On approximations and
accuracy 10 3.1.2 LU factorization 33
1.5 Summary: the theme of the 3.1.3 Cholesky factorization 35
book 15 3.1.4 QR decomposition 37
3.1.5 Singular value
decomposition 37
2. Numerical analysis in a 3.2 Iterative methods 38
nutshell 3.2.1 Jacobi, Gauss–Seidel, and
2.1 Computer arithmetic 17 SOR 39
Representation of real Successive overrelaxation 40
numbers 17 3.2.2 Convergence of iterative
Machine precision 19 methods 41
Example of limitations of 3.2.3 General structure of
floating point arithmetic 19 algorithms for iterative
2.2 Measuring errors 20 methods 42
2.3 Approximating derivatives with 3.2.4 Block iterative methods 44
finite differences 21 3.3 Sparse linear systems 45
Approximating first-order 3.3.1 Tridiagonal systems 45
derivatives 21 3.3.2 Irregular sparse matrices 47
Approximating 3.3.3 Structural properties of
second-order derivatives 22 sparse matrices 48
Partial derivatives 22 3.4 The Least Squares problem 50
How to choose h 22 3.4.1 Method of normal
Truncation error for equations 51
forward difference 23 3.4.2 Least Squares via QR
2.4 Numerical instability and factorization 54
ill-conditioning 24 3.4.3 Least Squares via SVD
Example of a numerically decomposition 55
unstable algorithm 24 3.4.4 Final remarks 56
The backslash operator in 6.2.2 Mersenne Twister 107
MATLAB 56 6.3 Nonuniform distributions 107
Appendix 3.A Solving linear 6.3.1 The inversion method 107
systems in R 56 6.3.2 Acceptance–rejection
solve 57 method 109
Least Squares 58 6.4 Specialized methods for
selected distributions 111
4. Finite difference methods 6.4.1 Normal distribution 111
6.4.2 Higher order moments
4.1 An example of a numerical and the Cornish–Fisher
solution 61 expansion 113
A first numerical 6.4.3 Further distributions 114
approximation 62 6.5 Sampling from a discrete set 116
A second numerical 6.5.1 Discrete uniform selection 116
approximation 63 6.5.2 Roulette wheel selection 117
4.2 Classification of differential 6.5.3 Random permutations and
equations 64 shuffling 118
4.3 The Black–Scholes equation 65 6.6 Sampling errors—and how to
4.3.1 Explicit, implicit, and reduce them 118
θ -methods 67 6.6.1 The basic problem 118
4.3.2 Initial and boundary 6.6.2 Quasi-Monte Carlo 119
conditions and definition 6.6.3 Stratified sampling 120
of the grid 67 6.6.4 Variance reduction 121
4.3.3 Implementation of the 6.7 Drawing from empirical
θ -method with MATLAB 71 distributions 122
4.3.4 Stability 73 6.7.1 Data randomization 122
4.3.5 Coordinate transformation 6.7.2 Bootstrap 122
of space variables 76 6.8 Controlled experiments and
4.4 American options 79 experimental design 127
Appendix 4.A A note on MATLAB’s 6.8.1 Replicability and ceteris
function spdiags 86 paribus analysis 127
6.8.2 Available random number
5. Binomial trees generators in MATLAB 128
6.8.3 Uniform random numbers
5.1 Motivation 89 from MATLAB’s rand
Matching moments 89 function 128
5.2 Growing the tree 90 6.8.4 Gaussian random numbers
5.2.1 Implementing a tree 91 from MATLAB’s randn
5.2.2 Vectorization 92 function 129
5.2.3 Binomial expansion 93 6.8.5 Remedies 131
5.3 Early exercise 95
5.4 Dividends 95 7. Modeling dependencies
5.5 The Greeks 97
Greeks from the tree 97 7.1 Transformation methods 133
7.1.1 Linear correlation 133
7.1.2 Rank correlation 138
Part II 7.2 Markov chains 144
Simulation 7.2.1 Concepts 144
6. Generating random 7.2.2 The Metropolis algorithm 146
7.3 Copula models 148
numbers 7.3.1 Concepts 148
6.1 Monte Carlo methods and 7.3.2 Simulation using copulas 150
sampling 103
6.1.1 How it all began 103 8. A gentle introduction
6.1.2 Financial applications 104
to financial simulation
6.2 Uniform random number
generators 104 8.1 Setting the stage 153
6.2.1 Congruential generators 104 8.2 Single-period simulations 154
8.2.1 Terminal asset prices 154 Part III
8.2.2 1-over-N portfolios 155 Optimization
8.2.3 European options 157
10. Optimization problems in
8.2.4 VaR of a covered put
portfolio 159 finance
8.3 Simple price processes 161 10.1 What to optimize? 219
8.4 Processes with memory in the 10.2 Solving the model 220
levels of returns 163 10.2.1 Problems 220
8.4.1 Efficient versus adaptive 10.2.2 Classical methods and
markets 163 heuristics 222
8.4.2 Moving averages 163 10.3 Evaluating solutions 222
8.4.3 Autoregressive models 164 10.4 Examples 224
8.4.4 Autoregressive moving Portfolio optimization
average (ARMA) models 165
with alternative risk
measures 224
8.4.5 Simulating ARMA models 166
Model selection 225
8.4.6 Models with long-term Robust/resistant
memory 167 regression 225
8.5 Time-varying volatility 169 Agent-based models 226
8.5.1 The concepts 169 Calibration of
8.5.2 Autocorrelated option-pricing models 226
time-varying volatility 170 Calibration of yield
8.5.3 Simulating GARCH structure models 227
processes 173 10.5 Summary 228
8.5.4 Selected further
autoregressive volatility 11. Basic methods
models 175
11.1 Finding the roots of f (x) = 0 229
8.6 Adaptive expectations and
11.1.1 A naïve approach 229
patterns in price processes 178
Graphical solution 230
8.6.1 Price–earnings models 178 Random search 231
8.6.2 Models with learning 179 11.1.2 Bracketing 231
8.7 Historical simulation 180 11.1.3 Bisection 232
8.7.1 Backtesting 180 11.1.4 Fixed point method 233
8.7.2 Bootstrap 181 Convergence 235
8.8 Agent-based models and 11.1.5 Newton’s method 238
complexity 185 Comments 240
11.2 Classical unconstrained
optimization 241
9. Financial simulation at Convergence 242
work: some case studies 11.3 Unconstrained optimization
in one dimension 243
9.1 Constant proportion portfolio 11.3.1 Newton’s method 243
insurance (CPPI) 189 11.3.2 Golden section search 244
9.1.1 Basic concepts 189 11.4 Unconstrained optimization
9.1.2 Bootstrap 191 in multiple dimensions 245
9.2 VaR estimation with Extreme 11.4.1 Steepest descent
Value Theory 192 method 245
9.2.1 Basic concepts 192 11.4.2 Newton’s method 247
9.2.2 Scaling the data 193 11.4.3 Quasi-Newton method 248
9.2.3 Using Extreme Value 11.4.4 Direct search methods 250
Theory 193 11.4.5 Practical issues with
MATLAB 254
9.3 Option pricing 195
11.5 Nonlinear Least Squares 256
9.3.1 Modeling prices 196 11.5.1 Problem statement and
9.3.2 Pricing models 199 notation 256
9.3.3 Greeks 208 11.5.2 Gauss–Newton
9.3.4 Quasi-Monte Carlo 210 method 257
11.5.3 Levenberg–Marquardt Appendix 12.B Parallel
method 258 computations in MATLAB 309
11.6 Solving systems of nonlinear 12.B.1 Parallel execution of
equations F (x) = 0 260 restart loops 311
11.6.1 General considerations 260 Appendix 12.C Heuristic methods
11.6.2 Fixed point methods 262 in the NMOF package 314
11.6.3 Newton’s method 263 12.C.1 Local Search 314
11.6.4 Quasi-Newton 12.C.2 Simulated Annealing 315
methods 268 12.C.3 Threshold Accepting 315
11.6.5 Further approaches 269 12.C.4 Genetic Algorithm 315
11.7 Synoptic view of solution 12.C.5 Differential Evolution 316
methods 270 12.C.6 Particle Swarm
Optimization 316
12.C.7 Restarts 317
12. Heuristic methods in a
nutshell
13. Heuristics: a tutorial
12.1 Heuristics 273
13.1 On Optimization 319
What is a heuristic? 274
13.1.1 Models 319
Iterative search 275
13.1.2 Methods 320
12.2 Single-solution methods 276
13.2 The problem: choosing few
12.2.1 Stochastic Local
from many 320
Search 276
13.2.1 The subset-sum
12.2.2 Simulated Annealing 277
problem 320
12.2.3 Threshold Accepting 278
13.2.2 Representing a
12.2.4 Tabu Search 279
solution 321
12.3 Population-based methods 279 13.2.3 Evaluating a solution 321
12.3.1 Genetic Algorithms 279 13.2.4 Knowing the solution 322
12.3.2 Differential Evolution 280 13.3 Solution strategies 323
12.3.3 Particle Swarm 13.3.1 Being thorough 323
Optimization 281 13.3.2 Being constructive 324
12.3.4 Ant Colony 13.3.3 Being random 325
Optimization 282 13.3.4 Getting better 327
12.4 Hybrids 282 13.4 Heuristics 330
12.5 Constraints 284 13.4.1 On heuristics 330
12.6 The stochastics of heuristic 13.4.2 Local Search 331
search 285 13.4.3 Threshold Accepting 336
12.6.1 Stochastic solutions 13.4.4 Settings, or: how (long)
and computational to run an algorithm 340
resources 285 13.4.5 Stochastics of LS and
12.6.2 An illustrative TA 340
experiment 287 13.5 Application: selecting
12.7 General considerations 289 variables in a regression 342
12.7.1 What technique to 13.5.1 Linear models 342
choose? 289 13.5.2 Fast least squares 343
12.7.2 Efficient 13.5.3 Selection criterion 344
implementations 289 13.5.4 Putting it all together 345
12.7.3 Parameter settings 293 13.6 Application: portfolio
12.8 Outlook 294 selection 347
Appendix 12.A Implementing 13.6.1 Models 347
heuristic methods with 13.6.2 Local-Search
MATLAB 294 algorithms 348
12.A.1 The problems 296
12.A.2 Threshold Accepting 298
14. Portfolio optimization
12.A.3 Genetic Algorithm 303
12.A.4 Differential Evolution 306 14.1 The investment problem 355
12.A.5 Particle Swarm 14.2 Mean–variance optimization 357
Optimization 308 14.2.1 The model 357
14.2.2 Solving the model 358 15.4 Backtesting portfolio
14.2.3 Examples of strategies 459
mean–variance models 358 15.4.1 Kenneth French’s data
14.2.4 True, estimated, and library 459
realized frontiers 366 15.4.2 Momentum 464
14.2.5 Repairing matrices 368 15.4.3 Portfolio optimization 470
14.3 Optimization with heuristics 377 Appendix 15.A Prices in btest 479
14.3.1 Asset selection with Appendix 15.B Notes on zoo 479
Local Search 377 Appendix 15.C Parallel
14.3.2 Scenario Optimization computations in R 480
with Threshold 15.C.1 Distributed computing 480
Accepting 383 15.C.2 Loops and apply
14.3.3 Portfolio optimization functions 481
with TA: examples 392 15.C.3 Distributing data 482
14.3.4 Diagnostics for 15.C.4 Distributing data,
techniques based on continued 484
Local Search 411 15.C.5 Other functions in the
14.4 Portfolios under parallel package 486
Value-at-Risk 413 15.C.6 Parallel computations
14.4.1 Why Value-at-Risk in the NMOF package 486
matters 413
14.4.2 Setting up experiments 414 16. Econometric models
14.4.3 Numerical results 415
16.1 Term structure models 487
Appendix 14.A Computing 16.1.1 Yield curves 487
returns 419 16.1.2 The Nelson–Siegel
Appendix 14.B More model 492
implementation issues in R 420 16.1.3 Calibration strategies 496
14.B.1 Scoping rules in R and 16.1.4 Experiments 518
objective functions 420 16.2 Robust and resistant
14.B.2 Vectorized objective regression 522
functions 422 16.2.1 The regression model 525
Appendix 14.C A neighborhood 16.2.2 Estimation 527
for switching elements 424 16.2.3 An example 532
16.2.4 Numerical
15. Backtesting experiments 535
16.2.5 Final remarks 540
15.1 What is (the problem with) 16.3 Estimating Time Series
backtesting? 427 Models 542
15.1.1 The ugly: intentional 16.3.1 Adventures with time
overfitting 428 series estimation 542
15.1.2 The bad: unintentional 16.3.2 The case of GARCH
overfitting and other models 543
difficulties 434 16.3.3 Numerical
15.1.3 The good: getting experiments with
insights (and Differential Evolution 545
confidence) in Appendix 16.A Maximizing the
strategies 436 Sharpe ratio 549
15.2 Data and software 437
15.2.1 What data to use? 437 17. Calibrating option pricing
15.2.2 Designing backtesting
software 439 models
15.2.3 The btest function 439 17.1 Implied volatility with
15.3 Simple backtests 440 Black–Scholes 552
15.3.1 btest: a tutorial 440 The smile 554
15.3.2 Robert Shiller’s 17.2 Pricing with the characteristic
Irrational-Exuberance function 555
data 450 17.2.1 A pricing equation 555
17.2.2 Numerical integration 560 A. The NMOF package
17.3 Calibration 580
17.3.1 Techniques 580 A.1 Installing the package 597
17.3.2 Organizing the A.2 News, feedback and
problem and discussion 597
implementation 582 A.3 Using the package 597
17.3.3 Two experiments 589
17.4 Final remarks 593
Appendix 17.A Quadrature rules Bibliography 599
for infinity 594 Index 609
List of figures

Fig. 1.1 A map of the beautiful country called Switzerland. 4


Fig. 1.2 Left: The S&P 500 in 2009. Right: Annual returns after jackknifing two observations. The vertical
line gives the realized return. 13
Fig. 1.3 Payoff of replicating portfolios with delta to double precision (left), and delta to two digits (right). 15
Fig. 2.1 Function f (x) = cos(x x ) − sin(ex ) (left panel) and relative error of the numerical approximation of
the derivative at x = 1.5 for forward difference (thick line) and central difference (thin line). 24
Fig. 2.2 Left panel: graphical solution of linear system (2.4). Right panel: details around the coordinate
(−3, 4.54) of the straight lines defining the linear system. 26
Fig. 2.3 Operation count for algorithm A1 (triangles) and A2 (circles). 29
Fig. 3.1 LU factorization. 33
Fig. 3.2 Grid search for optimal ω. 43
Fig. 3.3 Example of a decomposable matrix (left) and its block triangular form (right). 49
Fig. 3.4 Example of a matrix of order 8 with structural rank 6 and the corresponding 6 × 4 zero submatrix
(right panel). 50
Fig. 4.1 Graph of function f (x) = 2/x . 62
Fig. 4.2 Numerical approximation with the explicit method (circles) of the differential equation defined in
Eq. (4.1) for N = 10 (left panel) and N = 30 (right panel). 63
Fig. 4.3 Numerical approximation with the implicit method (circles) of the differential equation defined in
Eq. (4.1) for xN = 30 and N = 30 (left panel) and xN = 100 and N = 80 (right panel). 64
Fig. 4.4 Price of call option (left panel) and price of put option (right panel) for X = 20, r = 0.10, q = 0 and
σ = 0.60. 66
Fig. 4.5 Finite difference grid with terminal conditions (dark circles) and boundary conditions (gray circles). 68
Fig. 4.6 Four points involved in the numerical approximation of the derivatives of vij . 68
Fig. 4.7 Explicit method (left) and implicit method (right). 71
Fig. 4.8 M = 30 (left panel) and M = 500 (right panel). Common parameters: N = 100, σ = 0.20, r = 0.05,
q = 0, X = S = 10. 74
Fig. 4.9 σ = 0.20 (left panel) and σ = 0.40 (right panel). Common parameters: N = 100, M = 30, r = 0.05,
q = 0, X = S = 10. 74
Fig. 4.10 N = 310 (left panel) and N = 700 (right panel). Common parameters: M = 30, σ = 0.20, r = 0.05,
q = 0, X = 10. 74
Fig. 4.11 N = 700, M = 30, σ = 0.50, r = 0.05, X = 10. 75
Fig. 4.12 Price of down-and-out put (lower line) and European put (upper line) as a function of the price of the
underlying at the beginning of the option’s life. 76
Fig. 4.13 Coordinate transformations for N = 29 and X = 20. Left panel: Hyperbolic sine function with
λ = 10 and p = 0.4. Right panel: Logarithmic transformation with I = 2. 78
Fig. 4.14 Absolute errors for the hyperbolic sine transformation (circles) and the logarithmic transformation
(plus signs). 78
Fig. 4.15 Finite difference grid for Smin = 0, Smax = 80, N = 40, M = 50, and θ = 12 . Left panel: Dark nodes
indicate option price greater than payoff. Right panel: Option prices and early exercise boundary. 81
Fig. 4.16 Early exercise boundary (same setting as in Fig. 4.15). 82
Fig. 4.17 Left panel: Grid nodes used for interpolation. Right panel: Early exercise boundary resulting from
interpolation of the solutions obtained with the explicit payout method. 82
Fig. 4.18 Computation of early exercise boundaries with finer grid (Smax = 150, N = 300, and M = 50).
PSOR (left panel) and EP (right panel). 83
Fig. 4.19 Grid search for the optimal omega in the PSOR algorithm. 83
Fig. 4.20 Price of American strangle (dashed line is the payoff). 86
Fig. 6.1 Random numbers from the randu function, where ui = (65,539 ui−1 ) mod 231 . 106
Fig. 6.2 Confidence bands and probabilities for realizations within a region (x , xh ). 108
Fig. 6.3 Inversion principle. 108
Fig. 6.4 Numbers within the unit square before (top panel) and after (bottom panel) Box–Muller
transformation. 111
Fig. 6.5 Required skewness (left panel) and kurtosis (right panel), and sample moments with 500
observations per sample. 114
Fig. 6.6 Gaussian kernel density estimates with smoothing parameters h = 0.025, 0.1, and 0.5, respectively
(darker lines for larger h). 126
Fig. 6.7 Samples generated with the Taylor–Thompson algorithm; original observations (circles) from a
(0, 1) Gaussian distribution. 127
Fig. 6.8 First uniform PRN drawn for different seeds and RNGs. 129
Fig. 6.9 Scatterplots for first uniform PRNs drawn from seeds s and s + 1. 129
Fig. 6.10 First normal PRN (polar) drawn for seeds s = 1, . . . , 10,000. 130
Fig. 6.11 First normal PRN (ziggurat) drawn for seeds s = 1, . . . , 600,000. 130
Fig. 6.12 Scatterplot first draws from three subsequent seeds s = 1, . . . , 5000. 130
Fig. 6.13 First PRN drawn for seeds s = 1, . . . , 1000, x1,s plotted against x1,s+d with d = 100, 1000 and
10,000 (left to right) for the polar (top panel) and ziggurat (bottom panel) RNGs. 131
Fig. 6.14 101st PRN drawn for seeds s = 1, . . . , 1000, x101,s for the uniform (top panel) and normal (bottom
panel) RNGs. 131
Fig. 7.1 Left: scatter plot of three uncorrelated Gaussian variates. Right: scatter plot of three Gaussian
variates with ρ = 0.7. 135
Fig. 7.2 Bivariate normal samples, generated with the Metropolis algorithm.
Left panel: metropolisMVNormal(2000,2,0.2,0.75) (s = 0.2).
Right panel: metropolisMVNormal(2000,2,1.0,0.75) (s = 1.0). 147
Fig. 7.3 Densities for u and x (under the assumption of Gaussian marginals) and 500 samples. 151
Fig. 8.1 Simulated returns and stock prices for 100 (left) and 10,000 samples (right). White circles are
theoretical quantiles and black ones are their empirical counterparts. Theoretical and empirical
means are white and black stars, respectively. 155
Fig. 8.2 Returns for portfolios consisting of N equally weighted assets (“1-over-N ”). 157
Fig. 8.3 Histograms for the value of a stock (top panel) and a portfolio consisting of the stock plus one
European put (bottom panel) at times τ . 161
Fig. 8.4 Left panel: Value-at-Risk for stock (dotted lines) and portfolio (solid line) for 1%, 5%, and 10%
quantiles (light to dark). Right panel: 5%, 25%, 50%, 75%, and 95% percentiles of log returns for
stock (dotted) and portfolio (solid line). 161
Fig. 8.5 Returns and prices for different AR, MA, and ARMA models. 168
Fig. 8.6 Scaled white noise; dark lines indicate ±σt . 170
Fig. 8.7 Daily returns (gray lines) and μ ± 2σt confidence bands (dark lines) (left panel) for different
combinations of α1 + β1 = 0.95 and prices (right panel) for GARCH(1, 1) processes. 171

Fig. 8.8 Actual daily log returns for the FTSE and S&P 500 for the period Jan 2000 √ to Dec 2009 and ±2 ht
(top panel), and innovations standardized using time-varying volatility, ht , as fitted with a
GARCH(1, 1) (bottom panel). 173
Fig. 8.9 Actual FTSE daily log returns for July 2004 to June 2010 (left) and two simulations based on a
GARCH model fitted on actual data (center and right; 2525 observations each; seeds fixed with
randn(’seed’,10) and randn(’seed’,50), respectively). 174
Fig. 8.10 Kernel densities for cumulated returns (250 days; left panel) and prices (right panel) where daily
returns follow a GARCH(1, 1) process with α1 ∈ {0.05, 0.20.350.95} and β1 = 0.95 − α1 (darker
lines indicate lower α1 ). 175
Fig. 8.11 Stock prices under rational expectations from the Timmermann (1993) model (n = 25, μ = 0.01,
σ = 0.05). Rational expectation prices (thin gray line) and fundamental prices with known
parameters (thick gray line). 180
Fig. 8.12 Bootstrap samples for FTSE data. 182
Fig. 8.13 Moments of a buy-and-hold portfolio (thick gray lines) and its constituents (thin black lines; FTSE,
DAX, and EuroStoxx) over different lengths of investment horizons (x-axis: T = 1, 5, 21, 62 days;
corresponding to 1 day, 1 week, 1 month, 1 quarter); based on original data August 2005 to July
2010 and 1,000,000 bootstraps. Top panel, block length b = 1; center panel, block length b = 5; and
bottom panel, block length b = T . 184
Fig. 8.14 Price changes (top panel) and fraction of fundamentalists in the market (bottom panel) from an
agent-based simulation (Algorithm 27). 187
Fig. 9.1 CPPI composition for two sample price processes and different multipliers time to maturity of 2
years and daily readjustment (“no gap risk”) and quarterly readjustment (rightmost; “with gap risk”).
Light gray: safe asset, Bt ; dark gray: exposure, Et ; white line: floor, Ft . 191
Fig. 9.2 FTSE time series, 10-block bootstraps for risky asset and trajectories for CPPI. 192
Fig. 9.3 Terminal value of CPPI with FTSE as T = 1 year, multiplier m = 4, and different gap lengths,
simulated with block bootstrap (right). 192
Fig. 9.4 Median of ratios VaRHill /VaRtheo for different underlying distributions (panels) and different
fractions of data used, m/N (lines). 196
Fig. 9.5 20 paths of geometric Brownian motion. 199
Fig. 9.6 Five paths of a geometric Brownian bridge, starting at 100 and ending at 103. 199

Fig. 9.7 Speed of convergence for Monte Carlo pricing. S0 = 50, X = 50, r = 0.05, v = 0.30, and τ = 1.
The true Black–Scholes price is 7.12. 202
Fig. 9.8 Forward difference for Greeks: Boxplots for Delta estimates with M = 1 and N = 100,000 for
different values of h (y-axis). Parameters are S = 100, X = 100, τ = 1, r = 0.03, q = 0, and σ = 0.2.
The solid horizontal line is the true BS Delta of 0.60; the dotted lines give Delta ±0.05. 209
Fig. 9.9 Forward difference for Greeks: Boxplots for Delta estimates with M = 1 and N = 100,000 for
different values of h (y-axis), but using common random variates. Parameters are S = 100, X = 100,
τ = 1, r = 0.03, q = 0, and σ = 0.2. The solid horizontal line is the true BS Delta of 0.60; the dotted
lines give Delta ±0.05. 210
Fig. 9.10 Top left: scatter of 500 against 500 points generated with MATLAB’s rand. Top right: scatter of 500
against 500 generated with Halton sequences (bases 2 and 7). Bottom left: scatter of 500 against 500
generated with Halton sequences (bases 23 and 29). Bottom right: scatter of 500 against 500
generated with Halton sequences (bases 59 and 89). 214
Fig. 9.11 Convergence of price with quasi-MC (function callBSMQMC). The light gray lines show the
convergence with standard MC approach. 215
Fig. 9.12 Three paths generated with bases 3, 13, and 31. 215
Fig. 10.1 Objective function for Value-at-Risk. 225
Fig. 10.2 LMS objective function. 226
Fig. 10.3 Simulated objective function for Kirman’s model for two parameters. 227
Fig. 10.4 Left panel: Heston model objective function. Right panel: Nelson–Siegel–Svensson model objective
function. 228
Fig. 11.1 Synoptic view of methods presented in the chapter. 230
Fig. 11.2 Left panel: iteration function g1 . Right panel: iteration function g2 . 234
Fig. 11.3 Left panel: iteration function g3 . Right panel: iteration function g4 . 235
Fig. 11.4 Left panel: iteration function satisfies g  (x) < −1. Right panel: iteration function satisfies
−1 < g  (x) < 0. 236
Fig. 11.5 Shape of function ρk (left panel) and shape of function ρk (s) φ(s) (right panel). 237
Fig. 11.6 Behavior of the Newton method for different starting values. Upper left: x0 = 2.750 and
x sol = 2.4712. Upper right: x0 = 0.805 and x sol = 2.4712. Lower left: x0 = 0.863 and
x sol = 1.4512. Lower right: x0 = 1.915 and algorithm diverges. 240
 
Fig. 11.7 Minimization of f (x1 , x2 ) = exp 0.1 (x2 − x12 )2 + 0.05 (1 − x1 )2 with the steepest descent
method. Right panel: minimization of α for the first step (α ∗ = 5.87). 247
 
Fig. 11.8 Minimization of f (x1 , x2 ) = exp 0.1 (x2 − x12 )2 + 0.05 (1 − x1 )2 with Newton’s method. Contour
plots of the local model for the first three steps. 248
Fig. 11.9 Nonconvex local model for starting point (left panel) and convex local model but divergence of the
step (right panel). 249
Fig. 11.10 Comparison of the first seven Newton iterations (circles) with the first five BFGS iterations (triangles). 250
Fig. 11.11 Evolution of starting simplex in the Nelder–Mead algorithm. 251
Fig. 11.12 Detailed starting simplex and two reflections. 251
Fig. 11.13 Left panel: Plot of observations, model for x1 = 2.5, x2 = 2.5 and residuals. Right panel: plot of sum
of squared residuals g(x). 259
Fig. 11.14 Starting point (bullet), one step (circle) and contour plots of local models for Newton (upper right),
Gauss–Newton (lower left) and Levenberg–Marquardt (lower right). 260
Fig. 11.15 Solutions for varying values of parameter c of the systems of two nonlinear equations. 261
Fig. 11.16 Steps of the fixed point algorithm for the starting solution y0 = [ 1 5 ] . 264
Fig. 11.17 Oscillatory behavior of the fixed point algorithm for the starting solution y0 = [ 2 0 ] . 264
Fig. 11.18 Steps of the Newton algorithm for starting solution y0 = [ 1 5] . 266
Fig. 11.19 Steps of the Broyden algorithm for starting point y0 = [5/2 1] and identity matrix for B (0) . 268
Fig. 11.20 Steps of the Broyden algorithm for starting point y0 = [5/2 1] and Jacobian matrix evaluated at the
starting point for B (0) . 269
Fig. 11.21 Objective function minimizing F (y)2 for the solution of the system of nonlinear equations defined
in Example 11.4. 270
Fig. 11.22 Synoptic view of solution methods for systems of linear and nonlinear equations. 271
Fig. 12.1 Scheme for possible hybridizations. 283
Fig. 12.2 Shekel function. For better visibility −f has been plotted. 297
Fig. 12.3 Shekel function. Empirical distribution of solutions. 301
Fig. 12.4 Shekel function. Left panel: Threshold sequence. Right panel: Objective function values over the 6
rounds of the 5th restart. 301
Fig. 12.5 Subset sum problem. Left panel: Threshold sequence. Right panel: Objective function values over
the 6 rounds of the 5th restart. 303
Fig. 12.6 Subset sum problem: Evolution of the fittest solution in restart 3. The algorithm stops at generation 6. 305
Fig. 12.7 Differential Evolution optimization. Left panel: Empirical distribution of solutions. Right panel: Best
value over generations of restart 3. 307
Fig. 12.8 Particle Swarm optimization. Left panel: Empirical distribution of solutions. Right panel: Best value
over generations of restart 1. 309
Fig. 12.9 Code DCex.m. Theoretical speedup and efficiency as a function of processors. Dots represent
computed values. 311
Fig. 13.1 The upper panel shows the left tail of the distribution of objective function values for a sample of
random solutions. We sample uniformly, and thus the distribution function resembles a straight line.
The lower panel shows the distribution for the best 100 of the one million random solutions. 327
Fig. 13.2 Distributions of objective function values of best 100 random solutions and 100 greedy solutions.
Best possible solution would be zero. 330
Fig. 13.3 Objective function values of an unguided (random) walk through the search space. 333
Fig. 13.4 Distributions of changes in objective function values when 1 (dark gray), 5 (gray), and 10 (light gray)
elements are changed. 333
Fig. 13.5 No structure in the objective function. In search spaces such as the ones pictured a Local Search will
fail, because the objective function has no local structure, i.e. it provides no guidance. Being close to
a good solution cannot be exploited by the algorithm. 334
Fig. 13.6 Objective function value for two runs of Local Search over 50,000 iterations (log scale). 336
Fig. 13.7 Distributions of objective function values of best 100 random solutions, 100 greedy solutions, 100
solutions obtained by Local Search and 100 solutions obtained by Threshold Accepting. Best
possible solution would be zero. 342
Fig. 14.1 True frontier and estimated and realized frontiers (no short sales, nA is 25, nS is 100). 367
Fig. 14.2 Portfolio return distributions: the influence of rd . 387
Fig. 14.3 In-sample versus out-of-sample differences. Left: in-sample difference of a solution pair (QP minus
TA) against the associated out-of-sample difference. Right: same, but only for portfolios for which
the in-sample difference is less than one basis point. 398
Fig. 14.4 In-sample versus out-of-sample difference depending on the number of iterations. 399
Fig. 14.5 Squared returns (left panel) and ratio of conditional moments (right panel) for different portfolio
weights. 402
Fig. 14.6 Distributions of objective function values for random portfolios (light gray) and solutions of TAopt
and LSopt (darker gray). There is little difference between Local Search and Threshold Accepting. 411
Fig. 14.7 Distributions of objective function values with 500, 1000 and 2000 iterations. The lighter-gray lines
belong to TAopt, the darker-gray ones to LSopt. There is virtually no difference between the lines
for Local Search and Threshold Accepting with 2000 iterations; both methods converge quickly on
the same portfolio. 411
Fig. 14.8 Distribution of end-of-horizon wealth in a VaR setting. 416
Fig. 15.1 Daily returns of the S&P 500, including October 1987. 428
Fig. 15.2 S&P 500 without 10 best and 10 worst days. 428
Fig. 15.3 A random series created with function randomPriceSeries. 430
Fig. 15.4 Distribution of difference in final profits MA-crossover strategy vs buy-and-hold. A positive number
means that MA-crossover outperformed the underlying asset. The distribution is symmetric about
zero, which indicates that there is no systematic advantage to the crossover strategy. 431
Fig. 15.5 Final profits of MA-crossover strategy vs result of buy-and-hold of the same underlying series. 432
Fig. 15.6 Equity curves of optimized backtests. 433
Fig. 15.7 Distribution of in-sample excess profits for optimized MA crossover strategy. A positive number
means that MA crossover outperformed the underlying asset. 434
Fig. 15.8 Observational Equivalence: EURO STOXX Total Market and EURO STOXX Banks. 435
Fig. 15.9 Mechanics of a walk-forward. 437
Fig. 15.10 Survivorship bias when using current components of an index 438
Fig. 15.11 S&P stock market index since January 1871 452
Fig. 15.12 CAPE ratio 452
Fig. 15.13 Results of avoiding high valuation. The S&P index is shown in black; the strategy in gray. 455
Fig. 15.14 Results of avoiding high valuation compared with S&P 500. 455
Fig. 15.15 Performance of strategy for different values of q. The S&P index is shown in black; the strategy
variations in gray. 457
Fig. 15.16 Performance of industry portfolios as provided by Kenneth French’s data library, January 1990 to
May 2018. Time-series are computed from daily returns. 462
Fig. 15.17 Fanplot of the industry time-series shown in Fig. 15.16. 462
Fig. 15.18 Correlations of monthly returns. 463
Fig. 15.19 Benchmarks: Performance of market portfolio and of equally weighted portfolio. The gray shades
are the same as in Fig. 15.17 and indicate the range of performance of the different industries. 464
Fig. 15.20 Benchmarks: Outperformance of market versus equally weighted portfolio. A rising line indicates
the outperformance of the market portfolio, and vice versa. 464
Fig. 15.21 Performance of momentum strategy. The vertical axis uses a log scale. 466
Fig. 15.22 Performance of momentum strategy. The vertical axis uses a log scale. The gray line shows the
absolute performance of the market. The black line shows the relative performance of momentum
compared with the market, i.e. a rising line indicates an outperformance of momentum. 467
Fig. 15.23 Results from sensitivity check: Fanplots of paths with frequent rebalancing (the upper band) and less
frequent rebalancing (the lower band). 469
Fig. 15.24 Results from sensitivity check: The ratio of the median paths of the fanplots in Fig. 15.23. The
steadily rising line indicates that frequent rebalancing outperforms less frequent rebalancing. 470
Fig. 15.25 Results from sensitivity checks: densities of annualized returns of the fanplots in Fig. 15.23. The
distribution to the right belongs to the paths with more frequent rebalancing. 470
Fig. 15.26 Performance of the long-only minimum-variance portfolio (gray) vs market (black). On Nov 19,
1999, both series have a value of 100 (the MV time-series starts later because of its burn-in of
10 years). 472
Fig. 15.27 Performance of 100 walk-forwards with random rebalancing periods. The portfolios are computed
with QP. All randomness in the results comes through differences in the setup; there is no numeric
randomness. 475
Fig. 15.28 Correlation of daily returns when portfolios are rebalanced at random timestamps. 476
Fig. 15.29 Positions in a single industry across four backtest variations. The backtests differ only in their
rebalancing schedules; hence the positions are very similar. 477
Fig. 15.30 Performance of 100 walk-forwards with fixed rebalancing periods. The portfolios are computed with
Local Search. All variations in the results stem from the randomness inherent in Local Search
(though it is barely visible); there are no other elements of chance in the model. Compare with
Fig. 15.27. 478
Fig. 16.1 Interpolation and approximation. The panels show a set of market rates. Left: we interpolate the
points. Right: we approximate the points. 490
Fig. 16.2 The matrix C for the cash flows of the 44 bonds in bundData, each square represents a nonzero
entry. Each row gives the cash flows of one bond, each column is associated with one payment date. 492
Fig. 16.3 Level. The left panel shows y(τ ) = β1 = 3. The right panel shows the corresponding yield curve, in
this case also y(τ ) = β1 = 3. The influence of β1 is constant for all τ . 493
 
1−exp(−τ/λ1 )
Fig. 16.4 Short-end shift. The left panel shows y(τ ) = β2 τ/λ1 for β2 = −2. The right panel shows
 
1−exp(−τ/λ1 )
the yield curve resulting from the effects of β1 and β2 , that is, y(τ ) = β1 + β2 τ/λ1 for
β1 = 3, β2 = −2. The short end is shifted down by 2%, but then the curve grows back to the long-run
level of 3%.   493
1−exp(−τ/λ1 )
Fig. 16.5 Hump. The left panel shows β3 τ/λ1 − exp(−τ/λ1 ) for β3 = 6. The right panel shows the
yield curve resulting from all three components. In all panels, λ1 is 2. 493
Fig. 16.6 Nelson–Siegel–Svensson: The three panels show the correlation between the second and the third,
the second and the fourth, and the third and the fourth regressors in Eqs. (16.10) for different
λ-values (the x- and y-axes show λ1 and λ2 between 0 and 25). 496
Fig. 16.7 Distributions of estimated parameters. The true parameters are 4, −2, and 2. 498
Fig. 16.8 Negative interest rates with Nelson–Siegel model despite parameter restrictions. 499
Fig. 16.9 Results for Case 2, Nelson–Siegel (NS) model: True model yields, yield curves fitted with DEopt,
and yield curves fitted with nlminb. Plotted are the results of 5 restarts for both methods, though
for DEopt it is impossible to distinguish between the different curves. 507
Fig. 16.10 Results for Case 2, Nelson–Siegel–Svensson (NSS) model: True model yields, yield curves fitted
with DEopt, and yield curves fitted with nlminb. Plotted are the results of 5 restarts for both
methods, though for DEopt it is impossible to distinguish between the different curves. Compare
with Fig. 16.9: if anything, the results for nlminb are worse. 509
Fig. 16.11 Results for Case 3, Nelson–Siegel–Svensson (NSS) model: True model yields, yield curves fitted
with DEopt, and yield curves fitted with nlminb. Plotted are the results of 5 restarts for both
methods, though for DEopt it is impossible to distinguish between the different curves. 512
Fig. 16.12 Results for Case 4, Nelson–Siegel–Svensson (NSS) model: True model yields, yield curve fitted with
DEopt, and yield curve fitted with nlminb. Plotted are the results of 5 restarts for both methods,
though for DEopt it is impossible to distinguish between the different curves. 517
Fig. 16.13 Left: Convergence of solutions. Each distribution function shows the objective function value of the
best population member from 1000 runs of DEopt. As the number of generations increases, the
distributions become steeper and move to the left (zero is the optimal objective function value).
Right: Convergence of population, i.e. objective function values across members at the end of
DEopt run. The light-gray distributions are from runs with 100 generations; the darker (and steeper)
ones come from 500 generations. Those latter distributions typically converge: all solutions in the
population are the same. 518
Fig. 16.14 OF over time. The left panel shows F = 0.5, CR = 0.99. For the same problem, the right panel shows
F = 0.9, CR = 0.5. 520
Fig. 16.15 The true value of β1 is 5. Left: without constraint (penalty weight is zero). Right: with constraint
β1 ≤ 4. 522
Fig. 16.16 The effects of a single outlier. On 6 June 2006, Adidas, a large German producer of clothing and
shoes, made a stock split 4-for-1. The data was obtained from www.yahoo.com in April 2009.
According to www.yahoo.com the series in the upper panel is split adjusted. It is ironic that this is
even true: but the price was adjusted twice, which resulted in the price jump in June 2006 (topmost
panel). 526
Fig. 16.17 Objective function values obtained with lqs with increasing number of samples (see code example). 535
Fig. 16.18 True θ2 is zero. The outliers lead to a biased value of 0.9. 541
Fig. 16.19 True θ2 is zero. The outliers lead to a biased value of 1.5. 541
Fig. 16.20 Reported parameters for the GARCH(1, 1) estimates from different restarts, all with identical
(optimal) loglikelihood of −1106.60788104129. 547
Fig. 17.1 Implied volatilities for options on the S&P 500 (left) and DAX (right) as of 28 October 2010. 554
Fig. 17.2 Heston model: re-creating the implied volatility surface. The graphics show the BS-implied
volatilities obtained from prices under the Heston model. The panels on the right show the implied
volatility of ATM options. 567
Fig. 17.3 Bates model: re-creating the implied volatility surface. The graphics show the BS-implied volatilities
obtained from prices under the Bates model. The panels on the right show the implied volatility of
ATM options. 568
Fig. 17.4 Merton model: re-creating the implied volatility surface. The graphics show the BS-implied
volatilities obtained from prices under the Merton model. The panels on the right show the implied
volatility of ATM options. Importantly, jumps need to be volatile to get a smile (i.e., vJ ≥ 30%). 569
Fig. 17.5 Relative (left) and absolute (right; in cents) price errors for BS with direct integration with 25 nodes
(compared with analytical solution). 579
Fig. 17.6 Relative (left) and absolute (right; in cents) price errors for BS with direct integration with 50 nodes
(compared with analytical solution). 580
Fig. 17.7 Operations of Nelder–Mead. 581
Fig. 17.8 Absolute errors for Gauss–Legendre and Gauss–Laguerre compared with normcdf for an
increasing number of nodes. Errors are plotted on the y axis; the x axis shows the number of nodes. 595
List of tables

Table 1.1 Yearly returns of S&P 500 1987–2017. 14


Table 6.1 Series of pseudorandom numbers from a linear congruential RNG, where ui = (a ui−1 ) mod 7 and
u0 = 1 and different multipliers a. 106
Table 6.2 Popular kernels, K(y), for kernel density estimation. 125
Table 9.1 The first 10 VDC numbers in base 2. 213
Table 9.2 Prime numbers smaller than 1000. 214
Table 14.1 Average exceedance of out-of-sample shortfall probabilities compared to the target α (in-sample
shortfall probability). First column is the number of assets, while values in italics are those that do
not differ significantly from 0 (5% confidence level). 418
Table 15.1 Statistics of the industry series. Returns are annualized, in percent. Volatility is computed from
monthly returns and also annualized. 460
Table 16.1 A sample data set of German government bonds. Prices as of 31 May 2010. 491
Table 16.2 Errors in % for NSS model for different parameters of DE. Each block corresponds to
F ∈ {0.1, 0.3, 0.5, 0.7, 0.9} for a given F. FE stands for objective function evaluations. 521
Table 16.3 Parameter sensitivity DE. 537
Table 16.4 Parameter sensitivity PSO for δ = 1. 538
Table 16.5 Parameter sensitivity PSO for δ = 0.5. 538
Table 16.6 Parameter sensitivity PSO for δ = 0.75. 539
Table 16.7 Parameter sensitivity PSO for δ = 0.9. 539
Table 16.8 Median reported loglikelihoods for 100 runs with different numbers of generations (nG) and
population size nP. (In brackets: Number of times the overall optimum has been found.) 547
Table 16.9 Estimated GARCH(1, 1) parameters for the DEM2GBP time series and different model settings.
Fixed values in brackets: in models B and C, ȳ = E(y); in models A and B s 2 = E((y − μ̂)2 ). 548
Other documents randomly have
different content
The coach drew up in front of the largest building in the place. It
gave out a pleasant odour as of new pine and clean shavings; across
its front was hung a large sign which announced it to be the
Metropolitan Hotel.
A tall man in his shirt-sleeves, with a sandy beard, and a quill
toothpick held negligently between his teeth, stepped to the coach.
Stephen conjectured that this was no less a person than Mr. Jim
Youtsey himself.
“Friends of the general's?” he inquired affably.
“Yes,” said Stephen, stepping to the ground.
“The general asked me to keep an eye peeled for you. He's over in
the next county, will be back to-morrow if nothing happens—a
splendid man! You couldn't have a stronger indorsement, sir, I'm
glad to know you, glad to welcome you into our midst!” And Mr.
Youtsey shot him a sunny smile over the tip of his toothpick and held
out his hand. “Present me to the Madame—Youtsey's my name.”
Stephen did so, and Mr. Youtsey removed his hat with one hand
and his toothpick with the other; his hat was returned to his head,
and his toothpick to his mouth by a common movement of his two
hands, and he led the way toward his hotel.
“What you see, sir, is the newest thing in Kansas. A year ago there
was nothing here but sunshine and jack-rabbits.”
He further begged Stephen to particularly note that Grant City was
not a cow town; its wealth being derived entirely from the cultivation
of the soil; where were the farms? Just scattered about. Yonder was
the general's office; and through the falling twilight down the street,
Stephen, following the direction of Mr. Youtsey's useful toothpick,
was able to distinguish a very small building with a very large sign;
indeed the number and size of these signs greatly astonished him,
since no building seemed complete without one. Commenting upon
this fact, Mr. Youtsey kindly paused to explain that Grant City had
assembled itself on the prairie with such haste, and with so little
regard for the proper housing of its citizens, that such buildings as
had been erected were not only places of residence, but were used
as offices and stores as well—hence the signs. Having made this
point clear, Mr. Youtsey personally conducted them to their room,
still accompanied by his hat and toothpick, with both of which he
seemed loath to part.
He left them, and presently a small coloured boy appeared with a
pitcher of ice-water, and the information that supper was served. On
going down-stairs to the dining-room, they found Mr. Youtsey at the
head of a long table at which were seated half a score of men.
There immediately followed numerous introductions. Of the ten men,
five, Stephen gathered, were in the real estate business; four were
recent arrivals like himself who were looking about.
The last to be introduced was a small elderly man with a very red
face and a generally dissipated air, whom Mr. Youtsey presented as
Dr. Arling.
“I hope you'll find things home-like here, ma'am,” and Mr. Youtsey
addressed himself to Marian. “We are shy on ladies, it's strictly a
voting population.” Then he permitted his duties as host to absorb
him, and when he had seen that his guests were served, he seated
himself with a pleasant:
“Any one that hasn't had, just holler!”
Stephen's first impression of Grant City had been distinctly
unfavourable, but he said nothing of this to Marian; he felt it would
be wiser to wait until he saw Gibbs before he committed himself to
an opinion.
He saw Gibbs the next morning; on going down to the hotel office
he was welcomed by his friend who fell upon him and fairly
embraced him, then he held him at arm's length.
“Well, Landray, I am glad indeed,” he ejaculated.
The general was not less florid than of yore, but his face had a
battered look; for the rest, he was sleek and prosperous to the eye.
“They tell me you've brought your wife, Landray—that looks as
though you'd come to stay! I'm so sorry my Julia ain't here, but
she's visiting friends in St. Louis. What do you think of this year-old
child of mine? Something to have accomplished in a twelve month?”
and the general patted Stephen affectionately on the back.
Then Stephen must drink with him, and they retired to Mr. Yout-
sey's bar accompanied by Dr. Arling.
“This is to success, Landray!” said Gibbs smiling over the rim of his
glass, and Stephen smiled and nodded, too; Dr. Arling merely tilted
his glass into a toothless cavity and drew the back of his hand across
his lips, for as Mr. Youtsey was accustomed to observe, “He shot his
slugs without a rest.”
“Another round, Jim!” commanded the doctor.
Mr. Youtsey took the toothpick from between his teeth, he had
apparently acquired it along with his clothes when he was dressed,
and said affably:
“Ain't you a little early, Doc? You'll have a pair of hard-boiled eyes
for breakfast if you keep on.”
“Shove along the drinks, don't keep the gentlemen waiting!”
ordered the doctor huskily.
And Mr. Youtsey spun the glasses jingling across the bar.
The breakfast bell sounded a moment or two later, and they left
Dr. Arling leaning limply against the bar, while they repaired to the
dining-room where Gibbs met Marian, and did the honours with
great gallantry.
“Now, my dear lady,” said he, as they rose from the table, his
manner breathing benevolence and urbanity, “I am going to take this
husband of yours down to my office for a chat and smoke.”
When they reached the street, Landray said:
“Well, general, prosperity seems to be smiling on you.”
“It's grinning from ear to ear; Grant City ain't pretty;
architecturally she belongs to the year one—she's about the way
Rome was when Remus got himself disliked by jumping over the
wall; but it's as good as money in the bank.”
“It seems to offer an inviting field for real estate agents,” said
Stephen.
The general laughed.
“Oh, don't let that trouble you, Landray! It's Gibbs's eggs that'll
hatch out first, so don't you worry. We'll stand shoulder to shoulder.
You haven't any capital? Well, no matter—here brains and character
will see you far; and you got both, for you're a Landray.”
They had reached the office by this time, and Gibbs forced his
friend into his best chair; he then provided Stephen and himself with
cigars, and was ready for business.
“I want you to take right hold, Landray, and look after things,” he
said. “I'm going to branch out; I find I can't tie myself down—the
office work don't suit me. We're not only going to sell lots, we're
going to put up the buildings on them as well—this will be in your
department. I'm going to make it my business to keep Grant City
before the public. I'll have a weekly paper running here inside of the
next thirty days, and I got my eye on a seat in the State Legislature,
too; I want to play strong for that, or a worse man may fill it; but
you ain't interested in all this yet; naturally you're wanting to know
where you come in and what I got to offer you. I'm going to make
you the right sort of a proposition here and now, a proposition you
can't afford to turn down.”
And Gibbs was rather better than his word, for just twenty-four
hours later Stephen was established in his office with a satisfactory
salary and an interest in the business as well.
He had been inclined to look upon the town as something of a
farce, but he soon decided that in this hasty opinion he had formed,
his judgment had been at fault. Men straggled in by stage and
prairie schooner; there was a steady demand for lots, indeed the
demand was only exceeded by the supply; and he was rather
dubious as to the wisdom of the town-site speculators who with the
aid of tape line and stakes seemed willing to apportion the greater
part of Kansas to the future needs of Grant City.
“They'll overdo it,” he told Gibbs.
“I guess not,” said Gibbs. “You can't really overdo a good thing,
and Grant City's the best thing in Kansas. It's getting about that it is,
too—the public's waking up to the fact.”
But while Stephen never quite believed in the methods Gibbs
seemed to consider so admirable, he saw that the cheap and hastily
erected houses they were building, principally on credit, all found
tenants the moment they were habitable.
He and Marian lived in one of these houses, one of several that
formed an unpainted row that overlooked the dusty Main Street
which later when the winter rains set in, became a bog that the
citizens—knowing its perils—navigated with caution. It was here that
a son was born them, who was christened Stephen Mason Landray.
Gibbs would have had this event celebrated in some public
manner; for, as he said, little Stephen was the first native-born
citizen of Grant City, but out of consideration for Stephen's wishes in
the matter, he compromised by deeding the child a town lot.
“It will probably be worth thousands by the time he comes of age,
Landray,” he told the father. “But how is Marian?”
Stephen looked grave.
“Why, she don't seem to rally,” he said.
“What does Dr. Arling say?” asked Gibbs.
“He seems to think she will have her strength back as soon as it
gets warmer.”
“To be sure she will, a few mild days will see her up and about. I
have all confidence in Arling; I know—I know, his habits are not
what you'd look for in a man of his attainments; but it's no use to
expect him to be different from what he is.”
Gibbs had established his newspaper, the Kansas Epoch, which he
conducted with much noise and vigour.
“There's no use my denying it,” he told Stephen, “but I got the
editorial faculty. A newspaper in my hands becomes a personal
organ in the best sense. I reckon you can see me in every line. I
take the responsibility pretty seriously, too. I know you don't just
believe in the Epoch, because it don't show a profit in dollars and
cents; but the loss in money's a gain in prestige.”
Which was quite true, for Gibbs was the great man locally; an
admired and respected presence at Mr. Youtsey's bar. It was the
general here, and the general there; the echo of his name constantly
filled the ear. He was vain, bustling, ostentatious; the small fry of
struggling country newspapers heralded his passing to and fro in the
State with much noise; for that fall there had been a hotly contested
State election, and Gibbs with an eye single to his own advancement
had taken the stump. Even so far away as St. Louis one of the big
dailies had chanced to speak of him as “General Gibbs of Kansas,” a
tribute to his growing fame which seemed to argue that he had
already become almost a national figure. Perhaps it was this divided
interest that was responsible for his business methods, for they
struck Stephen as being entirely haphazard, nor could he induce him
to make any change here.
Mrs. Gibbs had returned to Grant City during the winter fully
prepared to make friends with Marian; but Marian conceived a dislike
for her which she was at no pains to conceal, and in the end Julia
informed Gibbs that Landray's wife was a stuck-up little fool, which
seemed to amuse the general immensely.
Stephen felt that it might have been to their advantage had
Marian sought to conciliate Mrs. Gibbs, since he did not know how
Gibbs himself might be affected; but Gibbs was affected not at all;
he commented upon their mutual hostility with characteristic
candour.
“God bless 'em! The ladies can't help it, Landray, and it might be
worse if they were intimate. Now, my Julia can't carry to Marian any
little thing I chance to let drop about you, and Marian can't take up
any slighting criticism you may make of me. I don't know but that
we have a good deal to be thankful for; anyhow—God bless 'em, let
'em fight it out! It won't upset our friendship. I knew from the first
that we were cut out for a business connection; you're long-headed
and conservative; I own I'm something of an idealist—my
imagination runs away with me. You're the steady hand, and by the
time we've talked a scheme over we get together on a pretty sane
conclusion.”
To Stephen this was a flattering opinion that he hoped the general
might never have cause to reconsider.
Apart from his anxiety concerning Marian, Stephen was on the
whole, happy and well content in Grant City. But Marian grew no
better; indeed as the summer advanced, he sometimes fancied he
noted a change for the worse, though Arling, still the only doctor in
the place, pooh-poohed this fear of his.
“Yes, but why don't she get better?” Stephen would demand.
“She will, Landray; give her time. I'm free to say it ain't just a
usual case, but there's no organic trouble, she ought to be a well
woman.” And he would scuttle off in the direction of Mr. Youtsey's
bar, where he could always be found when wanted.
And all this while Grant City seemed to grow and prosper. The
huddle of houses increased by the roadside; there were side streets
sparsely built upon it is true, but there was the beginning of a
perceptible movement in this direction. The place was taking on the
semblance of a town.
The stage carried a constantly increasing traffic; the tri-weekly
gave place to a daily service; next there was both a midmorning and
afternoon stage, but even this failed to entirely meet local needs, for
as Mr. Youtsey said, the town was throwing a fifty-inch chest.
There was talk of issuing bonds for what Gibbs called municipal
improvements; the town-site speculators clamoured for graded
streets and gas, and the general, in the Epoch, demanded a speedy
settlement of the vexing question of the county seat. More than this
he announced that he would personally make it his business to see
that they secured the county seat; and he took himself together with
the logic of the situation to the capital there to lobby for the
measure; and all this while the wilderness of pegs grew out upon
the prairie.
Winter came again and brought a lull to the trafficking; but early
in the spring a corps of railroad engineers pitched their camp on the
outskirts of the town, and more additions were plotted and more
pegs driven to keep pace with the impetus given by this most recent
development.
Gibbs was still at Topeka in the matter of the county seat which he
had not been able to adjust, since Grant City had a rival that dared
to claim this honour for itself, though no one seriously regarded its
claim—certainly no one in Grant City, where the town-site
speculators ridiculed the idea when it was advanced in their hearing,
and Gibbs, in the Epoch, wrote sarcastic editorials that were much
admired, and that proved as clearly as pen and ink can prove
anything, the county seat must come to Grant City.
But one night after Stephen had gone home to his sick wife, he
heard a loud knock at his door, and on going down to answer this
summons, found Gibbs, whom he had supposed to be still in Topeka,
standing on the threshold.
CHAPTER THIRTY-SEVEN

G
IBBS, his battered face red and perspiring, tramped the
length of the room without speech; then he turned and
tramped back again.
“When did you get to town?” asked Stephen, putting down the
lamp he carried. He knew that Gibbs had not come on the afternoon
stage for he had been in the crowd at the hotel when it arrived.
“Oh, I missed the stage, and got a man to drive me across,” said
the general, pausing and moodily mopping his face. “Well, what they
did to us up at Topeka ain't a little. We don't get the county seat! I
don't know who gets it, but we won't—you can everlastingly make
up your mind to that!”
“Is there nothing you can do?”
“Do!” cried the general in a tone of infinite disgust. “I didn't have
a chance to do!”
Stephen looked his dismay.
“A little public spirit and we might have landed it, but I was left to
give my time and spend my money, as if my personal influence
unsupported, was all that was needed! I been made to look foolish!
Well, when they ask me where those county buildings are, I'll tell
'em they can search me!” and the general shrugged his shoulders to
indicate the indifference he did not feel. “It's news to sleep on, ain't
it?” continued Gibbs. “Grant City's given itself county seat airs from
the start, now all we got to do is to lose the railroad, and we might
just as well shut up shop. Well, we ain't lost the railroad.”
“No,” said Stephen clutching at this hope. “The engineers are in
camp here now.”
“It's going to give us a set back,” said the general, who was still
thinking of the county seat. “But I reckon we're strong enough to
stand it—annoying though, ain't it?” and he paused in his tramping
which he had resumed, to stare hard at Stephen. “How do you feel
about it anyhow?” he asked.
“It's a calamity,” said Landray gravely.
“It certainly is,” agreed the general. “We thought we had it sure,
didn't we? Well, there is such a thing as being too sure. I tell you,
Steve, we ain't the only fellows in Kansas with land to sell; there was
plenty of others in the same line, and I went up against 'em hard at
the capital, and they sat down pat on our little scheme. We got to go
slow, things will come steady presently, but until they do we got to
slack off on the building. I am sorry I got you up to tell you nothing
better than this; don't worry though. How's Mrs. Landray?”
“About the same.”
“Humph! What does Arling say? I wonder if you hadn't better get
away from here, Steve?” asked the general with kindly concern.
“And leave you to face the crisis?”
“It won't be the first one I've faced,” and Gibbs expanded his
chest. “Maybe the climate just don't agree with her.”
“No, I sha'n'. leave you; and I doubt if Marian could make the
change now.”
“Well, you know best, Landray, only don't let any sense of honour
hold you here when it's to your interest to leave. I'm willing to split
with you to-morrow if you say the word, and I'll think none the less
of you. I ain't always been above running away myself. Just
remember there's some money to divide now, and there may not be
a red cent six months hence.”
“I'll chance it,” said Stephen firmly.
“Just as you like, but I want you to know that I don't consider you
in any ways obligated to me. You have earned all you have had out
of the business; it wouldn't set well for me to pretend otherwise.”
“Thank you, general,” said the younger man gratefully.
“Well, it's so. You sleep on what I've said.”
Then the general took his leave, and Stephen went back to his
bed, but not to sleep. He had thought himself on the high road to
fortune, and all at once things began to look black. In the morning
he went early to the office where Gibbs soon joined him. He looked
into Stephen's grave face and smiled.
“I guess my call didn't give you pleasant dreams, Landray. I've
thought it over and I guess we'll just keep quiet and see what
happens.”
But before the day was over, it was evident that the information
Gibbs brought from Topeka was not held by him alone. Before
night there was a well-developed stampede among the owners of
lots. This acted on Gibbs rather differently from what might have
been expected.
“We'll show 'em,” he said, “that we got confidence; let 'em get in a
panic, let 'em get rid of their lots, let 'em play the damn fool, Grant
City's here to stay! It's got a future in the very nature of things. We
don't need the county seat; they can locate that wherever they
blame please; but we do want to get rid of these speculators, I
always been opposed to speculation. I'd rather see things go back to
the normal, even if we do lose; I would so, Stephen.”
But hard on the news that Grant City was not to get the county
seat, came the news that neither was it to get the railroad.
“It's a lie! They can't afford to ignore us!” cried Gibbs when he
heard this. “It's a trick to depress real estate! Any one who's fooled
by it deserves to get skinned out of his last dollar.”
But the rumour was verified, and even Gibbs was forced to own
that for the present at least they had lost the railroad.
“You ain't involved, Landray,” he told Stephen. “You stood to share
only in the profits; and I can't lose much, for I hadn't five hundred
dollars when I came here, and I got seventy or eighty lots that are
worth something, and fifteen or twenty houses.”
Here Stephen recalled to his mind the fact that these houses were
in the main unpaid for, and that his creditors might swoop down
upon him at any moment.
“My creditors? Oh, hell—let 'em sweat!” and the general snapped
his fingers almost gaily, and said in the same breath, “I expect
you're cussing me, Steve, for fetching you here.”
“Nothing of the kind, general!” said Stephen stoutly.
“If I fooled you, I fooled myself, Steve; I knew there was some
foxy gentlemen in this part of Kansas, but I thought I was the foxiest
thing on two legs. I have had my ambitions too, Steve, and they
ain't dead yet, a man of my calibre don't give up readily; but I'm
adaptable. I don't sit down to weep over an unpropitious occasion,
in some fashion I dominate it. For the present I suppose there is
nothing for us to do but wait and see how things turn out.”
And through one long hot summer they watched things turn out.
They saw the departure of the town-site speculators; they saw the
settlement of tents and prairie schooners disappear from the waste
beyond the town, until at last there was only the rolling plain, with
its thousands of pegs that marked the lots of the various
subdivisions, or the streets by which they were to be reached; and
with this swift panoramic change came a leaden depression which
ate into the very soul. There was no more poker at the Metropolitan,
once a feature of the nightly gatherings there; the wilderness of
signs creaked idly in the breeze; half the houses stood tenantless;
the stages ceased to rumble in and out of town; the remnant of a
population loafed in heavy lethargetic idleness.
Gibbs and Stephen sat through that long summer, idly for the
most part, in their shirt-sleeves in the shade of their office; the only
business they had was to see Gibbs's creditors, and in the end these
ceased to trouble them.
At first Stephen had been somewhat sustained by Gibbs's
confidence; for Gibbs had been sure things would come right, that
the depression was only temporary; but in the end even his fine
courage failed him. He spent less and less time in the shade of his
office, and more and more time at Mr. Youtsey's bar; but at last Mr.
Yout-sey announced that he was about to leave to seek fortune
elsewhere.
“I've sat up with the corpse,” he told Gibbs jocosely, “and now I'm
going to pull out. I leave the last words to you, for you seem to be
chief mourner; I reckon you'll stay to the finish.”
“I don't know about that,” retorted Gibbs briskly. “You can't keep a
squirrel on the ground; but I'm free to say I don't want to make any
mistakes next time. I'm getting along to a time in life when I don't
expect to make more than two or three more fortunes before I quit.”
“Well, you don't squeal none, general, that's what I admire about
you; win or lose, there's a kind word still coming,” said Mr. Youtsey
admiringly.
In private to Stephen, Gibbs deplored the conditions which he
semed to think were largely responsible for their present situation;
he was seeing deeper than the mere surface of things.
“The war was a mistake, Steve; patriotism and sentiment aside, it
was a big mistake. It's knocked half the country into a cocked hat.
The South is dead, and in my opinion we won't live long enough to
see it come to life. If it had been just left alone, it would have been
mighty interesting to have seen how it would have settled the nigger
question. And what's been the result to the nation at large; we've
lost over half a million of men—young men who'd a pushed out into
the West here, and made this country a wonder. We wouldn't be
waiting for population if they'd lived. But it wasn't to be! The
country's filling up with all sorts of riff-raff from Europe, a class that
never used to come; not the good English and Scotch and Irish, who
came because they wanted elbow room; but greasy serfs, who ain't
caring a damn for anything but wages. I tell you in twenty years an
American will be a curiosity. And to think the way they were wasted,
just thrown at each other by the thousands, in the greatest and
crudest war of modern times.”
But Stephen could not sit there gossiping with Gibbs while Grant
City sank into the prairie, or its shabbily built houses collapsed about
his ears; he must do something, yet what was he to do? Gibbs came
to his rescue with a suggestion.
“I been thinking of your luck, Steve,” he said one day with kindly
concern. “I can hold up this building quite a while by myself just by
leaning against it; I reckon it will be about a hundred years before
anybody sells a lot again in Grant City, and you can't wait on that, on
the chance that you will be the lucky fellow.”
“If I could go!” cried Stephen, with savage earnestness.
“But you can't! Now look here, the farmers have had a pretty good
season; you can always sell a farmer improved machinery, and I
understand you can secure an agency without any capital to speak
of.”
“Why don't you try something of the sort?” asked Landray.
Gibbs shook his head.
“I'm too old a man, Steve, to knock about the country the way I'd
have to; besides, I'm getting ready to start up in business.”
“Start up in what?” cried Stephen.
“In a small way in the licker business,” said the general with
dignity.
“You mean a saloon?”
The general averted his eyes.
“Well, yes—it will have to be retail—I suppose you might almost
call it a saloon,” he admitted reluctantly.
“We are coming down between us, general,” said Landray with a
scornful laugh.
“Not at all, Steve, not at all. I've never claimed more than that I
was up to the occasion, and Youtsey's quit here; fact is, before he
left I made a dicker for what was left of his stock. His going makes
an opening for a commercial enterprise of this sort—in a small way,
of course.”
“Very small, I should say.”
“No, Steve, you must own there are a few people left; not many,
but a few; and they are not going to be any less thirsty in the future
than they have been in the past. I'm not counting on riches, but
merely enough to tide me over until I see an opening. Maybe the
venture might justify a partnership; if you think it will, I'm ready to
whack up,” concluded Gibbs generously.
“Excuse me, general,” said the young man haughtily, “but we'll
reserve that until the last.” He saw that Gibbs was hurt by his words
and manner, and hastened to add, “I think your first suggestion was
the best, do you know where I should write?”
“There are all sorts of catalogues in the office. I'd write to half a
dozen different firms. I merely suggested this as a temporary
makeshift; you might add insurance and lightning-rods.”
“We'll stop with the farm implements, more than that would only
be funny; and I'm in desperate need.”
“I really shouldn't wonder if you didn't do quite well, Landray, after
you get started,” Gibbs said, with ready faith in this new enterprise.
“The farmers farm the land, you'll farm the farmers;” and Stephen's
letters not appearing sufficiently hopeful in tone to his critical mind,
he ended by drafting them for him. “Now I reckon they'll get you
consideration,” said he, when their labours came to an end. “What!
You ain't got any stamps? Why didn't you tell me you were that hard
put? I ain't got much myself, but I always put down a little nest-egg.
I hoped you'd done the same—why didn't you tell me?”
He brought from an inner pocket an old leathern wallet; it was
limp enough, but it contained two twenty dollar bills, one of which
he forced Stephen to accept.
“I never give up quite all; and if I'd known about the county seat a
week earlier, there'd a been a few more gone broke here, but it
wouldn't been you or me, Steve?”
A week later Stephen was present at the formal opening of the
Golden West Saloon in their former office, Gibbs presiding with
typical versatility.
There still lingered in Grant City those who either could not get
away or were too indifferent to try, whose imagination had utterly
failed them; and occasionally, though rarely, the wagons of
emigrants paused for repairs at the blacksmith's; thus there was still
the semblance of life, though half the houses in the town stood
vacant; these seemed to fade away, to disappear in the rank grass
that had come back to flourish in the small trampled lots, and in the
midst of this universal decay, this final phase of the small tragedy of
settlement, Stephen waited and organized the business Gibbs had
suggested, but with no large measure of faith in it.
If Marian's condition would only improve sufficiently so that they
might quit Grant City, he was almost certain that by some
tremendous effort the money he would then require would be
forthcoming.
However when he finally got to work the effect on him was that of
a tonic; and through that fall, and the ensuing winter, over
desperately bad roads, he travelled far and wide; and by spring
when he began to reap the benefits of his winter's work, he saw that
he was not only clear of debt but that he had actually made some
money.
And while he toiled for her, Marian was left alone with a slatternly
unattached female, who was both housekeeper and nurse; and for
medical attendant there was Dr. Arling, who found Grant City entirely
congenial. Drunk and sodden, he was not unskillful; and when he
was needed he could always be found at the Golden West Saloon,
where he loafed tirelessly, and where he played countless games of
checkers with Gibbs.
It never occurred to Stephen that Marian might not recover. To
him it seemed only a question of time until her strength would
return; but he was the only one who was really ignorant of her
condition. Arling had no doubts as to what the end would be, nor
had Mrs. Bassett, the sick woman's attendant.
The end, when it came, came when he was least prepared to
meet it. It was fall again, and he had driven in from the country to
be with Marian over Sunday. He had stabled his horse, and had
come up from the barn, lantern in hand, cold and stiff from his drive
home from a farm far out on the plains.
Mrs. Bassett was in the little kitchen fussing over his supper when
he entered the house by the back door. She had seen his light in the
stable.
“I'll have your supper ready for you in just a minute, Mr. Landray,”
she said. “You've got time to run up and see Mis' Landray. She
knows you're back, I called up and told her you'd come.”
“How is she?” asked Stephen.
“Well, she seems to be about the same, I don't see no difference,”
answered Mrs. Bassett guardedly. She turned and followed him with
her eyes as he went through the sitting-room and entered the
narrow front hall from which the stairs led to the floor above.
He was gone but a moment, then he came quickly into the
kitchen, his face very white.
“She is worse!” he said in a husky whisper. “Why didn't you tell me
so?” but he did not wait for an answer. “Where is the child?” he
demanded.
“I carried him across the back lots to Mis' Gibbs a spell ago. I
couldn't tend him and her, too, he was real fretful.”
“I must go for the doctor,” said Stephen.
“You needn't, Mr. Landray, Mis' Gibbs said she'd go to the saloon
for him; I seen her lantern just a moment ago. You'd best have
something to eat,” she urged.
He turned away impatiently.
“I'm not hungry. Have the doctor come up-stairs as soon as he
gets here.”
But when Arling arrived a few moments later, accompanied by
Gibbs, and joined Stephen in the chamber above where he sat
holding his wife's hand in both his own, he merely shook his head. It
was as he had expected, only the end had been longer deferred
than he had thought possible. He stole from the room and rejoined
Gibbs in the kitchen.
“You tell him, Gibbs—I can't,” he said.
Gibbs rubbed his straggling unkempt beard with a tremulous
hand.
“Maybe he don't need to be told,” he suggested. “But if you think
he should be, I'll do it;” and he stood erect, with something of his
old air.
He mounted the creaking stairs. Stephen must have heard him
coming, for he opened the door and stepped out into the narrow
hall, that was barely large enough to hold the two men and the
small stand, where Mrs. Bassett had placed a smoky lamp with a
dirty chimney.
“Where's the doctor—why don't he come back?” Landray
demanded in a fierce whisper. “Is the drunken fool going to do
nothing?”
“Steve,” began the general, with white shaking lips, “Steve, bear
up, Arling says there ain't anything he can do.”
Stephen looked at him, scarce comprehending what it was he
said.
“He don't know what he's talking about—the fool's drunk!” he said
roughly.
“I reckon he is,” lamented Gibbs weakly. “I'd'a had him under the
pump if there'd been time, but my Julia said for him to hurry, and I
closed up and brought him along just as he was, he wasn't fit to
come by himself.”
“Send him up here again,” said Stephen with stern insistence.
“There is something he can do—my God—” and he broke off
abruptly and re-entered his wife's room.
“Come!” said Gibbs, when he had returned to the kitchen. “Come,
stir around!” he ordered, laying a hand on Arling's shoulder. “Come,
there's something you can do, and you've got to do it.” He was in a
panic of haste. He snatched up his dingy medicine case and thrust it
into Arling's shaking hands. “He's waiting for you, go up and do what
you can.”
“Haven't you told him, Gibbs? She is dying, all he's got to do is to
look at her to see that.”
“You're the one to tell him then—poor Steve—you go to him. I
must go fetch my Julia and the baby;” and he stumbled out into the
darkness with neither hat nor lantern, and fled across the back lots
toward the light that burned in his own window.
He soon returned with Julia, who went at once to the room above.
In the narrow hall she encountered Arling, who had just come from
Marian's bedside, where he had administered to her some simple
restorative. She brushed past him without a word.
“Thank you for coming, it's good to have a woman about,”
murmured Stephen, glancing toward the door as she entered.
Marian lay on the bed without speech or movement, but her eyes,
now brilliant and filled with a strange light, followed every
movement of the two. Julia, with Stephen's help, made her more
comfortable; they smoothed her pillows and raised her higher on
them, for Mrs. Gibbs had been quick to see that her breath came
with difficulty.
She had never liked Marian, and Marian had never liked her—but
she had forgotten all this—which, after all, was only that chance
which determines who shall love and who shall hate. Now she was
all tenderness, this brisk energetic woman, with the lines of a
shrewish temper already stamped upon her face; and her glance
always softened when she looked at Stephen.
There was little either could do but wait; and Marian, save for the
look in her eyes and their restless turning, gave no sign that she
knew what was passing about her.
Presently Julia stole down-stairs to the kitchen. She found Mrs.
Bassett, the general, and Arling still there; the boy fast asleep in her
husband's arms.
“Law!” she cried. “Haven't any of you had sense enough to put
that child to bed?” and she whisked him out of Gibbs's arms and
carried him into the adjoining room.
After that the four fell to watching the clock as if the slow moving
hands would tell them when all was over; and as they watched, the
row of ragged lights in the uncurtained windows that looked out
upon Grant City's Main Street, disappeared one by one, and it was
midnight and very still.
At last Julia rose from her chair and without a word went up-
stairs; she seemed to know that all was over. She noiselessly pushed
open the door and entered the room.
Stephen's face was buried in the pillow beside his wife's. A glance
told her what had happened during her absence from the room. She
stepped to the bedside and placed her hand gently on the man's
shoulder; she felt him shrink from the sudden touch.
“Come,” she said kindly. “You mustn't stay here any longer, Mr.
Landray. You go on down-stairs and ask Mrs. Bassett to come up
here to me.”
“Is she—” he gasped chokingly.
He had risen to his feet, and she urged him away from the
bedside with gentle but determined force.
“You must go down-stairs, Mr. Landray,” she insisted.
“She never spoke—never once,” he cried, turning his bloodshot
eyes on her.
“But she knew you; do go down-stairs, Mr. Landray, indeed, you
mustn't stay here any longer.”
She had pushed him from the room as she spoke, and he crossed
the hall and went slowly and heavily down the narrow steps.
CHAPTER THIRTY-EIGHT

I
F Virginia had been unable to influence Stephen's life as she
wished, this was far from being the case with Jane and Harriett,
who had wholly abandoned themselves to her care and control,
which had to do, unselfishly enough, with their comfort and
convenience. They were also indebted to her for their mental
outlook. They echoed her opinions and acquired her convictions, by
which they endured with unshaken pertinacity, and she had
furnished them with such prejudices as had found a home in their
gentle unworldly hearts. In sentiment they were quite as much
Landray as she was herself; and their pride in the name was quite
equal to her own pride in it; while their affection for her was, aside
from their affection for each other, quite the deepest emotion in their
simple lives.
Under these conditions Harriett had grown into young
womanhood, a shy pretty girl, who looked out upon the world with
soft, inexperienced eyes. But her father's death, and Stephen
Landray's, and perhaps more than all, Virginia's beauty and silent
devotion to her dead husband, had supplied a background of
romance and mystery of which she was never wholly unconscious.
Of society, as it was understood in Benson, she knew nothing; her
mother had never made any friends in the town, and Virginia's own
circle had narrowed; she went nowhere.
Some day Harriett knew she would teach; this Virginia and Jane
had decided for her. It was their conviction that it was the one thing
a young lady could do without compromising her position, it was
entirely dignified, a polite and unexceptional occupation where one
was so unfortunate as to have her own future to consider; and so to
teach, Harriett was fitting herself, when something happened which
materially changed all her plans.
Mr. Stark, who for many years had been the pioneer banker in
Benson, had long since gone to his reward, and now there were
several banks in the town; chief of these was the County Bank
where the interest on certain loans which Benson had made for
Virginia, with the money he had given her for the land in Belmont
County, was regularly paid. Here Harriett often went for Virginia, and
it was here she first met Mark Norton, whose uncle, Judge Norton,
was interested in the fortunes of the bank; indeed, young Norton
was supposed to be mastering the intricacies of the banking
business under the judge's eye.
He came of an excellent family in the county, and Harriett had
frequently observed the young fellow. She had even noted that after
business hours, the easy hours of banks, he indulged himself in the
pleasure of driving most excellent horses. His father was a rich
farmer, which doubtless had much to do with the soundness of the
son's judgment in the matter of horse-flesh; it also explained why it
was that he was able to keep fast horses, a luxury not within the
reach of the ordinary bank clerk. Harriett had seen all this, as he
frequently drove past the cottage presumably on his way into the
country beyond. It afterward developed that Norton had observed
the slight figure of the girl on the lawn in front of the cottage with
the two elder ladies, and he had noted that she was very pretty—
singularly pretty, he would have said.
But it was Harriett's privilege not only to see him in his hours of
recreation, but also when she went to the bank on some errand for
Virginia. He never ventured on anything that could be termed
conversation, though he occasionally appeared to be anxious to
discover Miss Walsh's opinion on such impersonal subjects as the
weather; or if it was a warm day he obligingly called her attention to
that fact. He always addressed her as Miss Walsh. He had been
almost the first person who found such formality necessary, and that
he did, had provoked her to a new and gratifying emotion.
But on one occasion when she stopped at the bank, he was rather
more disposed to talk than was usual with him, but Miss Walsh was
in some haste to go, once the business that had brough her there
was transacted; indeed, so great was her haste that she did not
observe that she had left her check-book.
Later in the afternoon, as she sat on the lawn with Virginia and
her mother, Norton appeared, striding briskly up the street. He
opened the gate, and crossed the lawn to them, smiling and at ease.
“I didn't give Miss Walsh her check-book,” he said. He addressed
himself to Virginia. “I thought she would find it out and come back—
but you didn't”—he turned to Harriett as he spoke—“and so I've
brought it.”
They were all very grateful to him; that is, Virginia and Jane
expressed their gratitude. They thought he had been most kind and
had put himself to a great deal of trouble in a really unimportant
matter. Harriett said nothing, but she suffered an accusing pang
when she recalled that she had shown no interest in the weather.
Virginia asked him to be seated, for though he had given the book
into her keeping, he still stood before them hat in hand.
Norton sat down with alacrity.
He was not under ordinary circumstances a garrulous young
fellow, but on the present occasion he talked hard and fast, as one
will who is trying to gain time; but the burden of what he had to say
was directed to Virginia. Instinct warned him that it would be her
opinion that would have weight with the others, that if he was ever
to return there, as he hoped he might be permitted to do, it would
be because she was willing he should come; and though Virginia
regarded him a little critically at first perhaps, there was nothing of
unkindness in her glance.
At last he quitted his chair; but he was manifestly most reluctant
to go; they rose, too; and the four walked slowly across the lawn.
Norton lagged more and more as they neared the gate; it involved a
positive effort for him to tear himself away. In this extremity he fell
to admiring the flowers; he was particularly fond of flowers, it
seemed; no doubt because he had always lived in the country and
was accustomed to having growing things about; he even ventured
the hope that Mrs. Landray would let him come later when the roses
were in bloom.
“I hope you may come again, Mr. Norton,” responded Virginia
kindly.
“Thank you—if it won't be an intrusion, I'll be only too glad to
come;” and he stole a swift glance at Harriett.
It may have been the merest chance, but after this in one way
and another, Harriett saw a good deal of Norton, for his love of the
country took him past the cottage very often. Harriett knew this
because she read much by the window in the small parlour.
One night as the three sat in the parlour, the girl's quick ear
caught the sound of wheels, and a horse at a rapid trot drew up at
the curb. She hid her face in the book she was reading, and her
heart beat rapidly. There was a brisk step on the path, and a brisk
knock at the door, which Mrs. Walsh opened, and there stood
Norton. To the girl's eyes he seemed wonderfully confident,
wonderfully sure of himself, and later she might have added,
wonderfully discreet, for he devoted himself almost exclusively to
Virginia.
He had, it developed, a lively interest in local history; his own
maternal grandfather having been a contemporary in the county
with General Landray; during the last war with England he had
served as a captain in the company of riflemen which the former had
raised; furthermore his father had known Mrs. Landray's husband, a
fact of which Mrs. Landray herself was well aware. These were all
points that were calculated to make her feel a certain liking for the
young fellow himself, which was only intensified by his quite evident
respect for the very name of Landray; nothing could have been more
commendable or better calculated to show him a person of proper
instincts. Virginia recalled that as a girl she had been a guest at his
mother's wedding; and that General Harrison—to whom Mrs. Norton
was distantly related—had been present.
But while they talked of these matters, his glance drifted on past
Virginia to the pretty silent girl.
When at last Norton took his leave, he was hospitably urged to
call again, an invitation he professed himself as fully determined to
make the most of.
He was the first young man who had ever called there, though
this was not because there was any dearth of young men in the
town; but Harriett was aware that Virginia's point of view regarding
strangers was conservative to say the least; here, however, was a
young man whose grandfather had been a prominent man in the
county when a Landray had been the prominent man of all that
region.
So Norton was welcomed graciously whenever he chose to call.
Yet somehow after that first call they avoided all mention of him;
even repeated visits did not provoke them to discussion; and at this,
Harriett wondered not a little.
At last Virginia astonished her small household by announcing that
she had invited Norton to tea. They dined in the middle of the day,
but on this particular evening tea became a very elaborate affair
indeed, for it was the first time in twenty years, or since Stephen
Landray's death, that Virginia had bidden a guest to her home. Even
Harriett, who thought she knew all the resources of the household,
was astonished at the old silver and glass and china that was
brought out for the occasion; nor had she ever before seen Virginia
dressed with such richness; and she did not wonder that Norton
whispered to her as Virginia quitted the room on some errand:
“What a beautiful woman Mrs. Landray is, I wonder she never
married again.”
“She will never marry, she is devoted to her husband,” said
Harriett.
“Odd, isn't it, that one should always be thinking of that?” he said.
“You mean her devotion to his memory?”
“No, not that—I mean that one should always wonder why a
pretty woman doesn't marry.”
“But when you have lost some one you love.”
“Of course—I suppose love only seems so important in our own
lives because we know what it has meant to some, and so hope it
may mean the same to us.”
“Does it seem so important?” she asked, the colour coming into
her cheeks.
“Doesn't it?” he asked quietly.
“Really I don't know, I had never thought of it—in that way.”
“It's a part of what we call success in life; it may be the better
part—it should be, Harriett.” His voice dwelt lingeringly and
caressingly on her name.
She gave him a frightened, embarrassed glance. It was the first
time he had ever called her anything else than Miss Walsh. She
hoped all at once that her mother or Virginia would come into the
room; but she knew they were busy elsewhere and would not
appear until tea was served.
“Don't you think that?” he asked.
“I don't know. I have never thought of it,” she said faintly.
“I wish you'd think of it now,” he insisted.
“Why?” she faltered.
“Can't you guess?” he asked. “As something that might affect you,
as something that might affect—us.” He leaned forward in his chair
until his face was very close to hers. “Don't you understand what I
mean, Harriett?” he went on, and his voice had become suddenly
tender. “I wonder if you could think it worth while to care for a
fellow like me; don't you know why I've been coming here?”
“To see my Aunt Virginia,” she faltered.
“Well, no—hardly, Harriett; but I fear you are not quite honest
with me. You know that you have brought me here. I wanted to
come long enough before I did; but there seemed no way. What are
you going to tell me about caring for a fellow like me; caring in a
particular way—I mean?”
The colour came and went in the girl's face.
“Of course I can wait—after all you don't know me so very well
yet; but I'd like to think that my case is not entirely hopeless. Won't
you tell me what I want to know, Harriett,”—he heard the swish of
heavy silks in the hall, it was Virginia returning. “I'm going to come
to-morrow for your answer,” he said quickly.
The next day, after the young man had taken his leave of her,
Harriett fled up-stairs to her mother's room with a burning face;
while Norton drove away from the house apparently in the best of
spirits, for he had the unmistakable air of a man who has just heard
something that was unqualifiedly pleasant to hear. The girl hesitated
nervously.
“Mr. Norton has just left,” she said.
“I thought I heard him, or some one, drive up to the gate.”
“It was he,” said Harriett.
Her mother went placidly on with her sewing.
“He wants me to tell you that he wishes to come and see you very
soon, mama,” said Harriett at last, with a little gasp.
“Wants to see me, dear?” in mild surprise.
“Yes.”
“But what about?”
“About—about me—he wants to tell you something.”
“He seems to have told you already,” said Mrs. Walsh.
The girl dropped on her knees before her mother, burying her face
in her lap. There was a little silence between them, and then Mrs.
Walsh said.
“We must tell Virginia. I hope, dear, that she will approve.”
Harriett glanced up quickly at this. She was very white of face.
“Oh, you don't think she won't—you don't think that?”
“Then you want her to approve?”
And Harriett nodded; a single little emphatic inclination of the
head.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy