Editor Stephen Satchell brings us a book that truly lives up to its title: optimizing optimization by taking the lessons learned about the failures of portfolio optimization from the credit crisis and collecting them into one book, providing a variety of perspectives from the leaders in both industry and academia on how to solve these problems both in theory and in practice. Industry leaders are invited to present chapters that explain how their new breed of optimization software addresses the faults of previous versions. Software vendors present their best of breed optimization software, demonstrating how it addresses the faults of the credit crisis. Cutting-edge academic articles complement the commercial applications to provide a well-rounded insight into the current landscape of portfolio optimization. Optimization is the holy grail of portfolio management, creating a portfolio in which return is highest in light of the risk the client is willing to take. Portfolio optimization has been done by computer modeling for over a decade, and several leading software companies make a great deal of money by selling optimizers to investment houses and hedge funds. Hedge funds in particular were enamored of heavily computational optimizing software, and many have been burned when this software did not perform as, er, expected during the market meltdown. The software providers are currently reworking their software to address any shortcomings that became apparent during the meltdown, and are eager for a forum to address their market and have the space to describe in detail how their new breed of software can manage not only the meltdown problems but also perform faster and better than ever before-that is, optimizing the optimizers!! In addition, there is a strong line of serious well respected research on portfolio optimization coming from the academic side of the finance world. Many different academic approaches have appeared toward optimization: some favor stochastic methods, others numerical methods, others heuristic methods. All focus on the same issues of optimizing performance at risk levels. This book will provide the forum that the software vendors are looking for to showcase their new breed of software. It will also provide a forum for the academics to showcase their latest research. It will be a must-read book for portfolio managers who need to know whether their current optimization software provider is up to snuff compared to the competition, whether they need to move to a competitor product, whether they need to be more aware of the cutting-edge academic research as well.
- Presents a unique ''confrontation'' between software engineers and academics
- Highlights a global view of common optimization issues
- Emphasizes the research and market challenges of optimization software while avoiding sales pitches
- Accentuates real applications, not laboratory results
Author(s): Stephen Satchell
Series: Quantitative Finance
Edition: 1
Publisher: Academic Press
Year: 2009
Language: English
Pages: 323
Optimizing Optimization: The Next Generation of Optimization Applications and Theory......Page 4
Copyright Page......Page 5
Contents......Page 6
List of Contributors......Page 12
Section One: Practitioners and Products......Page 18
1.1 Introduction......Page 20
1.2 Alpha uncertainty......Page 21
1.3 Constraints on systematic and specific risk......Page 23
1.4 Constraints on risk using more than one model......Page 29
1.5 Combining different risk measures......Page 33
1.6 Fund of funds......Page 35
References......Page 39
2.1 Introduction......Page 40
2.2 Portfolio construction using multiple risk models......Page 42
2.2.1 Out-of-sample results......Page 50
2.2.2 Discussion and conclusions......Page 51
2.3 Multisolution generation......Page 52
2.3.1 Constraint elasticity......Page 56
2.3.2 Intractable metrics......Page 58
2.4 Conclusions......Page 68
References......Page 69
3.1 Introduction......Page 70
3.1.4 BITA Optimizer(™)......Page 71
3.2.4 Long–short portfolio construction......Page 72
3.3.1 A technical overview......Page 73
3.3.2 The BITA optimizer—functional summary......Page 74
3.4.2 Introduction......Page 75
3.4.3 Reformulation of mean–variance optimization......Page 76
3.4.5 FE constraints......Page 78
3.4.6 Preliminary results......Page 79
3.4.8 Explicit risk budgeting......Page 82
3.5.1 Introduction......Page 83
3.5.2 Omega and GLO......Page 84
3.5.3 Choice of inputs......Page 85
3.5.4 Analysis and comparison......Page 86
3.5.7 Down-trimming of emerging market returns......Page 87
3.5.8 Squared losses......Page 88
3.5.9 Conclusions......Page 89
3.6.1 Introduction......Page 90
3.6.2 Discussion......Page 91
3.6.3 The model......Page 92
3.6.4 Incorporation of alpha and risk model information......Page 93
3.7.2 Why endowments matter......Page 95
3.7.3 Managing endowments......Page 96
3.7.4 The specification......Page 97
3.7.5 Trustees' attitude to risk......Page 99
3.7.6 Decision making under uncertainty......Page 100
3.7.7 Practical implications of risk aversion......Page 101
3.8.2 Request: how to optimize in the absence of forecast returns......Page 103
3.9 Conclusions......Page 104
Appendix A: BITA Robust optimization......Page 105
Appendix B: BITA GLO......Page 106
References......Page 107
4.1 Introduction......Page 110
4.2.2 The WPA solution......Page 111
4.3.2 The WPA solution......Page 114
4.4.2 The WPA solution......Page 118
4.5.2 The WPA solution......Page 121
4.5.3 Summary......Page 124
Appendix—WPA features......Page 128
References......Page 130
Section Two: Theory......Page 132
5.1 Introduction......Page 134
5.2 Empirical evidence from the Dow Jones Industrial Average components......Page 136
5.3.1 The portfolio dimensionality problem......Page 138
5.3.2 Generation of return scenarios......Page 143
5.4 The portfolio selection problem......Page 147
5.4.1 Review of performance ratios......Page 149
5.4.2 An empirical comparison among portfolio strategies......Page 151
5.5 Concluding remarks......Page 153
References......Page 157
6.1 Introduction......Page 160
6.2.1 Definition and properties......Page 162
6.2.2 Modeling EVaR dynamically......Page 164
6.3 The asset allocation problem......Page 167
6.4 Empirical illustration......Page 170
6.5 Conclusion......Page 175
References......Page 176
7.1 Introduction......Page 178
7.2.1 Basic assumptions......Page 179
7.2.2 Optimize or measure performance......Page 182
7.3 Part 2: The DTR optimizer......Page 184
Appendix: Formal definitions and procedures......Page 188
References......Page 194
8.1 Introduction......Page 196
8.2 Main proposition......Page 197
8.3 The case of two assets......Page 201
8.4 Conic results......Page 207
8.5 Simulation methodology......Page 211
References......Page 215
9.1 Introduction......Page 218
9.2.1 Risk and reward......Page 221
9.2.2 The problem summarized......Page 226
9.3.1 The algorithm......Page 227
9.3.2 Implementation......Page 228
9.4 Stochastics......Page 232
9.5.2 Arbitrage opportunities......Page 235
9.5.4 The neighborhood and the thresholds......Page 236
9.6 Conclusion......Page 237
References......Page 238
10.1 Section 1......Page 242
10.2 Section 2......Page 243
10.3 Remark 1......Page 246
10.4 Section 3: Finite sample properties of estimators of alpha and tracking error......Page 247
10.5 Remark 2......Page 252
10.7 Section 4......Page 253
10.8 Section 5: General linear restrictions......Page 255
10.9 Section 6......Page 258
Acknowledgment......Page 261
References......Page 262
11.1 Introduction......Page 264
11.2 A brief history of portfolio optimization......Page 265
11.3.1 Basic properties......Page 268
11.3.2 Density estimation......Page 271
11.3.3 Simulating Johnson random variates......Page 273
11.4.1 The maximization problem......Page 274
11.4.2 The threshold acceptance algorithm......Page 277
11.5 Data reweighting......Page 278
11.6 Alpha information......Page 279
11.7 Empirical application......Page 282
11.7.1 The decay factor, ρ......Page 283
11.7.3 The importance of non-Gaussianity......Page 285
11.8 Conclusion......Page 288
11.9 Appendix......Page 289
References......Page 295
12.1 Introduction: Risk measures and their axiomatic foundations......Page 300
12.2 A simple algorithm for CVaR optimization......Page 302
12.3.2 How much momentum investing is in a downside risk measure?......Page 305
12.3.3 Will downside risk measures lead to "under-diversification"?......Page 307
12.4.1 Estimation error......Page 309
12.4.2 Approximation error......Page 310
12.5 Scenario generation II: Conditional versus unconditional risk measures......Page 312
12.6 Axiomatic difficulties: Who has CVaR preferences anyway?......Page 313
References......Page 315
C......Page 318
G......Page 319
M......Page 320
P......Page 321
S......Page 322
Z......Page 323
