Generative AI for Trading for Asset Management.pdf
大小:(13.15 MB)

只需: RMB 19元  马上下载

生成式 AI 在交易与资产管理中的应用资料
内容新，2026最新上线！
内容丰富精彩，560页的大型资料包！
Part I: Generative AI for Trading and Asset Management: A No-code Introduction
Chapter 1: No-code Generative AI for Basic
Quantitative Finance
1.1 Retrieving Historical Market Data
1.2 Computing Sharpe Ratio
1.3 Data Formatting and Analysis
1.4 Translating Matlab Codes to Python Codes
1.5 Conclusion
Chapter 2: No-code Generative AI for Trading
Strategies Development
2.1 Creating Codes from a Strategy
Specification
2.2 Summarizing a Trading Strategy Paper and
Creating Backtest Codes from It
2.3 Searching for a Portfolio Optimization
Algorithm Based on Machine Learning
2.4 Explore Options Term Structure Arbitrage
Strategies
2.5 Conclusion
2.6 Exercises
Appendix 2A.1 Computing Next-day’s Return
Appendix 2A.2 Uploading the Fama-French
Factors
Appendix 2A.3 Combining Fama-French Factors
with Next-day’s Returns
Chapter 3: Whirlwind Tour of ML in Asset
Management
3.1 Unsupervised Learning
3.2 Supervised Learning
3.3 Deep Reinforcement Learning
3.4 Data Engineering
3.5 Feature Engineering
3.6 Conclusion
Part II: Deep Generative Models for Trading and Asset
Management
Chapter 4: Understanding Generative AI
4.1 Why Generative Models
4.2 Difference with Discriminative Models
4.3 How Can We Use Them?
4.4 Illustrating Generative Models with
ChatGPT
4.5 Hybrid Modeling: Combining Generative
and Discriminative Models
4.6 Taxonomy of Generative Models
4.7 Conclusion
Chapter 5: Deep Autoregressive Models for
Sequence Modeling
5.1 Representation Complexity
5.2 Representation and Complexity Reduction
5.3 A Short Tour of Key Model Families
5.4 Model Fitting
5.5 Conclusions
Chapter 6: Deep Latent Variable Models
6.1 Introduction
6.2 Latent Variable Models
6.3 Examples of Traditional Latent Variable
Models
6.4 Learning
6.5 Variational Autoencoder (VAE)
6.6 VAEs for Sequential Data and Time Series
6.7 Conclusion
Chapter 7: Flow Models
7.1 Introduction
7.2 Model Training
7.3 Linear Flows
7.4 Designing Nonlinear Flows
7.5 Coupling Flows
7.6 Autoregressive Flows
7.7 Continuous Normalizing Flows
7.8 Modeling Financial Time Series with Flow
Models
7.9 Conclusion
Chapter 8: Generative Adversarial Networks
8.1 Introduction
8.2 Training
8.3 Some Theoretical Insight in GANs
8.4 Why Is GAN Training Hard? Improving GAN
Training Techniques
8.5 Wasserstein GAN (WGAN)
8.6 Extending GANs for Time Series
8.7 Conclusion
Chapter 9: Leveraging LLMs for Sentiment Analysis
in Trading
9.1 Sentiment Analysis in Fed Press Conference
Speeches Using Large Language Models
9.2 Data: Video + Market Prices
9.3 Speech-to-text Conversion
9.4 Sentiment Analysis
9.5 Experiment Results
9.6 Conclusion
Chapter 10: Efficient Inference
10.1 Introduction
10.2 Scaling Large Language Models: High
Performance, High Computational Cost, and
Emergent Abilities
10.3 Making FinBERT Faster
10.4 Model Quantization
10.5 Customizing Your LLM: Adapting Models
to Your Needs
10.6 Conclusions
Chapter 11: Afterword
11.1 Diffusion Models
11.2 Combining Generative Model Variants
11.3 LLMs as Financial Advisors
References
Appendix
A.1 Retrieving Adjusted Closing Prices and
Computing Daily Returns
A.2 Installing Python
A.3 Plotting the Risk-free-rate over the Years
A.4 Computing the Sharpe Ratio of SPY
A.5 Matlab Code for Computing Efficient Frontier
and Finding the Tangency Portfolio
Index
End User License Agreement
List of Illustrations
Chapter 1
Figure 1.1 Efficient frontier based on Python code
generated by ChatGPT.
Chapter 2
Figure 2.1 Cumulative returns of Fama-French
three-factor strategy.
Figure 2.2 Incorrect plot of annualized time value
of put options as function o...
Figure 2.3 Incorrect plot of implied volatility of put
options as function of t...
Figure 2.4 Annualized put option prices as function
of time to expiration based...
Figure 2.5 Annualized call option prices as function
of time to expiration base...
Chapter 3
Figure 3.1 Dendrogram of five stocks based on the
correlations of their daily r...
Figure 3.2 Principal components of two correlated
series.
Figure 3.3 Illustration of a sigmoid function.
Figure 3.4 Illustration of why L1 regularization can
more easily get us zero we...
Figure 3.5 The Anscombe quartet. All four data sets
have same means, variances,...
Figure 3.6 A typical precision-recall curve.
Figure 3.7 A typical ROC (Receiver Operating
Characteristics) curve.
Figure 3.8 A one-hidden-layer MLP that takes a
vector of features to classify...
Figure 3.9 A simple MLP for time-series prediction.
Figure 3.10 Tying the weights to the same values
and connecting the hidden nodes...
Figure 3.11 Same as Figure 3.10, but rotated 90°
counterclockwise so time procee...
Figure 3.12 Daily minimum and maximum
sentiment scores that show structural brea...
Chapter 4
Figure 4.1 Model taxonomy.
Figure 4.2 Conditional probability of the first token
given the prompt .
Figure 4.3 Conditional probability of the second
token given the prompt and p...
Figure 4.4 Conditional probability of the third token
given the prompt and pr...
Figure 4.5 Conditional probability of the fourth
token given the prompt and p...
Figure 4.6 Conditional probability of the fourth
token given the prompt and p...
Chapter 5
Figure 5.1 Model taxonomy: Autoregressive
models.
Figure 5.2 Sample from the Binarized MNIST
dataset. Larochelle and Murray (2011).
Figure 5.3 MADE Generation on MNIST. Left:
samples from a MADE model. Right: Ne...
Figure 5.4 Visualization of a stack of causal
convolutional layers. Figure 2 fr...
Figure 5.5 Visualization of a stack of dilated causal
convolutional layers. Fig...
Figure 5.6 Visualizing attention.
Figure 5.7 Scaled Dot-Product Attention. Figure 2
(left) from Vaswani et al. (2...
Figure 5.8 The Transformer encoder-decoder
architecture, developed for machine ...
Figure 5.9 Multi-head Attention. Figure 2 (right)
from Vaswani et al. (2023).
Figure 5.10 An illustration showing how the current
and lagged values of the ser...
Figure 5.11 An illustration showing how the model
input, comprising the time-ser...
Chapter 6
Figure 6.1 Model taxonomy: variational
autoencoders.
Figure 6.2 Illustration of model selection for
homoscedastic noise.
Figure 6.3 Illustration of model selection for
heteroscedastic noise.
Figure 6.4 Illustration of clustering using Gaussian
Mixture Models (GMMs). (a)...
Figure 6.5 Gaussian Mixture Model for market
regime detection.
Figure 6.6 VAE.
Figure 6.7 Illustration of the learned data manifold
for generative models with...
Figure 6.8 Illustration of the encoder-decoder
architecture of Base TimeVAE. Th...
Figure 6.9 Illustration of the main components in
Interpretable TimeVAE. Specia...
Chapter 7
Figure 7.1 Model taxonomy: flow models.
Figure 7.2 Illustration of the operations involved in
coupling flows, including...
Figure 7.3 Illustration of unbiased samples
generated by the NICE model when tra...
Figure 7.4 Illustration of unbiased samples
generated by the NICE model when tra...
Figure 7.5 Samples generated by the Real-NVP
model across four datasets: CIFAR-10...
Figure 7.6 Illustration of new images generated
through interpolations between f...
Chapter 8
Figure 8.1 Model taxonomy: GANs.
Figure 8.2 Evolution of GAN-generated images over
time, illustrating the progre...
Figure 8.3 The rightmost column displays the
nearest training example to each c...
Figure 8.4 Table 1 from Goodfellow et al. (2014):
Parzen window-based log-likel...
Figure 8.5 Illustration of one reason why training
GANs can be difficult. In th...
Figure 8.6 Illustration of two different cases for the
data distribution and th...
Figure 8.7 Illustration of the optimal discriminator
and critic when distinguis...
Chapter 9
Figure 9.1 Fed Chair Powell discusses latest Fed
rate hike.
Figure 9.2 System block diagram.
Figure 9.3 SPY Price series during Fed press
conference.
Figure 9.4 Enriched price series.
Figure 9.5 Scatter plot of sentiment signal vs
forward returns.
Figure 9.6 Available models and languages.
Figure 9.7 Whisper output on FED data.
Figure 9.8 Figure 3 from Devlin et al. (2019):
Illustration of differences in p...
Figure 9.9 Figure 1 from Devlin et al. (2019):
Illustration of the input/output...
Figure 9.10 Figure 2 from Devlin et al. (2019):
Illustration of the BERT input r...
Figure 9.11 Time-series sentiment signal and
forward returns.
Figure 9.12 Scatter plot of sentiment signal vs
forward returns.
Chapter 10
Figure 10.1 Figure 1 from Kaplan et al. (2020):
Illustration of how language mod...
Figure 10.2 Figure 4 from Wei et al. (2023): An
illustration of how performance ...
Figure 10.3 Figure 2 from Wei et al. (2022):
Illustration of performance, measur...
Figure 10.4 Softmax distribution.
Figure 10.5 Weights distribution.
Figure 10.6 Distribution of quantized weights.
Figure 10.7 Figure 1 from Hu et al. (2021): This
figure illustrates the reparame...
Appendix
Figure A.1 BIL annualized returns.
Figure A.2 Twenty-day moving average of
annualized BIL returns.
Figure A.3 Efficient Frontier produced by Matlab
codes.
List of Tables
Chapter 3
Table 3.1 Results of cluster-based features
importance ranking based on their ...
Table 3.2 Hypothetical design matrix for three
features with four samples.
Table 3.3 The confusion matrix for binary
classification.
Table 3.4 Confusion matrix for three-class
classification.
Chapter 7
Table 7.1 Computational complexity of the inverse
and determinant of the Jacob...
Chapter 9
Table 9.1 Performance table.
Chapter 10
Table 10.1 Performance metrics of teacher vs.
student models.
Table 10.2 Inference speed of teacher vs. student
models.
Table 10.3 Integer range for different bit widths.
Table 10.4 Performance table for speedups because
of quantization.
Table 10.5 Inference speed after quantization.