Although investing in environmental, social, and governance (ESG)-driven portfolios is already a large and growing portion of global assets under management, applications of quantitative techniques to improve and standardize ESG scoring and the construction of ESG portfolios are underutilized. In this article, the authors propose an approach to automatically convert unstructured text data into ESG scores by using the latest advances in deep learning for natural language processing (NLP). They also show how a state-of-the-art NLP technique, BERT, can be incorporated to improve the accuracy of assessing relevance and content of documents in an ESG context using social media data as an example and discuss the relevance of this approach to automating ESG scoring and constructing ESG portfolios. TOPICS:ESG investing, big data/machine learning, portfolio construction Key Findings ▪ The authors demonstrate the feasibility and advantages to applying state-of-the art natural language processing (NLP) to identify environmental, social, governance (ESG) risks using social media data. ▪ The authors discuss how advances in modern NLP can be leveraged to continuously build up algorithmic capabilities for processing ESG-relevant documents, by leveraging the capabilities of deep-learning models for learning general representations of text data, which can then be applied across many tasks in the ESG domain. ▪ The authors discuss results of NLP models can be used for creating aggregated ESG scores, as well design considerations for creating fully or semi-autonomous ESG scoring systems.

The authors propose an approach that combines modern machine learning techniques in natural language processing with portfolio optimization to incorporate views of companies’ environment, social, and governance (ESG) performance. This is automatically done through curating and subsequently converting large-scale news data into portfolio management decisions. They train a machine learning news data classifier to automatically identify several key ESG issues in news data over time. They then aggregate these issues over time to generate a views vector under the Black–Litterman portfolio framework and finally compare the performance of an ESG-tilted portfolio against a standard Black–Litterman portfolio. They also show how this can be achieved at scale, in a fully automated manner, and with consistency over large periods of time. Their methodology thus demonstrates a reasonable and agile method for asset managers to incorporate ESG considerations into their portfolios free of any exclusionary frameworks and without sacrificing performance. TOPICS:ESG investing, quantitative methods, statistical methods, big data/machine learning, portfolio construction Key Findings ▪ The authors describe a theoretical framework for using an automated NLP system to incorporate ESG criteria into portfolio optimization decisions. ▪ The authors demonstrate the technical implementation details for incorporating ESG signals into augmented portfolio weights. ▪ The authors demonstrate the competitive performance of such a portfolio through a long-term historical backtest with the S&P 500 Index.

The complexity of financial data, characterized by its variability and low signal-to-noise ratio, necessitates advanced methods in quantitative investment that prioritize both performance and interpretability.Transitioning from early manual extraction to genetic programming, the most advanced approach in the alpha factor mining domain currently employs reinforcement learning to mine a set of combination factors with fixed weights. However, the performance of resultant alpha factors exhibits inconsistency, and the inflexibility of fixed factor weights proves insufficient in adapting to the dynamic nature of financial markets. To address this issue, this paper proposes a two-stage formulaic alpha generating framework AlphaForge, for alpha factor mining and factor combination. This framework employs a generative-predictive neural network to generate factors, leveraging the robust spatial exploration capabilities inherent in deep learning while concurrently preserving diversity. The combination model within the framework incorporates the temporal performance of factors for selection and dynamically adjusts the weights assigned to each component alpha factor. Experiments conducted on real-world datasets demonstrate that our proposed model outperforms contemporary benchmarks in formulaic alpha factor mining. Furthermore, our model exhibits a notable enhancement in portfolio returns within the realm of quantitative investment and real money investment.

In this paper, we propose a method to price collateralized debt obligations (CDO) within Merton's structural model on underlyings with a stochastic mean-reverting covariance dependence. There are two key elements in our development, first we reduce dimensionality and complexity using principal component analysis on the assets' covariance matrix. Second, we approximate this continuous multidimensional structure using a tree method. Trinomial-tree models can be developed for both the principal components and the eigenvalues assuming the eigenvectors are constant over time and the eigenvalues are stochastic. Our method allows us to compute the joint default probabilities for k defaults of stochastically correlated underlyings and the value of CDOs in a fast manner, without having lost much accuracy. Furthermore we provide a method based on moments to estimate the parameters of the model.Stochastic covariance matrix, CDO, Trinomial-trees, Principal component analysis, Method of moments,