EMNLP 2025 in Suzhou

Serious, Sassy, or Sad? Teaching Machines to Read the Room (From Speech Embeddings)

Disentangling Speech Embeddings: Removing Text Content from Audio Embeddings with Regression

Personal Reflections on NeurIPS 2024

What’s in a Name? AI Meets the Sociology of Naming

[paper] Assisting in Writing Wikipedia-like Articles From Scratch with Large Language Models

Bayesian Networks for Business: Modeling Profit and Loss of a Cafe in Hong Kong

Preparing for ICML 2024: Main themes

Basic DSPy RAG tutorial on DataGrapple blog posts

Basic DSPy RAG tutorial on DataGrapple blog posts

Prompting is Programming with LMQL

Selected ML Papers from ICML 2023

[Active Reading with ChatGPT] Systematic Investing in Credit

[Active Reading with ChatGPT] Quantitative Portfolio Management: The Art and Science of Statistical Arbitrage

Building a S&P 500 company classification from Wikipedia articles (guided by ChatGPT)

[Book] Volatility Trading

SetFit: Fine-tuning a LLM in 10 lines of code and little labeled data

Performance attribution of a crypto market-neutral book on a statistical risk model

[Book] Interview with ChatGPT about its book 'From Data to Trade: A Machine Learning Approach to Quantitative Trading'

AWS S3 to Lambda to Aurora to Lambda to Binance | Serverless architecture for crypto trading

Takeaways from Complex Networks 2022 in Palermo, Italy

Getting ready for Complex Networks 2022 in Palermo, Italy

Hierarchical PCA x Hierarchical clustering on crypto perpetual futures

Crypto PCA First Eigenvector

Bayesian net and Boparan 7.625% 30 Nov 2025 Prospectus

Speaker identification in 'Marine Le Pen vs. Emmanuel Macron 2017 French Presidential Debate'

Standard readability measures (applied on Shakespeare's plays)

Naive modelling of Matalan defaulting on its MTNLN 9.5 01/31/24 Notes

Naive modelling of credit defaults using a Markov Random Field

Back to basics: PCA on stocks returns

[Book] Advanced Portfolio Management -- A Quant's Guide for Fundamental Investors

[paper] Top2Vec: Distributed Representations of Topics

with application on 2020 10-K business descriptions

[HKML] Hong Kong Machine Learning Meetup Season 4 Episode 2

[HKML] Hong Kong Machine Learning Meetup Season 4 Episode 2

Embeddings of Sectors and Industries using Graph Neural Networks

node2vec embeddings of industries projected onto the 2d plane

Mind the Jensen Gap!

The Swelling Effect: Think twice before averaging covariance matrices

A few ellipsoids representing the associated covariance matrices along the geodesic path from the leftmost to the rightmost matrices.

Conditional CorrGAN: An example in Google Colab

A few cCorrGAN-generated correlation matrices, and the confusion matrix of a SPDNet + RBN classification.

Classification of Correlation Matrices using SPDNet with Riemannian Batch Normalization

Illustration from "A Riemannian Network for SPD Matrix Learning" https://arxiv.org/pdf/1608.04233.pdf

[Paper] Summary of 'Interpretable Machine Learning – A Brief History, State-of-the-Art and Challenges'

[Paper] Summary of Interpretable Machine Learning – A Brief History, State-of-the-Art and Challenges

AutoGL and the Open Graph Benchmark: Datasets for Machine Learning on Graphs

Some further assessment of the original CorrGAN model (2019)

[Paper] Summary of 'Explaining by Removing: A Unified Framework for Model Explanation'

Illustration from Explaining by Removing: A Unified Framework for Model Explanation

From Fundamental To Quantamental Investing

[HKML] Hong Kong Machine Learning Meetup Season 3 Episode 4

[HKML] Hong Kong Machine Learning Meetup Season 3 Episode 4

[HKML] Hong Kong Machine Learning Meetup Season 3 Episode 3

[HKML] Hong Kong Machine Learning Meetup Season 3 Episode 3

[HKML] Hong Kong Machine Learning Meetup Season 3 Episode 2

[HKML] Hong Kong Machine Learning Meetup Season 3 Episode 2

[HKML] Hong Kong Machine Learning Meetup Season 3 Episode 1

[HKML] Hong Kong Machine Learning Meetup Season 3 Episode 1

Some Thoughts on the Applications of Deep Generative Models in Finance

Can we predict a market regime from correlation matrix features?

[Book] Commented summary of Probabilistic Graphical Models -- A New Way of Thinking in Financial Modelling

Which portfolio allocation method to choose? Look at the correlation matrix!

Portfolio construction methods and risk metrics: in- and out-of-sample comparisons on simulated data

Extraction of features from a given correlation matrix

Release of a few pretrained CorrGAN models

Release of a few pretrained CorrGAN models

How to combine a handful of predictors using empirical copulas and maximum likelihood

CopulaGAN (the 3d case) - A first naive attempt

Clustering Marginal Distributions of Stocks Returns, and Sampling from their Wasserstein Barycenter

Sampling from Empirical Copulas of Stocks

Wasserstein Barycenters of Stocks Empirical Copulas

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 8

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 8

[Paper + Implementation] Hierarchical PCA and Applications to Portfolio Management

Mutual Information Is Copula Entropy

Measuring non-linear dependence with Optimal Transport

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 7

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 7

[Book] Commented summary of Machine Learning for Factor Investing by Guillaume Coqueret and Tony Guida

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 6

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 6

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 5

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 5

[Book] Commented summary of Machine Learning for Asset Managers by Marcos Lopez de Prado

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 4

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 4

[Paper + Implementation] The Hierarchical Equal Risk Contribution Portfolio (Part I)

[Paper + Implementation] The Hierarchical Equal Risk Contribution Portfolio (Part I)

[Paper + Experimentation] CartoonGAN applied to Hong Kong landscapes using Streamlit

CorrVAE: A VAE for sampling realistic financial correlation matrices (Tentative II)

CorrVAE: A VAE for sampling realistic financial correlation matrices (Tentative II)

CorrVAE: A VAE for sampling realistic financial correlation matrices (Tentative I)

CorrVAE: A VAE for sampling realistic financial correlation matrices (Tentative I)

S&P 500 Sharpe vs. Correlation Matrices - Building a dataset for generating stressed/rally/normal scenarios

S&P 500 Sharpe vs. Correlation Matrices - Building a dataset for generating stressed/rally/normal scenarios

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 3

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 3

How to define an intrisic Mean of Correlation Matrices in a Riemannian sense?

Comparison of Network-based and Minimum Variance Portfolios Using CorrGAN

Comparison of Network-based and Minimum Variance Portfolios Using CorrGAN

Hierarchical Risk Parity - Implementation & Experiments (Part III)

Hierarchical Risk Parity - Implementation & Experiments (Part III)

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 2

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 2

TF 2.0 DCGAN for 100x100 financial correlation matrices

TF 2.0 DCGAN for 100x100 financial correlation matrices

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 1

[HKML] Hong Kong Machine Learning Meetup Season 2 Episode 1

TF 2.0 GAN MLP for 100x100 financial correlation matrices

TF 2.0 GAN MLP for 100x100 financial correlation matrices

[HKML] Supercharge your Marketing with Data & ML . [HKML <> IAB] . Off-Season #1

[HKML] Supercharge your Marketing with Data & ML . [HKML <> IAB] . Off-Season #1

Permutation invariance in Neural networks

Permutation invariance in Neural networks

Using LIME to 'explain' Snorkel Labeler

Using LIME to “explain” Snorkel Labeler

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 12 (Season Finale)

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 12 (Season Finale)

Stylized Facts of Financial Correlations

Stylized Facts of Financial Correlations

CorrGAN: A GAN for sampling correlation matrices (Part II)

CorrGAN: A GAN for sampling correlation matrices (Part II)

CorrGAN: A GAN for sampling correlation matrices (Part I)

CorrGAN: A GAN for sampling correlation matrices (Part I)

[ICML 2019] Day 5 - Workshop Time Series

[ICML 2019] Day 5 - Workshop Time Series

[ICML 2019] Day 4 - Interpretability, Natural Language Processing, Smarter than AI four year old kids, Unsupervised Learning

[ICML 2019] Day 4 - Interpretability, Natural Language Processing, Smarter than AI four year old kids, Unsupervised Learning

[ICML 2019] Day 3 - Robotics, Good ol' Sparse Coding, misc. applications, Transfer, Multitask and Active Learning

[ICML 2019] Day 3 - Robotics, Good ol’ Sparse Coding, misc. applications, Transfer, Multitask and Active Learning

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 11

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 11

[ICML 2019] Day 2 - U.S. Census, Time Series, Hawkes Processes, Shapley values, Topological Data Analysis, Deep Learning & Logic, Random Matrices, Optimal Transport for Graphs

[ICML 2019] Day 2 - U.S. Census, Time Series, Hawkes Processes, Shapley values, Topological Data Analysis, Deep Learning & Logic, Random Matrices, Optimal Transport for Graphs

[ICML 2019] Day 1 - Tutorials

[ICML 2019] Day 1 - Tutorials

Experimenting with LIME - A tool for model-agnostic explanations of Machine Learning models

Experimenting with LIME - A tool for model-agnostic explanations of Machine Learning models

[ICML 2019] Reading list of accepted papers

[ICML 2019] Reading list of accepted papers

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 10

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 10

May the Fourth: VADER for Credit Sentiment?

May the Fourth: VADER for Credit Sentiment?

Snorkel Credit Sentiment - Part 1

First experiment with Snorkel Metal – Credit Sentiment on DataGrapple blogs

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 9

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 9

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 8

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 8

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 7

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 7

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 6

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 6

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 5

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 5

[Paper] A Backtesting Protocol in the Era of Machine Learning

[Paper] A Backtesting Protocol in the Era of Machine Learning

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 4

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 4

[Book] The UnRules - Man, Machines and the Quest to Master Markets

[Book] The UnRules - Man, Machines and the Quest to Master Markets

Hierarchical Risk Parity - Implementation & Experiments (Part II)

Hierarchical Risk Parity - Implementation & Experiments (Part II)

Network-based vs. Minimum Variance portfolios: Any deep connections?

Network-based vs. Minimum Variance portfolios: Any deep connections?

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 3

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 3

How to sample uniformly over the space of correlation matrices? The onion method

How to sample uniformly over the space of correlation matrices? The onion method

Hierarchical Risk Parity - Implementation & Experiments (Part I)

Hierarchical Risk Parity - Implementation & Experiments (Part I)

[Book] Neural Network Methods for Natural Language Processing

[Book] Neural Network Methods for Natural Language Processing

``Combination of Rankings'' - The Stationary Distribution

``Combination of Rankings’’ - The Stationary Distribution

``Combination of Rankings'' - The Full Coverage Case

``Combination of Rankings’’ - The Full Coverage Case

[Bloomberg Meetup] The Forefront of Technologies in Finance

[Bloomberg Meetup] The Forefront of Technologies in Finance

A Monte Carlo study of the ``Combination of Rankings'' methods

A Monte Carlo study of the ``Combination of Rankings’’ methods

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 2

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 2

Combination of Rankings

Combination of Rankings - A Proper Merging of Experts Views

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 1

[HKML] Hong Kong Machine Learning Meetup Season 1 Episode 1

[ICML 2018] Retrospectives

[ICML 2018] Retrospectives

[ICML 2018] Day 4 - Time-Series Analysis, NLP, and More Human-like Learning Machines

[ICML 2018] Day 4 - Time-Series Analysis, NLP, and More Human-like Learning Machines

[ICML 2018] Day 3 - Energy, GANs, Rankings, Curriculum Learning, and our paper

[ICML 2018] Day 3 - Energy, GANs, Rankings, Curriculum Learning, and our paper

[ICML 2018] Day 2 - Representation Learning, Networks and Relational Learning

[ICML 2018] Day 2 - Representation Learning, Networks and Relational Learning

[ICML 2018] Day 1 - Tutorials

[ICML 2018] Day 1 - Tutorials

On the difficulty of reading numbers in different languages

On the difficulty of reading numbers in different languages

Neural Style Transfer applied to paintings

In this short blog, we apply the fast style transfer as implemented in tensorflow/magenta.

How to compute the Planar Maximally Filtered Graph (PMFG)

How to compute the Planar Maximally Filtered Graph (PMFG)

How to detect false strategies? The Deflated Sharpe Ratio

Deflated Sharpe Ratio

APIs for getting crypto related data

Following my blog post Download & Play with Cryptocurrencies Historical Data in Python, I got several times questions on how to get the historical data. In the previous blog, I used a Python wr...

Physiological analytics for sports

I recently got a Garmin Forerunner 935 (advised by a good friend of mine). After using it for two months, I can say that I’m happy with it so far. It has lots of functionalities and analytics that ...

AQR Academic Factors

AQR has released an implementation of the well-known academic factors in its AQR Data Library:

Quant Blogs

Here a tentative of a list of interesting blogs to keep up with quant best practices to study financial markets. I hope that my readers will help me curate this list by suggesting new blogs, removi...

Tail Dependence Coefficients

Research material:

Riemannian Geometry of Correlation Matrices

Research material:

PhD defense - Some contributions to the clustering of financial time series

Here are the slides. The PhD studies were generously funded by Hellebore Capital.

[Correlation] How to visualize dependence between two variables?

In this blog, we provide a snippet of code to explore the dependence between two variables. We illustrate its use on visualizing the dependence between a few of the main cryptocurrencies: Bitcoin (...

[Clustering] How to sort a distance matrix

Following the Ecole Polytechnique - Data Science Summer School where I got several times questions about how I produced the sorted correlation matrices displayed in my poster, I decided to write ...

[Field report] Data Science Summer School at Ecole Polytechnique (with Bengio, Russell, Bousquet, Archambeau and others)

A small field report with personal viewpoint about the Data Science Summer School (Ecole Polytechnique) Monday, Aug. 28 – Friday, Sept. 1, 2017.

Download & Play with Cryptocurrencies Historical Data in Python

To access the CryptoCompare public API in Python, we can use the following Python wrapper available on GitHub: cryCompare.

Reading list of NLP stuff

General NLP:

Quick correlation study between BTC/USD and ETH/USD

import numpy as np import scipy import pandas as pd import matplotlib.pyplot as plt %matplotlib inline import seaborn as sns import json from datetime import timedelta

Field reports from ICML 2017 in Sydney

My colleague, Mikolaj Binkowski, at Hellebore Capital was at the 34th International Conference on Machine Learning ICML 2017 in Sydney to represent the company and present his work on Deep Learnin...

Study of US Stocks Correlations, Hierarchies and Clusters

In this small study, we use hierarchical clustering techniques to explore the structure of correlations between US stocks. To do so, we first download a dataset of adjusted close prices for the US ...

Ether vs. Bitcoin -- Part 0 bis

For the last few days, ETH has lost 1/3 of its value with -20% several days in a row. We update the initial study with up-to-date data to take into account this recent drawback. We add a comment on...

Ether vs. Bitcoin -- Part 0

In this introduction notebook, we simply displayed the distribution of the returns and see that tails are heavy, meaning that standard quant models cannot be applied to cryptocurrencies either…

June Ethereum London Meetup at Imperial College

I attended this evening the June Ethereum London Meetup at Imperial College (I have been there a couple of times before). Imperial College’s big amphitheater in Sir Alexander Fleming Building was f...

Swap Data Repositories for Credit Default Swaps

What are Swap Data Repositories?

[Clustering] How and Where should you cut a dendrogram?