The tutorials are organized as a Summer Course by the COST Action HiTEc and are chaired by Prof. Erricos Kontoghiorghes and Prof. Ana Colubi on behalf of the Action. The participation of COMPSTAT 2026 attendees is sponsored by IASC. Any other participant must register using the link below.

The Summer Course will consist of three modules plus two keynote talks as described below. Participants are requested to bring their own laptop.

Dates: 25-27 August 2026
Venue: Athens University Law School building, Solonos 57, Athens 106 79, Greece.
Room: TBA.

Trainers:
Module I: Weining Wang, University of Bristol, UK.
Module II: Michael Pitt, King's College London, UK.
Module III: Matus Maciak, Charles University in Prague, Czech Republic.

Keynote talks:
Peter Buehlmann, ETHZ, Switzerland.
Martina Mittlboeck, Medical University of Vienna, Austria.

• In order to apply for the grants, candidates should submit their CV by e-mail to hiteccostaction@gmail.com.
• Deadline for applications: 15th March 2026.
• Granted candidates will be informed by e-mail after the deadline and must send their flight tickets and registration 7 days after the notification to secure their grants. Otherwise, their grants will be revoked and assigned to other candidate.
• The granted candidates must attend all the sessions and sign the attendance list in order to obtain their grants.

• The granted candidates must attend all the sessions of the course in order to obtain their grants.

Time Series Forecasting: From Classical Methods to Transformers and Diffusion Models
Weining Wang, University of Bristol, UK.

Description: This module provides a concise yet rigorous introduction to modern time series forecasting, bridging classical econometric methods and recent advances in machine learning. We begin with foundational concepts, including ARIMA, highlighting their probabilistic structure and forecasting principles. The course then reformulates forecasting as a supervised learning problem, introducing machine learning approaches such as neural networks.

The second part of the module focuses on attention-based models and Transformers, explaining the self-attention mechanism and its advantages for modeling long-range temporal dependencies. We discuss global forecasting models and cross-series learning in high-dimensional settings. Finally, we introduce diffusion models for probabilistic forecasting, emphasizing their ability to generate full predictive distributions and capture complex, multimodal uncertainty.

issues of estimation and inference. The aim is to equip participants with both practical tools and conceptual understanding of current research frontiers in time series forecasting.

Monte Carlo Methods for financial time series: MCMC, particle filters and pseudo-marginal inference
Michael Pitt, King's College London, UK.

Description: This module provides a concise introduction to modern Monte Carlo methods for inference in financial time series models. The focus is on latent-variable and state-space models, including stochastic volatility, regime-switching and related models used in empirical finance, where nonlinear dynamics, non-Gaussian observations or intractable likelihoods make standard methods difficult to apply.

The module first reviews key ideas in Markov chain Monte Carlo, including Metropolis-Hastings, Gibbs sampling, data augmentation and simulation smoothing. It then introduces sequential Monte Carlo methods, including particle filters, auxiliary particle filters and likelihood estimation for state-space models. These methods are subsequently connected to particle MCMC and pseudo-marginal inference, which combine MCMC with unbiased likelihood estimates for Bayesian inference in complex time series models.

Throughout the module, the methods will be illustrated using financial time series examples, with accompanying code in Python and R. The aim is to give participants both a conceptual understanding of the algorithms and practical experience of how Monte Carlo methods are used in modern financial econometrics.

Functional-based reserving: From scratch to CRAN
(why and how to create and deploy your own R package)

Matus Maciak, Charles University in Prague, Czech Republic.

Description: In this tutorial, we present a concise narrative of the full lifecycle of developing and deploying own R package.

The key stages of the process range from organising personal R code into a coherent package, through development, testing, and versioning, to the final submission to CRAN (the Comprehensive R Archive Network).

Throughout the course, we highlight both the practical challenges and the tangible benefits of this journey, using the R package ProfileLadder as an example.

A short story, a few lessons learned, and—hopefully—some encouragement to take the leap into package development.

Tuesday, 25 August 2026

• 09:30 – 10:30 Keynote Talk
• 10:30 – 11:00 Coffee break
• 11:00 – 12:30 Module I (Weining Wang)
• 12:30 – 14:00 Lunch break
• 14:00 – 15:00 Module I (Weining Wang)
• 15:00 – 15:30 Coffee break
• 15:30 – 17:30 Module I (Weining Wang)

Wednesday, 26 August 2026

• 09:00 – 10:30 Module II (Michael Pitt)
• 10:30 – 11:00 Coffee break
• 11:00 – 12:30 Module II (Michael Pitt)
• 12:30 – 14:00 Lunch break
• 14:00 – 15:30 Module II (Michael Pitt)

Thursday, 27 August 2026

• 09:00 – 10:00 Module II (Michael Pitt)
• 10:00 – 10:30 Coffee break
• 10:30 – 12:30 Module III (Matus Maciak)
• 12:30 – 14:00 Lunch break
• 14:00 – 15:30 Module III (Matus Maciak)
• 15:30 – 16:00 Coffee break
• 16:00 – 17:30 Module III (Matus Maciak)
• 17:40 – 18:30 Keynote Talk

To register please click here.