Global Market Making Games 2026 — Singapore Qualifier
An individual global market-making competition open to university students across Singapore.
News
Research, releases, events and guides from QRS. RSS feed
Post-training, explained: from SFT to GRPO
A long-read explainer on how language models are post-trained, from supervised fine-tuning and reward models to PPO, DPO and GRPO, with derivations and ten interactive figures.
GeoAFT: new project page and draft explainer
Geometric Attention Fine-Tuning now has its own project page, with an interactive geometry explorer, the controlled synthetic results and a draft explainer video.
MH3100, MH2100 and HE2003 solutions republished
Solutions for MH3100, MH2100 and HE2003 are republished in a multi-method, fully worked style.
Diverse Scenarios, Selective Evaluation: Certified Spectral-Risk Decisions under Costly Robustness Tests
Workshop paper accepted at the EvoRobust Workshop, NeurIPS 2026.
Five works accepted to NeurIPS 2026 workshops
Five works from our inaugural QRS Research Academy cohort (2026 Fall #1) were accepted to NeurIPS 2026 workshops, spanning softmax-attention geometry, policy improvement, risk-sensitive decision-making and AI-native research workflows.
Geometric Attention Fine-Tuning Is Representation-Relative: Autonomous ML Research with AI-Native Workflow
Workshop paper accepted at the AutoMLR Workshop, NeurIPS 2026.
PolicyAttention: Robustness and Failure Modes of Softmax Policy Improvement
Workshop paper accepted at the EvoRobust Workshop, NeurIPS 2026.
PolicyAttention: When Softmax Attention Becomes Policy Improvement
Workshop paper accepted at the PTA Workshop, NeurIPS 2026.
The Approximation Rank of Softmax Attention: Sharp Geometric Laws and Robust Interaction Dimension
Extended abstract accepted at the NeurReps Workshop, NeurIPS 2026.
QRS Research Academy
The QRS Research Academy takes members from guided research to independent research leadership. Applications are open from the Academy page.
Exact Trade-Level Attribution of FRTB-IMA Capital
The FRTBTrace paper is on arXiv (2609.32238).
PolicyAttention: Softmax Attention Implements Policy Mirror Descent for Closed-Loop Control
The PolicyAttention paper is on arXiv (2609.30500).
SoftRank preprint updated to version 2
Version 2 of “Intrinsic Interaction Geometry Controls the Low-Rank Complexity of Softmax Attention” is on arXiv (2608.28150).
IMC Tech Race @ NTU
An on-campus coding race by IMC featuring 20 rapid computer-science problems, a live leaderboard, networking with IMC technologists, and prizes.
Intrinsic Interaction Geometry Controls the Low-Rank Complexity of Softmax Attention
The SoftRank paper is on arXiv (2608.28150).
NTU-QRS AlphaAllocator: Bounded Transformer News Overlays for Chinese Macro-Asset Allocation
Presented at NLPCC 2026 Shared Task 4, as a poster and an oral.
Singapore vs Hong Kong Quant Trading Hackathon
A cross-border university quantitative-trading competition in which Singapore and Hong Kong teams built and deployed live crypto trading strategies, with city qualifiers followed by a SG-vs-HK final.
Quant & FinTech Events Guide, January 2026
Campus events, recruiting programmes, conferences and industry expos for students interested in quant and fintech.
QRS@NTU major update, December 2025
Website fixes to folders, search and ranking, and final examination papers added for over 70 SPMS and NBS courses.