Research map / Time series / Foundation models for time series
Reasoning on Time-Series for Financial Technical Analysis
Kelvin J.L. Koa, Jan Chen, Yunshan Ma, Zheng Huanhuan, Tat-Seng Chua
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| first RL fine-tuning (Time-GRPO) alone (ablation) | base reasoning model · all data | MSE: 1.6% improvement | Improved |
| add rejection sampling and SFT before second Time-GRPO RL (ablation) | VTA reasoning model · all data | MSE: 20.3% average improvement | Improved |
| condition reasoning on the forecasting model (joint conditional training) (ablation) | VTA · all data | MSE/MAE | Improved |
| remove technical indicators from reasoning (ablation) | VTA · StockNet | forecast error | Hurt |
| add adversarial noise to indicator values (ablation) | VTA · StockNet | forecast error | No clear effect |
| use verbal reasoning with inference-only LLMs instead of fine-tuned Transformer | GPT-4.1 mini · All Data | MSE: 2.6047 -> 0.2014 | Improved |
| add verbal reasoning to latent forecasting (VTA vs CALF) | CALF · All Data | MSE: 0.1235 -> 0.1178 | Improved |
| use VTA forecasts in Markowitz portfolio | TimeLLM · Portfolio | Sharpe ratio: 1.5230 -> 1.7190 | Improved |
| apply VTA to ILI instead of financial data | VTA · ILI | MSE: 1.0795 -> 1.4366 | Hurt |
| apply VTA to ETTh1 | CALF · ETTh1 | MSE: 0.0361 -> 0.0346 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.
- CALF — CALF: Aligning LLMs for Time Series Forecasting via Cross-modal Fine-Tuning (AAAI 2025) · better on 11 of 11 shared results
- TimeLLM — Time-LLM: Time Series Forecasting by Reprogramming Large Language Models (ICLR 2024) · better on 11 of 11 shared results
- TimesNet — TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis (ICLR 2023) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- FinBen: A Holistic Financial Benchmark for Large Language Models (NeurIPS 2024 · similarity 0.85)
- TimeXL: Explainable Multi-modal Time Series Prediction with LLM-in-the-Loop (NeurIPS 2025 · similarity 0.84)
- A Closer Look at Transformers for Time Series Forecasting: Understanding Why They Work and Where They Struggle (ICML 2025 · similarity 0.84)
- MASTER: Market-Guided Stock Transformer for Stock Price Forecasting (AAAI 2024 · similarity 0.84)
- Market-GAN: Adding Control to Financial Market Data Generation with Semantic Context (AAAI 2024 · similarity 0.84)
- A Set-Sequence Model for Time Series (NeurIPS 2026 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
