Paper: arXiv 2601.10591

Authors: Arundeep Chinta, Lucas Vinh Tran, Jay Katukuri

Abstract

Time Series Foundation Models (TSFMs) have emerged as a promising approach for zero-shot financial forecasting, demonstrating strong transferability and data efficiency gains. However, their adoption in financial applications is hindered by fundamental limitations in uncertainty quantification: current approaches either rely on restrictive distributional assumptions, conflate different sources of uncertainty, or lack principled calibration mechanisms. While recent TSFMs employ sophisticated techniques such as mixture models, Student’s t-distributions, or conformal prediction, they fail to address the core challenge of providing theoretically-grounded uncertainty decomposition. For the very first time, we present a novel transformer-based probabilistic framework, ProbFM (probabilistic foundation model), that leverages Deep Evidential Regression (DER) to provide principled uncertainty quantification with explicit epistemic-aleatoric decomposition. Unlike existing approaches that pre-specify distributional forms or require sampling-based inference, ProbFM learns optimal uncertainty representations through higher-order evidence learning while maintaining single-pass computational efficiency. To rigorously evaluate the core DER uncertainty quantification approach independent of architectural complexity, we conduct an extensive controlled comparison study using a consistent LSTM architecture across five probabilistic methods: DER, Gaussian NLL, Student’s-t NLL, Quantile Loss, and Conformal Prediction. Evaluation on cryptocurrency return forecasting demonstrates that DER maintains competitive forecasting accuracy while providing explicit epistemic-aleatoric uncertainty decomposition. This work establishes both an extensible framework for principled uncertainty quantification in foundation models and empirical evidence for DER’s effectiveness in financial applications.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 6.0/10 (capped from 6.5: transaction costs not modelled; see penalty flags)
  • Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 6.5)

Why this score: The paper introduces advanced Bayesian concepts (Deep Evidential Regression, NIG priors) and derives uncertainty decompositions, yet backs it with a rigorous controlled comparison study, uncertainty-aware trading backtests, and specific performance metrics on cryptocurrency data.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Zero-Shot TSFM with<br>Principled Uncertainty Quantification"] --> B["Methodology<br>Deep Evidential Regression DER"]
  B --> C["Data<br>Cryptocurrency Time Series"]
  C --> D["Training<br>LSTM with DER Evidence Learning"]
  D --> E{"Evaluation"}
  E --> F["Comparison Methods<br>Gaussian NLL, Student-t NLL, Q-Loss, CP"]
  E --> G["DER Results"]
  G --> H["Outcome: Epistemic-Aleatoric<br>Uncertainty Decomposition"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes ProbFM, a transformer-based probabilistic foundation model that uses Deep Evidential Regression to provide principled epistemic-aleatoric uncertainty decomposition.

Source sentence
For the very first time, we present a novel transformer-based probabilistic framework, ProbFM (probabilistic foundation model), that leverages Deep Evidential Regression (DER) to provide principled uncertainty quantification with explicit epistemic-aleatoric decomposition.

Assumes 3

  • A directional trading strategy is used where positions are taken based on the sign of the predicted return. execution
    Source sentence
    Trading Strategy: We implement a directional trading strategy where positions are taken based on predicted returns: • Long (BUY): If predicted return > 0 • Short (SELL): If predicted return ≤0
  • Trades are filtered out when the predicted total uncertainty exceeds the 75th percentile threshold. execution
    Source sentence
    Uncertainty-Aware Filtering: For probabilistic methods, we evaluate an additional strategy where trades are executed only when predicted total uncertainty is below the 75th percentile threshold.
  • The cryptocurrency dataset consists of daily 1‑day log returns calculated from closing prices. data
    Source sentence
    Construction: 1-day log returns calculated as rt = log(Pt/Pt−1) where Pt represents the closing price at time t

Data

Backtest sample: BTC · crypto · daily

  • Cryptocurrency Dataset · Stooq crypto-onchain public
    January 1, 2020 to October 3, 2025 · 11 most liquid cryptocurrencies (ADA, BNB, BTC, DASH, DOGE, ETH, LTC, SOL, USDC, USDT, XRP
    Source sentence
    We conduct experiments on a Cryptocurrency Dataset consisting of the 11 most liquid cryptocurrency assets downloaded from Stooq.1 The dataset spans from January 1, 2020 to October 3, 2025

Costs

Not modeled A trading or execution result is reported without costs.

Finds 3

Reported performance (Evidential Regression (ProbFM core)): Sharpe 1.33

  • Evidential Regression (ProbFM core) achieves comparable RMSE and MAE to baseline methods, with slightly lower correlation, showing uncertainty‑focused approach does not sacrifice predictive accuracy. predictive-accuracy
    Source sentence
    Evidential Regression (ProbFM core) achieves comparable RMSE and MAE to baseline methods, with slightly lower correlation.
  • Evidential Regression attains the highest trading performance with Annual Sharpe 1.33, Annual Sortino 2.27, Calmar 3.04 and win rate 0.52, outperforming other probabilistic methods. performance
    Source sentence
    Evidential Regression (ProbFM core) 1.33 2.27 -15.14 3.04 0.52
  • Evidential Regression maintains the highest win rate (0.52) alongside superior downside risk management, as reflected in the Calmar ratio of 3.04. performance
    Source sentence
    Evidential Regression maintains the highest win rate (0.52) alongside superior downside risk management.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 23 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (19)

  1. Lag-Llama: Towards Foundation Models for Probabilistic Time Series Forecasting (2023) — link
  2. Evidential Deep Learning to Quantify Classification Uncertainty (2018)
  3. Deep Evidential Regression (2020)
  4. DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks (2020)
  5. Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting (2021)
  6. TimeGPT-1 (2023) — link
  7. Toto: Time Series Optimized Transformer for Observability (2024) — link
  8. Unified Training of Universal Time Series Forecasting Transformers (2024) — link
and 11 more
  1. A Time Series is Worth 64 Words: Long-term Forecasting with Transformers (2023) — link
  2. Long-term Forecasting with TiDE: Time-series Dense Encoder (2023) — link
  3. Bayesian Data Analysis (2013)
  4. Attention is All You Need (2017)
  5. GLU Variants Improve Transformer (2020) — link
  6. Decoupled Weight Decay Regularization (2017)
  7. SGDR: Stochastic Gradient Descent with Warm Restarts (2016)
  8. Obtaining well calibrated probabilities using bayesian binning (2015)
  9. On calibration of modern neural networks (2017)
  10. Portfolio Selection: Efficient Diversification of Investments (1959)
  11. A New Interpretation of Information Rate (1956) — link

0 of 19 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.