Paper: arXiv 2512.24621

Authors: Lucas A. Souza

Abstract

We study short-horizon forecasting in financial time series under strict causal constraints, treating the market as a non-stationary stochastic system in which any predictive observable must be computable online from information available up to the decision time. Rather than proposing a machine-learning predictor or a direct price-forecast model, we focus on \emph{“constructing”} an interpretable causal signal from heterogeneous micro-features that encode complementary aspects of the dynamics (momentum, volume pressure, trend acceleration, and volatility-normalized price location). The construction combines (i) causal centering, (ii) linear aggregation into a composite observable, (iii) causal stabilization via a one-dimensional Kalman filter, and (iv) an adaptive ``forward-like’’ operator that mixes the composite signal with a smoothed causal derivative term. The resulting observable is mapped into a transparent decision functional and evaluated through realized cumulative returns and turnover. An application to high-frequency EURUSDT (1-minute) illustrates that causally constructed observables can exhibit substantial economic relevance in specific regimes, while degrading under subsequent regime shifts, highlighting both the potential and the limitations of causal signal design in non-stationary markets.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 6.0/10 (capped from 7.0: transaction costs not modelled; no out-of-sample test; see penalty flags)
  • Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 7.0)

Why this score: The paper employs advanced mathematical tools such as Kalman filtering, causal derivative approximations, and stochastic process theory, indicating high math complexity. It also demonstrates empirical rigor through application to real high-frequency EURUSDT data with realized returns and turnover evaluation, though it stops short of presenting full backtest code or exhaustive parameter optimization.

Research Flowchart

  flowchart TD
  A["Research Goal: Construct Causal Observable<br>for Non-Stationary FX Forecasting"] --> B{"Data Input:<br>High-Freq EURUSDT 1min"}
  B --> C["Key Methodology Steps"]
  C --> C1["Causal Centering &<br>Micro-feature Aggregation"]
  C1 --> C2["Kalman Filter<br>Stabilization"]
  C2 --> C3["Forward-Like Operator<br>Adaptive Mixing"]
  C3 --> D["Computational Process:<br>Interpretable Decision Functional"]
  D --> E["Key Outcomes:<br>Economic Relevance in Regimes<br>Limitations under Shifts"]

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 constructing an interpretable causal signal from heterogeneous micro‑features, using causal centering, linear aggregation, Kalman filtering and a forward‑like operator, and evaluates it via a transparent decision rule.

Source sentence
Rather than proposing a machine-learning predictor or a direct price-forecast model, we focus on constructing an interpretable causal signal from heterogeneous micro-features that encode complementary aspects of the dynamics (momentum, volume pressure, trend acceleration, and volatility-normalized price location).

Assumes 5

  • Predictive observables must be computable online from information available up to the decision time (strict causality). model
    Source sentence
    any predictive observable must be computable online from information available up to the decision time.
  • The market is treated as a non‑stationary stochastic system. market
    Source sentence
    treating the market as a non-stationary stochastic system
  • All empirical results are reported without transaction costs, isolating intrinsic signal effects. execution
    Source sentence
    Importantly, all results in this section are computed without transaction costs.
  • Trading activity is measured by counting state changes (turnover) and assumes the ability to trade at observed prices. execution
    Source sentence
    Trading activity is quantified by counting state changes.
Show 1 more
  • The raw price series is filtered to retain only active Forex trading hours, removing weekends and illiquid periods. data
    Source sentence
    Prior to computing any indicators, the raw price series is filtered to retain only active Forex trading hours (local time). Saturdays are removed entirely, as well as Sundays before 18:00 and Fridays after 18:00.

Data

Backtest sample: EURUSDT · crypto · intraday

  • EURUSDT 1‑minute exchange rate crypto-onchain unknown
    EURUSDT pair
    Source sentence
    An application to high-frequency EURUSDT (1‑minute) illustrates that causally constructed observables can exhibit substantial economic relevance in specific regimes.

Costs

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

Source sentence
Importantly, all results in this section are computed without transaction costs.

Finds 4

  • The causal signal delivers strong outperformance relative to buy‑and‑hold from 2023 through approximately September 2024. performance
    Source sentence
    First, the induced dynamics exhibit substantial outperformance relative to buy‑and‑hold during extended intervals, most visibly throughout 2023 and up to approximately September 2024.
  • After September 2024 the cumulative performance plateaus while trading activity remains high, indicating regime‑dependent degradation. performance
    Source sentence
    From late 2024 onward, the cumulative performance partially retraces and then fluctuates around a near‑plateau, while trading activity remains elevated.
  • A purely smoothed or delayed version of the raw composite signal F0(t) lacks predictive structure, confirming the necessity of the derivative‑based forward component. theoretical
    Source sentence
    This contrast demonstrates that the derivative‑based component is essential: a purely smoothed or delayed version of F0(t) does not introduce additional predictive structure, whereas the causal derivative contributes local directional information.
  • Even without transaction costs, the high turnover of the strategy suggests that small frictions would materially reduce net performance. performance
    Source sentence
    Given the high turnover, even small frictions would materially reduce net performance, so the reported outcomes should be interpreted as evidence of predictive structure under idealized conditions rather than as an execution‑ready trading strategy.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 24 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 1 paper in the archive

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 (26)

  1. Technical Analysis of the Financial Markets: A Comprehensive Guide to Trading Methods and Applications (1999)
  2. Adaptive Feature Subset and Dynamic Trend Indicators for Medium-Term Stock Market Predictions: A 70 Trading Days Forecasting Approach (2024) — link
  3. Causality-driven multivariate stock movement forecasting (2024) — link
  4. Cryptocurrency trading: a comprehensive survey (2022) — link
  5. Application of artificial neural network models and principal component analysis method in predicting stock prices on Tehran Stock Exchange (2015) — link
  6. Inside the black box: a simple guide to quantitative and high-frequency trading (2013)
  7. When Finance Meets Physics: The Impact of the Speed of Light on Financial Markets and Their Regulation (2014) — link
  8. Physics Applications for Finance and their Practical Implementation (2025) — link
and 18 more
  1. Theory of Financial Risk and Derivative Pricing: From Statistical Physics to Risk Management (2003) — link
  2. Herd behavior and aggregate fluctuations in financial markets (2000) — link
  3. Exogenous and endogenous price jumps belong to different dynamical classes (2022) — link
  4. Momentum crashes (2016) — link
  5. Few-Shot Learning Patterns in Financial Time-Series for Trend-Following Strategies — → 2310.10500
  6. Stock price forecasting for companies listed on Tehran stock exchange using multivariate adaptive regression splines model and semi-parametric splines technique (2015) — link
  7. Forecasting financial signal for automated trading: An interpretable approach (2023) — link
  8. Financial news-based stock movement prediction using causality analysis of influence in the Korean stock market (2019) — link
  9. Analysing the information flow between financial time series: An improved estimator for transfer entropy (2002) — link
  10. Time series momentum (2012) — link
  11. Measuring Information Transfer (2000) — link
  12. Spectra of Some Self-Exciting and Mutually Exciting Point Processes (1971) — link
  13. A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle (1989) — link
  14. Scaling behaviour in the dynamics of an economic index (1995) — link
  15. A New Approach to Linear Filtering and Prediction Problems (1960) — link
  16. The limits of statistical significance of Hawkes processes fitted to financial data (2020) — link
  17. New Concepts in Technical Trading Systems (1978)
  18. The Moving Average Convergence–Divergence Trading Method: Advanced Techniques for Operating in Bull and Bear Markets (1979)

1 of 26 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.