Paper: arXiv 2408.02973

Authors: Yaoyue Tang, Karina Arias-Calluari, Michael S. Harré

Abstract

This paper compares and contrasts stationarity between the conventional stock market and cryptocurrency. The dataset used for the analysis is the intraday price indices of the S&P500 from 1996 to 2023 and the intraday Bitcoin indices from 2019 to 2023, both in USD. We adopt the definition of `wide sense stationary’, which constrains the time independence of the first and second moments of a time series. The testing method used in this paper follows the Wiener-Khinchin Theorem, i.e., that for a wide sense stationary process, the power spectral density and the autocorrelation are a Fourier transform pair. We demonstrate that localized stationarity can be achieved by truncating the time series into segments, and for each segment, detrending and normalizing the price return are required. These results show that the S&P500 price return can achieve stationarity for the full 28-year period with a detrending window of 12 months and a constrained normalization window of 10 minutes. With truncated segments, a larger normalization window can be used to establish stationarity, indicating that within the segment the data is more homogeneous. For Bitcoin price return, the segment with higher volatility presents stationarity with a normalization window of 60 minutes, whereas stationarity cannot be established in other segments.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 3.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper employs advanced mathematical concepts such as the Wiener-Khinchin Theorem, Fourier transforms, and detailed derivations of windowing functions, placing it in the high math complexity range. However, the empirical implementation is primarily theoretical and methodological, with no mention of backtesting, live data integration, or specific quantitative performance metrics, resulting in low empirical rigor.

Research Flowchart

  flowchart TD
  A["Research Goal: Compare Stationarity of S&P500 vs. Bitcoin"] --> B["Data Collection & Preprocessing"]
  
  subgraph B ["Inputs"]
      B1["S&P500 Intraday<br>1996-2023"]
      B2["Bitcoin Intraday<br>2019-2023"]
  end

  B --> C{"Core Methodology<br>Wiener-Khinchin Theorem"}
  C --> D["Detrending & Normalization<br>Segmentation of Data"]
  D --> E["Stationarity Testing<br>Power Spectrum Analysis"]

  subgraph E ["Computational Process"]
      E1["Calculate Autocorrelation"]
      E2["Calculate Power Spectral Density<br>via Fourier Transform"]
  end

  E --> F{"Key Findings"}

  subgraph F ["Outcomes"]
      F1["S&P500: Stationary<br>Full 28-year period achievable<br>12m Detrend / 10min Norm"]
      F2["Bitcoin: Segmented Stationarity<br>Only in high volatility segments<br>60min Norm required"]
  end

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 a novel stationarity testing approach based on the Wiener‑Khinchin theorem, applying detrending and normalization windows to intraday price returns.

Source sentence
The testing method used in this paper follows the Wiener-Khinchin Theorem, i.e., that for a wide sense stationary process, the power spectral density and the autocorrelation are a Fourier transform pair. This pa-per explores alternative approaches for testing stationar-ity based on the Wiener-Khinchin Theorem, which states the that the power spectrum of a stationary random pro-cess is the Fourier transform of its

Assumes 3

  • The time series are treated as wide‑sense stationary processes, requiring constant mean and autocovariance. statistical
    Source sentence
    We adopt the definition of `wide sense stationary', which constrains the time independence of the first and second moments of a time series. A con-tinuous time random process {Xt} (as above) is a Wide-Sense Stationary Process (WSP) if it constrains the arXiv:2408.02973v1 [q-fin.ST] 6 Aug 2024 2 mean µX(t) ≜E[Xt] and auto-covariance KXX(t1, t2) ≜ E[(Xt1 −µX(t1))(Xt2 −µX(t2))] in the following fashion: µX(t) = µX(t +
  • Detrending and normalization are necessary preprocessing steps before applying the stationarity test. model
    Source sentence
    We demonstrate that localized stationarity can be achieved by truncating the time series into segments, and for each segment, detrending and normalizing the price return are required. Both detrending and normalizing processes are required for each dataset before the stationarity test can be performed.
  • The intraday price indices are measured in USD and are representative of the underlying markets. data
    Source sentence
    The dataset used for the analysis is the intraday price indices of the S&P500 from 1996 to 2023 and the intraday Bitcoin indices from 2019 to 2023, both in USD. The S&P500 index data cover the period from Jan 1st, 1996 to Dec 31st, 2023, and Bitcoin data span from Apr 2nd, 2019 to Dec 31st, 2023.

Data

  • Intraday S&P500 price index market-prices unknown
    1996-01-01 to 2023-12-31 · S&P500
    Source sentence
    The dataset used for the analysis is the intraday price indices of the S&P500 from 1996 to 2023 and the intraday Bitcoin indices from 2019 to 2023, both in USD. The S&P500 index data cover the period from Jan 1st, 1996 to Dec 31st, 2023, and Bitcoin data span from Apr 2nd, 2019 to Dec 31st, 2023.
  • Intraday Bitcoin price index (BTC/USD) crypto-onchain unknown
    2019-04-02 to 2023-12-31 · Bitcoin
    Source sentence
    The dataset used for the analysis is the intraday price indices of the S&P500 from 1996 to 2023 and the intraday Bitcoin indices from 2019 to 2023, both in USD. The S&P500 index data cover the period from Jan 1st, 1996 to Dec 31st, 2023, and Bitcoin data span from Apr 2nd, 2019 to Dec 31st, 2023.

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • The S&P500 price return achieves stationarity for the full 28‑year period using a 12‑month detrending window and a 10‑minute normalization window. statistical
    Source sentence
    These results show that the S&P500 price return can achieve stationarity for the full 28-year period with a detrending window of 12 months and a constrained normalization window of 10 minutes.
  • When the S&P500 series is segmented, a larger normalization window can be used, indicating greater homogeneity within segments. statistical
    Source sentence
    With truncated segments, a larger normalization window can be used to establish stationarity, indicating that within the segment the data is more homogeneous.
  • For Bitcoin, only the high‑volatility segment shows stationarity with a 60‑minute normalization window; other segments do not achieve stationarity. statistical
    Source sentence
    For Bitcoin price return, the segment with higher volatility presents stationarity with a normalization window of 60 minutes, whereas stationarity cannot be established in other segments.
  • The full 28‑year S&P500 dataset is globally non‑stationary, but local segments achieve stationarity with appropriate detrending and normalization. statistical
    Source sentence
    Testing results for the S&P500 demonstrate global non-stationarity for the full 28-year dataset.
  • Reducing the normalization window (Δ2t) increases the degree of stationarity for the S&P500, with Δ2t = 10 min yielding stationarity for the full dataset. statistical
    Source sentence
    The gap between | c x∗∗(f)|2 and [ Cx∗∗(f) de-creases with a gradual decrease in the size of ∆2t, indi-cating an increase in the degree of stationarity. The nor-malized price return achieves stationarity with ∆2t = 10 min for the full data set of the S&P500 from Jan 1996 to Dec 2023.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 20 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.