Paper: SSRN 7323419
Authors: José Enrique Delgado
Abstract
Zarattini, Aziz, and Barbon report that a simple intraday time-series momentum strategy applied to SPY produced a total return of 1,985%, an annualized return of 19.6%, and a Sharpe Ratio of 1.33 from May 2007 through April 2024. This paper performs an independent replication and a post-publication extension using approximately 2.6 million one-minute SPY observations from TradeStation spanning January 2000 through August 2026. After reconciling two material implementation details against the authors’ published Python reference code—the use of typical-price Volume Weighted Average Price (VWAP) and VWAP confirmation at entry—the original result is reproduced almost exactly: 1,989% total return and a Sharpe Ratio of 1.317 on the paper window under the closest paper-matching convention. The independent update also closely reproduces the authors’ September 2025 FAQ update: 2024 returns are 31.8% versus their 32.2%, and January–August 2025 returns are 1.4% versus their 1.0%. The longer sample changes the interpretation, but not in the simple form suggested by the pooled May 2024–August 2026 Sharpe of 0.35. Disaggregating the post-publication path reveals two sharply different regimes. Under a Q24-comparable monthly convention, May 2024–29 August 2025 returns 19.3% with Sharpe 1.057; using the exact first trading day after the 10 May 2024 publication, the corresponding return is 23.3% with Sharpe 1.284. By contrast, the subsequent 2 September 2025–14 August 2026 stretch returns -8.1% with Sharpe -0.461. Thus the evidence does not support a narrative of immediate post-publication alpha decay; it supports a strong initial post-publication period followed by a materially weak recent regime. A systematic search then evaluates asymmetric short filters, relative-volume sizing, volatility overlays, persistence filters, profit locks, breakeven rules, trade caps, daily loss limits, VWAP-dominance rules, intraday decay, regime overlays, and performance-based overlays. Only a small subset survives the consistent multi-window standard applied during the extension stage. The final frozen selected specification reaches a valid-period Sharpe Ratio of 1.274, a -14.8% maximum drawdown, and a 2110.6% total return over 2000–2026, while deliberately sacrificing terminal wealth relative to the baseline. To address multiple testing, a research ledger of 129 trials is reconstructed and 80 SPY search alternatives are identified. Combinatorially Symmetric Cross-Validation yields a Probability of Backtest Overfitting of 38.49% under the historical convention and 40.87% on a common-valid sample. The Deflated Sharpe Ratio of the final candidate, using 𝑁 = 80 as the primary multiple-testing count, is 99.999986796%. These statistics support a strong retrospective edge but do not provide a genuine prospective validation because the search was adaptive and the 2024–2026 data were observed during model development. The correct next experiment is therefore not another retrospective optimization, but the passage of time with frozen parameters.
Complexity vs Empirical Score
- Math Complexity: 6.5/10
- Empirical Rigor: 9.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
- Basis: provisional. Scored from the abstract alone because the full text was not available, so both scores reflect what the abstract claims, not a reading of the paper.
Why this score: This paper demonstrates high empirical rigor through extensive backtesting, data validation, and advanced statistical methods for overfitting. The mathematical complexity is present in the statistical techniques applied, particularly in addressing multiple testing. Its novelty lies in the detailed replication and post-publication analysis, revealing nuanced regime shifts and a systematic search for improvements.
Research Flowchart
flowchart TD
A["Research Goal: Replicate & Audit Intraday Momentum Strategy"] --> B{"Data: SPY 1-Min Observations (Jan 2000 - Aug 2026)"};
B --> C["Methodology: Independent Replication & Statistical Audit"];
C --> D{"Computational Process: Strategy Implementation, Performance Calculation, Multi-Window Analysis"};
D --> E["Outcome 1: Original Strategy Replicated - 1989% Total Return, Sharpe 1.317 (Paper Window)"];
D --> F["Outcome 2: Post-Publication Analysis - Strong Initial Period (Sharpe 1.284), Weak Recent Regime (Sharpe -0.461)"];
D --> G["Outcome 3: Systematic Search for Enhancements - Final Spec Sharpe 1.274, 2110.6% Total Return (2000-2026)"];
D --> H["Outcome 4: Multiple Testing & Overfitting Analysis - PBO ~40%, Deflated Sharpe Ratio ~100%"];
H --> I["Conclusion: Strong Retrospective Edge, but Not Prospective Validation - Next Step: Frozen Parameters & Passage of Time"];
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 independently replicates and extends a simple intraday time-series momentum strategy applied to SPY, evaluating many variations and selecting a final specification.
Source sentence
After reconciling two material implementation details against the authors’ published Python reference code—the use of typical-price Volume Weighted Average Price (VWAP) and VWAP confirmation at entry—the original result is reproduced almost exactly: 1,989% total return and a Sharpe Ratio of 1.317 on the paper window under the closest paper-matching convention.
Assumes 2
- Implementation uses typical-price Volume Weighted Average Price (VWAP) and VWAP confirmation at entry. execution
Source sentence
After reconciling two material implementation details against the authors’ published Python reference code—the use of typical-price Volume Weighted Average Price (VWAP) and VWAP confirmation at entry—the original result is reproduced almost exactly: 1,989% total return and a Sharpe Ratio of 1.317 on the paper window under the closest paper-matching convention.
- One‑minute SPY price data are available from TradeStation for the full sample period. data
Source sentence
This paper performs an independent replication and a post-publication extension using approximately 2.6 million one-minute SPY observations from TradeStation spanning January 2000 through August 2026.
Data
2000 – 2026 · S&P 500 ETF (SPY) · equities · intraday
- one‑minute SPY observations market-prices commercial
Source sentence
This paper performs an independent replication and a post-publication extension using approximately 2.6 million one-minute SPY observations from TradeStation spanning January 2000 through August 2026.
Costs
Unclear
Finds 5
Sharpe 1.27 · total return 2110.6% · max drawdown 14.8%
- Original paper reported a total return of 1,985%, annualized return of 19.6% and Sharpe Ratio of 1.33 for May 2007–April 2024. performance
Source sentence
Zarattini, Aziz, and Barbon report that a simple intraday time-series momentum strategy applied to SPY produced a total return of 1,985%, an annualized return of 19.6%, and a Sharpe Ratio of 1.33 from May 2007 through April 2024.
- Independent replication reproduces almost exactly a total return of 1,989% and Sharpe Ratio of 1.317 on the original window. performance
Source sentence
After reconciling two material implementation details against the authors’ published Python reference code—the use of typical-price Volume Weighted Average Price (VWAP) and VWAP confirmation at entry—the original result is reproduced almost exactly: 1,989% total return and a Sharpe Ratio of 1.317 on the paper window under the closest paper-matching convention.
- Post‑publication update shows 2024 returns of 31.8% (vs. 32.2% reported) and Jan–Aug 2025 returns of 1.4% (vs. 1.0%). performance
Source sentence
The independent update also closely reproduces the authors’ September 2025 FAQ update: 2024 returns are 31.8% versus their 32.2%, and January–August 2025 returns are 1.4% versus their 1.0%.
- Disaggregated post‑publication periods: May 2024–29 Aug 2025 returns 19.3% with Sharpe 1.057 (or 23.3%/1.284 using exact first trading day); Sep 2025–14 Aug 2026 returns –8.1% with Sharpe –0.461. performance
Source sentence
Under a Q24-comparable monthly convention, May 2024–29 August 2025 returns 19.3% with Sharpe 1.057; using the exact first trading day after the 10 May 2024 publication, the corresponding return is 23.3% with Sharpe 1.284. By contrast, the subsequent 2 September 2025–14 August 2026 stretch returns -8.1% with Sharpe -0.461.
- Final frozen specification achieves Sharpe Ratio 1.274, maximum drawdown –14.8% and total return 2,110.6% over 2000–2026. performance
Source sentence
The final frozen selected specification reaches a valid-period Sharpe Ratio of 1.274, a -14.8% maximum drawdown, and a 2110.6% total return over 2000–2026, while deliberately sacrificing terminal wealth relative to the baseline.
Checked by others
Supported by
- Out-of-Sample Evaluation of an Intraday Momentum Strategy for the S&P 500 ETF (SPY)
Compare the findings
This paper: Independent replication reproduces almost exactly a total return of 1,989% and Sharpe Ratio of 1.317 on the original window.
That paper: In-sample replication closely matches the original results, confirming reproducibility with a Sharpe Ratio of 1.34 over 2015-2024.
Code
No repository by the authors found in the paper.
Extracted 2026-11-15 · same card as JSON · A language model chose these items: check the source sentence before relying on one.