SPB Git

spb/anomaly-atlas Public License

Systematic discovery & rigorous validation of statistical anomalies in open HF market data (hfmarketdata.io) — pre-registered, artifact-null-driven, fully reproducible. Live atlas: www.anomaly-atlas.io

quant-researchmarket-microstructureanomaliesreproducibilityduckdbapple-silicon
Python 61.4% JavaScript 28.7% CSS 8.6% Shell 0.7% Makefile 0.5%
4.7 KB
View source Raw Blame History

# project: anomaly-atlas document: Phase 1 notes — calendar & intraday seasonal effects author: Simon-Pierre Boucher contact: contact@spboucher.ai data_source: hfmarketdata.io created: 2026-08-12 status: reviewed

# Calendar & intraday seasonal effects (Q3)

Phase 1 literature notes. Citations verified via OpenAlex, accessed 2026-08-12. The calendar space is the p-hacking capital of finance — every claim below must be read against the Sullivan–Timmermann–White result (see multiple-testing notes).

# 1. The classic effects

  • French, K. R. (1980). Stock returns and the weekend effect. Journal of Financial Economics 8(1), 55–69. https://doi.org/10.1016/0304-405X(80)90021-5 — Systematically negative Monday returns. Status: decayed — one of the cleanest documented post-publication disappearances.
  • Ariel, R. A. (1987). A monthly effect in stock returns. Journal of Financial Economics 18(1), 161–174. https://doi.org/10.1016/0304-405X(87)90066-3 — Returns concentrate in the first half of the month (turn-of-month). Status: disputed/partially persistent.
  • Lakonishok, J. & Smidt, S. (1988). Are Seasonal Anomalies Real? A Ninety-Year Perspective. Review of Financial Studies 1(4), 403–425. https://doi.org/10.1093/rfs/1.4.403 — 90 years of DJIA data confirm turn-of-week/month/year and holiday effects, while explicitly warning about data-snooping — remarkably ahead of its time. Status: the effects it confirmed have mostly decayed; the warning is robust.
  • Wood, R. A., McInish, T. H. & Ord, J. K. (1985). An Investigation of Transactions Data for NYSE Stocks. Journal of Finance 40(3), 723–739. https://doi.org/10.2307/2327796 — Minute-data intraday U-shape in returns/volatility at open and close. Status: robust as a volatility/spread pattern; as a return anomaly, likely-artifact (auction mechanics, spread seasonality).

# 2. The modern intraday variant

  • Gao, L., Han, Y., Li, S. Z. & Zhou, G. (2018). Market intraday momentum. Journal of Financial Economics 129(2), 394–414. https://doi.org/10.1016/j.jfineco.2018.05.009 — First half-hour SPY return predicts the last half-hour. Testable 1:1 on our data (SPY 1min, 2000→2026). Status: disputed — decay after publication is itself a good expE sub-hypothesis.
  • Baltussen, G., Da, Z., Lammers, S. & Martens, M. (2021). Hedging demand and market intraday momentum. Journal of Financial Economics 142(1), 377–403. https://doi.org/10.1016/j.jfineco.2021.04.029 — Mechanism: option gamma-hedging flows; the effect varies with hedging conditions. Gives us a mechanism prior — and our options chains (67 quarters) can proxy the hedging-demand state.

# 3. Decay evidence specific to calendars

  • Schwert, G. W. (2003). Anomalies and Market Efficiency. Handbook of the Economics of Finance, 939–974. https://doi.org/10.1016/S1574-0102(03)01024-0 — Size, weekend, dividend, and January effects weaken or vanish after publication.
  • Marquering, W., Nisser, J. & Valla, T. (2006). Disappearing anomalies: a dynamic analysis of the persistence of anomalies. Applied Financial Economics 16(4), 291–302. https://doi.org/10.1080/09603100500400361 — Rolling-window analysis: weekend, holiday, time-of-month, January all gone post-publication; turn-of-month persisted (as of 2006). Makes turn-of-month the single most interesting calendar hypothesis to re-test on 2000–2026 data.

# 4. Artifact confounds specific to Q3 (feed the taxonomy)

  1. Spread/staleness intraday seasonality: spreads and staleness are widest at the open — a "first-30-minutes effect" in returns can be pure T1/T2 artifact. expE must measure the intraday profile of our bounce null first (taxonomy open item).
  2. Auction-close convention (T4): close-anchored calendar effects change magnitude depending on which close is used — daily bars vs last 1min bar differ by ~8 bp even on calm days (expA).
  3. Session-length changes & DST: ET wall-clock timestamps mean DST weeks shift the UTC session; naive UTC grouping manufactures day-of-week effects.
  4. Hypothesis-space explosion: day-of-week (5) × month (12) × turn-of-X × holiday × intraday half-hours (13) → hundreds of implicit tests. expE's budget must be pre-counted and corrected (White RC / SPA — see multiple-testing notes).

# 5. Implications for expE

Pre-register a SMALL set: (i) turn-of-month (the survivor per Marquering et al.), (ii) Monday effect (the canonical corpse — expected negative finding), (iii) intraday momentum first→last half-hour (Gao et al., with the gamma-hedging state split of Baltussen et al.), (iv) the intraday U-shape as an artifact demonstration, not an anomaly claim. Everything else is exploratory Level 0, reported as such.