← Research Library
New to these reports? Start here
  • Dotted-underlined words have a plain-English definition — hover or tap them. Every term is also on the glossary page.
  • "Null" means we found nothing, not that something broke. Most reports here are negative results, on purpose — knowing an idea doesn't work is the point.
  • Two questions get asked separately. First, is the effect real? Second, is it already priced into the betting odds? An effect can be completely real and still useless to bet on.
  • A "calibration" row is a self-check. It runs the same method on something already known to be true. If that fails, the whole report is unreliable — so it's reported alongside the findings.
  • If a confidence interval includes zero, the real effect might be nothing at all, so no claim gets made.

Strategy 008: Sell-in-May Calendar Effect

Source

Folk wisdom + Bouman & Jacobsen, The Halloween Indicator, "Sell-in-May-and-Go-Away": Another Puzzle (American Economic Review, 2002). Subsequent decades of replication.

Thesis

US equity returns from November–April materially exceed returns from May–October. The pattern has held in 36+ developed markets across many decades. Hypothesized mechanism: summer vacation liquidity drain, tax-loss harvesting October → January, year-end rebalancing.

Formula

  • Hold SPY November 1 – April 30
  • Hold cash (or short-term bonds) May 1 – October 31
  • Rebalance twice per year on May 1 and November 1

Why it might still work

  • Statistically significant over 100+ years of data in academic studies
  • Mechanism (vacation liquidity) hasn't disappeared
  • The "calendar effect" survives publication far better than most anomalies (Bouman & Jacobsen tracked it post-publication)

Why it might NOT work

  • Median May-Oct return is still positive — you're sitting in cash while the market grinds up most years
  • Modern algo trading has compressed many calendar anomalies
  • You miss occasional huge summer rallies (e.g., summer 2020 post-COVID)
  • Tax friction on entering/exiting twice per year

Implementation

  • Universe: SPY only (plus cash placeholder)
  • Rebalance: monthly (signal fires at May 1 + Nov 1 boundaries)
  • Position: 100% SPY Nov 1 – Apr 30, 0% rest of year (cash)
  • Note: harness doesn't model interest on cash — real-world would earn 4-5% short-term Treasury during May-Oct, which is meaningful

Data requirements

  • SPY price only

Expected outcome

  • Academic excess return: ~3-5%/yr historically
  • Risk-adjusted: better SharpeSharpe ratioReturn relative to how much it bounced around. Higher means smoother returns for the same profit. than buy-and-hold (lower vol from spending 6 months in cash)
  • But absolute CAGR is lower (cash returns 0%, market still drifts up most years)
  • In our 2015-2025 window: probably modest negative alpha because the period had unusually strong summer/fall (especially 2020-2024 pandemic + AI rallies)

Pass/fail criteria

  • Pass: Sharpe > SPY's despite lower CAGR (drawdown reduction is real)
  • Marginal: CAGR -2 to -4pp vs SPY, but Sharpe comparable
  • Fail: both CAGR AND Sharpe worse than SPY (no risk-adjusted edge)
  • Note: this is mostly a calibrationcalibrationA deliberate sanity check on the method itself: run it on something already known to be true. If it fails to detect the known thing, the method is broken and its other results mean nothing. check on the harness for on/off binary signals. Outcome is roughly known a priori.

On this page

Terms in this report

Sharpe ratio calibration

Related

Strategy 002: 12-1 Cross-Sectional Momentum Strategy 009: Insider Cluster Buying Strategy 007: Short-Term Reversal (1-week loser-winner) Strategy 006: Net-Net (Graham Deep Value) Strategy 005: VIX Term Structure Strategy 004: Piotroski F-Score Strategy 003: Post-Earnings Announcement Drift (PEAD) Strategy 001: Magic Formula (Greenblatt)

Source

backtests/docs/008_sell_in_may.md
updated 2026-05-24 23:24