Guide · Calendar
Does Halloween Indicator work?
Across 375 stocks, Halloween Indicator cut the worst fall on 89%, beat buying and holding on risk-adjusted return on 33% and on raw return on 27%. Mainly a seatbelt: it softened the big falls on most stocks, but usually gave up return to do it.
What it predicts
Own stocks from November through April; sit in T-bills from May through October.
- In at the close on the last trading day of October.
- Out at the close on the last trading day of April.
Calendar plays bet that some parts of the year and month have been reliably stronger than others. The family guide
How to use it
- It trades rarely (about 2.0 switches a year), so it suits someone who checks in occasionally, and a single bad call can last months.
- It worked best on Metals & mining (63% beat holding on risk-adjusted return) and worst on Big tech (0%).
- Use it as one input. Your signal combines the plays you count with the other factors you weight into one call.
When it worked
Halloween Indicator did best against holding in the 2008 crash (+20% median gap) and worst in the 2020–21 rally (−17%). Median gap in return against holding the same stock (per year over stretches longer than 18 months; crypto excluded).
| 2008 crash | Recovery 2009–19 | COVID drop | 2020–21 rally | 2022 bear | Since Oct 2022 | |
|---|---|---|---|---|---|---|
| Halloween Indicator | +20% | −5% | 0% | −17% | +14% | −8% |
Where it worked
By kind of stock: how often it beat holding on risk-adjusted return, and the median Sharpe ratio gap.
| Group | Beat holding, risk-adjusted | Sharpe gap |
|---|---|---|
| Metals & mining | 63% | +0.02 |
| Micro caps | 50% | +0.04 |
| Media & telecom | 40% | -0.12 |
| Energy & power | 39% | -0.04 |
| Semiconductors | 35% | -0.04 |
| Industrials & materials | 33% | -0.07 |
| Real estate | 32% | -0.02 |
| Consumer & retail | 29% | -0.06 |
| Indexes & ETFs | 29% | -0.04 |
| Sectors | 25% | -0.05 |
| Healthcare | 22% | -0.08 |
| Industrials & rentals | 22% | -0.08 |
| Financials | 15% | -0.09 |
| Software & devices | 12% | -0.22 |
| Big tech | 0% | -0.33 |
Where it helped most
Sharpe ratio gap against holding, stocks with 3+ years of history
Right now
Halloween Indicator holds 0 of 377 names as of the Oct 13, 2026 close.
No changes of call in the last two weeks.
Backtest results since 2005 (or as long as each stock has traded), against holding the same stock over the same dates, after 0.05% per switch. Past results don't promise future ones. This analysis does not constitute trading advice. Please meet with an advisor or independently review sources before making any decision.