Fractiz
Pattern Detail
Santa Rally and January Effect
Whether the late-December rally and early-January push produce different RTH returns than the rest of the year.
Baseline Avg Return
+0.063%
Across 4,493 NQ sessions
Largest Deviation
+0.068%
January Effect vs baseline mean
Sample Range
1d
2008-01-02 to 2026-02-24
Trigger: Santa Rally = last 5 sessions of December · January Effect = first 5 sessions of January · Baseline = all other sessions
Per-Window Stats
| Window | Sessions | Avg Return | Median | Win Rate | Std Dev | Δ vs Baseline |
|---|---|---|---|---|---|---|
| Santa Rally | 90 | +0.024% | -0.061% | 45.6% | 1.09% | -0.039% |
| January Effect | 95 | +0.130% | +0.122% | 55.8% | 1.48% | +0.068% |
| Baseline (all sessions) | 4,493 | +0.063% | +0.121% | 55.7% | 1.41% | — |
Detection scan: NQ 1d · 2008-01-02 to 2026-02-24 · generated Apr 27, 2026
What this pattern measures
Two adjacent calendar-window phenomena reported side by side:
- Santa Rally: the last 5 trading sessions of December.
- January Effect: the first 5 trading sessions of January.
Both are compared to a baseline of every other session in the sample (all sessions outside both windows). Windows are kept disjoint, so each session is in at most one bucket and there’s no double-counting.
Definitions used on this page:
- Sessions are aggregated from RTH bars only (08:30 to 15:00 Central Time for CME equity index futures).
- Return is
(close − prior close) / prior close.
Why it matters
The Santa Rally and January Effect are two of the most-cited calendar claims in equity markets. The Santa Rally story attributes year-end strength to thin participation, tax-loss buy-backs, and bonus reinvestment. The January Effect story points to retirement-account contributions, fresh capital allocations, and small-cap mean reversion.
These are also the most-tortured claims in the literature. The data below shows whether the effect actually shows up in the specific instrument and date range, separate from generic-equity-index folklore.
Sample sizes are small: about 5 sessions per year per window, so even 18 years of data leaves under 100 sessions per bucket. A modest edge in either bucket can fail significance testing easily.
How to read the numbers
- Each bucket reports its own return distribution.
- Delta vs baseline is the difference of means. Positive means the bucket beat the rest of the year.
- Win rate is informative independently of the mean.
- Comparing the two buckets to each other is also useful: a strong Santa with a weak January (or vice versa) tells you which half of the bridge actually carries the effect for this instrument.
What’s not here
- The “if Santa fails to call, bears may come to Broad and Wall” claim. That conditional version requires a separate measurement.
- Different window sizes (some authors use last 5 + first 2, others use last 7 + first 5).
- Year-by-year breakdown.