IWM day-of-week seasonality over the last ~3 years
The classic weak-Monday/strong-Friday effect in small caps looks like a ghost: over the last ~3 years of IWM data the weekday means are statistically indistinguishable and the old weekend seasonality has largely washed out. We looked at 751 trading days; Monday averaged +0.233% and Friday −0.016% (Mon–Fri gap ≈ 0.249%), but the Kruskal–Wallis test shows no cross-weekday difference (H=4.425, p=0.3515) and the Monday vs. Friday Welch test only gives a suggestive lean (p≈0.0908), not a robust signal.
Below is the full breakdown — methodology, per-weekday moments, distributional plots and the hypothesis tests — so you can see why a small mean gap, high daily volatility, and these p-values argue against a dependable day-of-week edge.
For IWM over the past ~3 years, does the calendar weekday still drive small-cap returns — is the classic 'weak Monday, strong Friday' weekend effect alive, or has it washed out? Mean returns are statistically indistinguishable across weekdays with Mondays no longer negative, so the famous day-of-week seasonality has been arbitraged out even in small caps.
How this was measured
Resampled IWM minute bars to daily closes, computed close-to-close daily returns, and restricted the analysis to the last 36 months (or the available window if shorter). Grouped returns by calendar weekday (Mon–Fri), summarized per-weekday moments, ran a Kruskal–Wallis test to assess cross-weekday distribution differences, and a Welch two-sample t-test for Monday vs Friday mean returns. Monday returns naturally embed the weekend gap.
The key numbers
Reading the numbers
Kruskal–Wallis p = 0.3515 (>0.05) — mean returns do not differ clearly across weekdays. Monday's mean is 0.0023265712 and Friday's is -0.0001639348, a small Mon–Fri gap of 0.00249050599 that is not statistically significant (Welch p = 0.0908).
The charts
The box plot lays out the full daily return distribution for each weekday: Monday has mean 0.0023 (n=141) with min -0.0251 and max 0.0329, while Friday has mean -0.0002 (n=153) with min -0.0446 and max 0.038. Boxes and whiskers overlap heavily across all weekdays and Wednesday even shows a large positive extreme (max 0.1065), so there isn't a clean shift in the central tendency from Monday to Friday. In plain terms, the day-to-day spread is large relative to the tiny differences in means, which argues against a robust 'weak Monday / strong Friday' pattern in this sample.
The bar chart of weekday means shows Monday highest at 0.0023 and Friday lowest at -0.0002, with the five-day range only 0.00249050599. Those bars are visually very close together, and combined with the Kruskal–Wallis p = 0.3515 and Welch Mon vs Fri p = 0.0908, the small Mon–Fri gap does not hold up as a statistically clear weekend effect. So while Monday is slightly positive on average here, the classic seasonality appears washed out over this period.
IWM daily returns by weekday — summary
| Weekday | N | Mean | Median | Std | Pos fraction |
|---|---|---|---|---|---|
| Monday | 141 | 0.0023 | 0.003 | 0.0107 | 0.6383 |
| Tuesday | 155 | 0.0014 | -0.0002 | 0.0143 | 0.4839 |
| Wednesday | 153 | 0.0007 | -0.0003 | 0.0151 | 0.4902 |
| Thursday | 149 | -0.0001 | 0.001 | 0.0137 | 0.5101 |
| Friday | 153 | -0.0002 | 0 | 0.0143 | 0.4967 |
The takeaway
No — the classic weak-Monday/strong-Friday pattern has essentially washed out in IWM over the last ~3 years: weekday mean returns are not statistically different. Over 751 trading days, Monday averaged +0.233% (N=141) and Friday averaged −0.016% (N=153), a Monday–Friday mean gap of about 0.249%. The Kruskal–Wallis test (H=4.425, p=0.3515) finds no clear cross-weekday difference, and the Mon vs Fri Welch test (p≈0.0908) only gives a weak, suggestive lean rather than a solid signal. Put plainly, p=0.3515 is roughly a 35-in-100 chance these weekday differences are noise, and the Mon–Fri p≈0.09 is only a ~9-in-100 chance the gap isn’t luck — not strong enough to call reliable. Mondays still had a lower frequency of negative days (about 36.2% negative) but daily volatility (stds roughly 0.0107–0.0151) is large compared with these tiny mean gaps. Bottom line: there’s at best a weak hint of a Monday edge in the sample, not a dependable seasonal effect to trade on.
The fine print
- Window limited to the last ~36 months (751 trading days); longer-term seasonality can differ.
- This is an in-sample comparison only; no out-of-sample validation or multiple-comparison adjustment was performed.
- Close-to-close returns embed the weekend gap into Mondays; intraday or open-to-close slices could show different patterns.
- Per-weekday sample sizes differ (Mon 141, Fri 153) and daily volatility (~0.0107–0.0151) is large relative to mean gaps.