AI Research SPY

SPY weekday seasonality over the last ~3 years: is Monday still weakest?

Surprising outcome: over the trailing ~3 years (2023-07-03 to 2026-06-30) SPY’s Monday has been the strongest weekday, not the weakest. Across 751 trading days, Monday’s mean close‑to‑close return was about +0.24% versus roughly +0.04% for Tuesday–Friday, and the difference is statistically meaningful (Welch p = 0.0082; Kruskal–Wallis p = 0.0327). In short, the old “stocks are weak on Mondays” seasonality does not hold up in this window.

This study used minute bars resampled to daily returns and grouped observations by weekday, then tested distributional differences with Kruskal–Wallis and pairwise Welch tests. Below you’ll find the full tables, charts, and robustness checks that justify the finding and show its magnitude and limits.

The research question

For SPY over the past ~3 years, is there still a day-of-week edge — is Monday reliably the weakest session while Friday or midweek quietly does the heavy lifting, or do all five weekdays average out to statistically the same return? Thesis: the old Monday-effect seasonality has washed out, with average returns across Monday–Friday indistinguishable from one another, so 'stocks are weak on Mondays' is folklore the modern tape no longer honors.

How this was measured

Minute bars resampled to daily close-to-close returns, then restricted to the trailing ~3 years anchored to the available SPY data window (2023-07-03 to 2026-06-30). Returns grouped by weekday (Mon–Fri). We summarize mean/median/std and fraction-positive per weekday, run a Kruskal–Wallis test across all five weekdays to assess distributional differences, and run Welch t-tests for Monday vs Tue–Fri and Friday vs Mon–Thu to quantify the classic 'Monday effect' and Friday strength narratives.

The key numbers

Trading days analyzed
751
2023-07-03 to 2026-06-30
Kruskal-Wallis H-statistic (Mon–Fri)
10.504
Higher H = stronger weekday effect
Kruskal-Wallis p-value (Mon–Fri)
0.0327
p=0.0327 < 0.05 → at least one weekday's return distribution differs
Mean Monday return
0.2420%
N=143 Mondays
Mean Tue–Fri return
0.0406%
N=608 Tue–Fri
Gap (Monday − Tue–Fri)
0.2014%
Welch p-value (Mon vs Tue–Fri)
0.0082
p=0.0082 < 0.05 → Monday differs from Tue–Fri
Fraction positive — Monday
67.832%
Fraction positive — Tue–Fri
52.961%
Mean Friday return
-0.0074%
N=151 Fridays
Gap (Friday − Mon–Thu)
-0.1081%
Welch p-value (Fri vs Mon–Thu)
0.2384
p=0.2384 ≥ 0.05 → no Fri vs Mon–Thu difference

Reading the numbers

Kruskal–Wallis H=10.504 (p=0.0327) says weekday return distributions aren’t all identical. Monday’s mean (0.0024198) exceeds Tue–Fri (0.0004059) with Welch p=0.0082, and Mondays were positive 0.6783 vs 0.5296 for Tue–Fri.

The charts

SPY daily returns by weekday (last ~3 years)
What this chart says

The box plot shows Monday's distribution sits higher on the y-axis: mean Monday return 0.0024 (n=143) with min −0.0284 and max 0.0303, while other weekdays cluster much closer to zero. Note the big single upside on Wednesday (max 0.1125) and the deeper negative tail on Friday (min −0.0575) — these are individual extremes, not a consistent Monday weakness. In plain terms, the center of Monday’s returns is shifted up relative to most other days, which argues against the old 'Mondays are the weakest' story.

Mean SPY daily return by weekday
What this chart says

The bar chart makes the same point cleanly: Monday’s mean (0.0024) is visibly taller than Tuesday (0.0008), Wednesday (0.0013), Thursday (−0.0004) and Friday (−0.0001). The numerical gap Monday minus Tue–Fri is 0.002014 and is statistically supported (Welch p=0.0082; Kruskal–Wallis p=0.0327), while Friday’s slightly negative mean is small and not a significant outlier versus Mon–Thu (Welch p=0.2384). So on average over this ~3-year window Monday is actually stronger, not weaker.

Weekday return summary (last ~3 years)

WeekdayNMeanMedianStdFraction positive
Monday1430.00240.00240.00750.6783
Tuesday1560.00080.0010.0090.5321
Wednesday1530.00130.00170.01240.549
Thursday148-0.0004-0.00050.00930.4865
Friday151-0.00010.00050.01010.5497

The takeaway

No — the old ‘stocks are weak on Mondays’ story does not hold over the last ~3 years: Mondays have been the strongest weekday, not the weakest. Monday’s mean daily return was about +0.24% versus roughly +0.04% for Tue–Fri (gap ≈ +0.20%), and the Kruskal–Wallis test (p = 0.0327) says at least one weekday differs. The Monday vs Tue–Fri comparison is statistically strong: Welch p = 0.0082 (only about an 8‑in‑1000 chance this pattern is pure luck), and 67.8% of Mondays were positive versus ~53.0% on other weekdays (751 trading days total; 143 Mondays). Friday shows no evidence of carrying the gains — its mean was essentially zero (≈ −0.007%) and the Fri vs Mon–Thu gap (≈ −0.108%) is not significant (p = 0.238). Practically: for this 2023‑07‑03 to 2026‑06‑30 window, you can’t rely on “Mondays are weak”; if anything, Monday has been the quietly strong day, though the effect size is modest in absolute terms.

The fine print