AAPL overnight vs intraday return decomposition — last ~3 years
Surprising: over the last ~3 years AAPL’s gains accrued during the regular session, not overnight. Stacking only open→close legs produces +69.73% while stacking close→open legs produces −9.74% — an asymmetry that runs counter to the overnight-drift claim.
I tested this by decomposing each trading day from minute bars into the overnight leg (prior close → next open) and the intraday leg (open → close), then comparing cumulative growth paths and summary statistics across 749 days. The detailed charts, numbers, and methodology below show the full evidence: intraday returns drove the period’s performance and the overnight leg, on a log basis, actually detracted from the total.
For AAPL over the past ~3 years, was essentially all of its cumulative gain earned overnight (prior close → next open) while the regular session (open → close) netted out flat-to-negative — the overnight-drift anomaly? Thesis: stacking only the close-to-open moves captures the bulk of AAPL's total return while the open-to-close moves add almost nothing, so the stock does its real work while the market is shut.
How this was measured
Filtered AAPL minute bars to US regular trading hours (Mon–Fri, 09:30–16:00 ET). For each session, defined the session open as the first RTH minute's 'open' and the session close as the last RTH minute's 'close'. Computed overnight return for day t as open_t/close_{t-1}−1 and intraday return as close_t/open_t−1, so close-to-close equals (1+overnight)*(1+intraday)−1. Compared cumulative growth paths (stacking only overnight vs only intraday legs) and summarized means, medians, and positive-day rates over the past ~36 months.
The key numbers
Reading the numbers
Across 749 trading days the close-to-close series compounded to 0.532 (about +53.2%), but the overnight-only stacked return was -0.0974 (−9.74%) while the intraday-only stacked return was 0.6973 (+69.73%). In short: most long-term gain came intraday and overnight was a drag.
The charts
This growth-of-$1 chart shows three diverging paths: total (close→close) starts near 1.0024 and ends at 1.532, overnight-only starts 0.9922 and falls to 0.9026, while intraday-only climbs from 1.0102 to 1.6973. The eye should go to the rising intraday line and the falling overnight line — the intraday leg delivers the bulk of accumulation, and overnight moves subtract from that gain, leaving the total in between.
The end-of-window bars lay it out plainly: Total C→C = 0.532, Overnight-only = -0.0974, Intraday-only = 0.6973. The intraday bar is the tallest and exceeds the total because the negative overnight bar offsets part of those intraday gains — directly contradicting the idea that the cumulative gain was earned mainly overnight.
The daily-return boxes show overlapping but different shapes: overnight mean ≈ -0.0001 with min -0.0942 and max 0.0784, while intraday mean ≈ 0.0008 with min -0.0746 and max 0.1564. Intraday returns have a larger right tail (bigger upside days) and a positive mean, whereas overnight has a slightly negative mean and a deeper left tail; combine that with similar win rates (overnight ~51.00% positive vs intraday ~53.40%) and you see why intraday stacking produced long-term growth while overnight stacking slightly reduced it.
Calendar-year cumulative returns (overnight vs intraday vs total)
| year | days | overnight_cum_ret | intraday_cum_ret | total_cum_ret |
|---|---|---|---|---|
| 2,024 | 252 | -0.0212 | 0.3352 | 0.3069 |
| 2,025 | 250 | -0.1165 | 0.2344 | 0.0905 |
| 2,026 | 123 | 0.0904 | -0.0229 | 0.0654 |
The takeaway
No — the data do not support the overnight-drift thesis for AAPL over the last ~3 years: the gains came during the regular session, not between close and next open. Close-to-close compounded to +53.20% over 749 trading days, while stacking only overnight legs produced -9.74% and stacking only intraday (open→close) legs produced +69.73%. Mean daily overnight return was essentially zero (≈ -0.008%) versus about +0.080% intraday, and intraday had a slightly higher hit-rate (53.4% vs 51.0%). On a log basis overnight actually detracted from total return (overnight log-share ≈ -24.0%), so the overnight leg did not drive the period’s performance. With 749 days and an ~80 percentage-point spread between intraday and overnight cumulative results, this is a strong, clear signal rather than a coin flip. Practical takeaway: over this window, you would have captured AAPL’s multi-year gain by being long through the regular session — the close-to-open leg would have reduced returns, not created them.
The fine print
- Session timestamps use the first 09:30 open and the last 15:59–16:00 close; small vendor timing differences can shift a few bps between legs.
- Half-days and altered schedules are included; shorter sessions can distort the intraday leg on those dates.
- Corporate actions in the minute bars (splits/dividends) can show up as overnight gaps if not adjusted and inflate the overnight leg.
- This attribution covers ~36 months only; different multi-year windows or regimes can produce different splits between overnight and intraday.