NFLX earnings-day vs non-earnings-day contribution to cumulative return (last ~3 years)
Contrary to the common narrative that a handful of earnings days drive Netflix’s moves, the dozen earnings-reaction sessions over the past ~3 years contributed almost nothing to the stock’s three‑year cumulative return. Across 730 trading sessions (12 earnings days, ~1.6% of sessions) the total cumulative return was ~118.30%; compounding only the earnings reaction days yields about 0.40%, and those days account for only ~0.51% of the total log‑cumulative return.
Below I decompose NFLX’s close-to-close returns (resampled from minute bars) into earnings‑reaction days and all other sessions using a log-return sum so the parts add to the whole. The full analysis, charts, and the step-by-step numbers follow.
For NFLX over the past ~3 years, how much of its total cumulative return was generated on just the handful of quarterly earnings-reaction days versus every other session? Thesis: the dozen-or-so earnings days — roughly 5% of all sessions — captured a wildly outsized share of NFLX's cumulative return while the other ~95% of days netted out close to flat, exposing the stock as a binary event-driven name rather than a steady compounder.
How this was measured
Resampled NFLX minute bars to daily closes and computed close-to-close returns over the last ~3 years (ending at the most recent available day). Identified quarterly earnings reaction days as the first trading day on-or-after NFLX_earnings.reported_date. Decomposed total cumulative return using log-returns so that total log-cumulative equals the sum of log-cumulatives from earnings days and all other sessions. Reported (i) session share, (ii) multiplicative cumulative returns for earnings-only and non-earnings-only subsets, and (iii) the share of total log-cumulative attributed to earnings days.
The key numbers
Reading the numbers
Across 730 sessions only 12 (~1.64%) were earnings-reaction days. Those 12 accounted for only ~0.40% of the simple cumulative return (and ~0.51% of log-cumulative), while the other 718 sessions produced about 117.43% of the roughly 118.30% total.
The charts
The three lines let you compare the full path to the counterfactuals: the 'All days' curve rises from 0.0253 to 1.183, the 'Earnings-only' path barely moves (starts 0.0, ends 0.004), and the 'Non-earnings-only' path captures almost the full rise (starts 0.0253, ends 1.1743). Look at how close the All-days and Non-earnings lines are at the end — the earnings-only line is essentially flat by comparison. That means most of the net cumulative gain came on non-earnings sessions, not on the dozen earnings days.
This bar chart shows share of total log-cumulative return: Earnings days are 0.0051 (≈0.51%) and Non-earn days are 0.9949 (≈99.49%). The earnings bar is a sliver next to the non-earn bar, so in additive (log) return space earnings days contribute almost nothing. In plain terms, earnings days did not drive a material share of the total log-return.
The box plots show earnings-day returns are wider but not consistently large: earnings-day range is about -9.86% to +15.48% with mean ~0.34% (n=12), while non-earn days run about -7.33% to +10.57% with mean ~0.13% (n=718). So earnings days are indeed more volatile and can produce big moves, but their average outcome is only modestly higher and they were too few to dominate cumulative performance. That supports the idea of earnings as binary risk events, but it contradicts the claim they captured a dominant share of total return.
Earnings reaction-day returns (chronological)
| date | daily_return |
|---|---|
| 2023-07-19 | -0.0888 |
| 2023-10-18 | 0.1024 |
| 2024-01-23 | 0.0996 |
| 2024-04-18 | -0.0545 |
| 2024-07-18 | -0.0113 |
| 2024-10-17 | 0.0296 |
| 2025-01-21 | 0.1548 |
| 2025-04-17 | 0.0398 |
| 2025-07-17 | -0.0009 |
| 2025-10-21 | -0.0677 |
| 2026-01-20 | -0.0633 |
| 2026-04-16 | -0.0986 |
The takeaway
No — the dozen earnings-reaction days did not drive Netflix’s three‑year return. Over the window there were 730 trading sessions and 12 earnings reaction days (about 1.6% of sessions); the total cumulative return for the window was roughly 118.30%. If you compound only the earnings reaction days you get about 0.40% total, while the non‑earnings sessions alone compound to about 117.43%. In log‑return space those 12 days account for only ~0.51% of the total log‑cumulative return (non‑earnings ~99.49%). With 730 days of data and a 12‑day earnings sample this is a clear result — the “earnings‑days‑drive‑it” thesis is not supported here. Practical takeaway: over the last ~36 months NFLX’s gains came from ordinary sessions, not a handful of quarterly reaction days.
The fine print
- Decomposition is done in log‑return space; shares are additive in logs, not in simple returns.
- Reaction day = first trading day on or after the reported date; after‑hours moves may be split between sessions.
- Window is the most recent ~36 months; a different multi‑year window could change concentration materially.
- If any earnings releases were missing from the dataset, some reaction days could be unflagged and contribution understated.