SPX −2% Days — what comes next

Every session SPX closed down ≥2%, and the 10 trading days that followed. Filter by weekday below — defaults to Friday.
How to read these tables

A row is triggered when the SPX cash index closes down 2% or more on a given session (close-to-close). The day-of-week filter defaults to Friday — switch it to any other weekday, or show all. Rows are newest-first.

To the left of the date, two context blocks: trailing 1M / 3M / 6M returns into the drop (≈21 / 63 / 126 trading days — was the market already weak or extended?), then the 3 prior trading days' daily returns. The drop column is the trigger day itself (its weekday is shown beside the date). To the right are the next 10 trading days — +1 is the following session, through +10. The +1 column is marked with a blue edge.

The 21-EMA @ dump pair describes the daily-21-EMA trend posture: the slope arrow is tilted to match the EMA's actual rate of change (a steeper arrow = a faster-moving EMA; green up / red down), measured heading into the day so the dump's own close doesn't distort it, and %D21 is where the dump closed relative to the 21-EMA (above = green, below = red). The footer shows the share of dumps where the EMA was still rising.

The CBOE VIX block shows the VIX close the prior day (pre), on the dump day (post), and its % change (Δ%). Level cells are tinted by how high VIX is (fear heat); Δ% is red when VIX spikes / green when it falls. VIX rises on ~97% of −2% days. You can filter by VIX regime and sort by level or spike. VIX data starts 2008.

The ES 4–5p column (blue) is the E-mini S&P futures move in the hour after the 4pm cash close (4:00pm → 5:00pm ET) — did futures bounce or keep selling once the cash session ended? ES data starts 2008, so pre-2008 rows are blank there.

A LIVE row is a selloff that is still unfolding — its forward cells are blank and fill in as the days pass. The live row is pulled from Yahoo ^GSPC / ES=F (the static files lag a few weeks); the rest is FirstRateData.

Daily table: every cell is that day's own close-to-close return. Cumulative table: the +1…+10 cells are the return measured from the drop day's close to that day's close (the 3 prior days stay daily for context).

Cells are green when positive, red when negative, with a faint heat-tint scaled to size. The bottom row is the median across all events for each column.

Day of week (drop day · default Friday)
VIX regime (dump-day close · click to toggle)
Sort rows
Exclude years (click to drop · applies to both tables & medians)
include allexclude all
Table 1 — daily returns each day
Each cell = that single day's close-to-close % return. +1 (blue edge) is the next session after the drop.
Table 2 — cumulative return since that drop day's close
+1…+10 cells = cumulative % from the drop day's close to that day's close (so +10 is the total 2-week move off the selloff). The 3 prior days remain daily returns for context.
Does the post-close futures hour predict the next days?
For the rows shown above, split by whether ES bounced (4–5pm > 0) or kept selling (≤ 0) in the hour after the 4pm cash close. Median forward move and % positive for each side. Updates with the weekday/year filters — try Friday vs all. ES data starts 2008.
positive negative +1 (next session) trailing / prior-day context LIVE still unfolding
Method & caveats

Data: FirstRateData SPX cash index, daily bars, 2000-11-27 → 2026-05-01. Returns are close-to-close.

Trigger: any session with daily return ≤ −2.0%. The day-of-week filter (default Friday) selects which drop days to show. +1 is simply the next trading session.

ES 4–5pm: E-mini S&P 500 futures, FirstRateData 1-min continuous (ratio-adjusted), measured as the 16:00 bar open → 16:59 close (4:00→5:00pm ET). Coverage 2008-01 → 2026-01; the live row uses Yahoo ES=F (16:59 open, to dodge a last-bar artifact). Because it's an ES/ES ratio, the futures' own reaction is isolated from the cash-vs-futures basis.

Why some ES cells are blank: pre-2008 (no data), genuine half-days, and pre-~2013 Fridays — back then the E-mini closed at 4:15pm ET (the weekly futures close only extended to 5pm ET around 2012–13), so there was no full 4–5pm hour to measure. Net effect: the Friday ES sample is effectively ~2013→present.

Caveats: SPX is the cash index — it does not trade the gap, so the next session's open-to-open behavior differs from SPY/ES. Trailing windows are truncated (blank) for the most recent drops where 10 forward sessions don't yet exist. This is a descriptive table, not a strategy backtest — no transaction costs, no overnight risk modeled.