FOMC announcement days are wider than ordinary days and lean toward taking out the prior day’s high. On the 232 scheduled FOMC statement days from February 4, 1994 to October 28, 2026, SPY’s high-low range averaged 1.57% against 1.29% on other sessions, and it traded above the previous day’s high 76% of the time, against 53% normally. It traded below the previous low about as often as any other day (46% vs 45%), and the upward tilt in the close has faded since 2019.
Every number below was calculated from the daily data listed under Sources. The meeting dates are parsed from the Federal Reserve Board’s own meeting calendars (2021 to 2026) and its historical materials pages (1994 to 2020), and saved in _data/fomc_dates.csv. A “statement day” is the date on the policy statement issued after a regular meeting. More studies are on the Market Stats hub.
How FOMC days compare with the days around them
The statement day has the widest range of the three days around a meeting and the biggest average gain, though the median gain is close to an ordinary day’s.
| Days | n | Avg high-low range | Median range | Avg close-to-close | Median close-to-close | % up |
|---|---|---|---|---|---|---|
| Day before FOMC | 232 | 1.18% | 0.89% | 0.10% | 0.07% | 56% |
| FOMC statement day | 232 | 1.57% | 1.31% | 0.23% | 0.10% | 56% |
| Day after FOMC | 232 | 1.43% | 1.24% | -0.06% | 0.10% | 54% |
| All other sessions | 7557 | 1.29% | 1.05% | 0.04% | 0.07% | 54% |
| Unscheduled statement days | 15 | 4.26% | 4.87% | -0.08% | -0.47% | 47% |
Range is the day’s high minus its low, as a percent of the prior close. “All other sessions” excludes the statement days, the days either side of them and the unscheduled days, so it is a clean baseline of 7,557 sessions. The 0.23% average on statement days is pulled up by a few big rallies; the median of 0.10% is only a little above normal. The 15 unscheduled days (conference calls, unscheduled meetings and notation votes that issued a statement, April 1994 to March 2020, listed in the data file) are a different animal: a median range near 5% and too few cases to say anything about direction.
Above the prior high, not much more often below the prior low
The extra range on FOMC days shows up mostly as a push through the previous day’s high.
| Days | n | Traded above prior high | Traded below prior low | Avg close location (0 = low, 100 = high) | Closed in top quarter | Closed in bottom quarter |
|---|---|---|---|---|---|---|
| Day before FOMC | 232 | 53% | 41% | 57 | 35% | 20% |
| FOMC statement day | 232 | 76% | 46% | 52 | 31% | 29% |
| Day after FOMC | 232 | 47% | 47% | 54 | 38% | 27% |
| All other sessions | 7557 | 53% | 45% | 55 | 35% | 23% |
| Unscheduled statement days | 15 | 53% | 47% | 52 | 27% | 33% |
Three in four statement days traded above the prior high, a big gap over the 53% base rate. The close does not follow it up, though. Close location measures where the close sat inside the day’s range, and FOMC days finished slightly lower in their range (52) than ordinary days (55), with more closes in the bottom quarter (29% vs 23%). Put plainly: Fed days often poke above yesterday’s high at some point, then give a fair part of it back by the bell more often than a normal day does.
The quiet day before, the wide day itself
Volatility changes over the decades, so the table below compares each day’s range with the average range of the 20 sessions before it. A ratio of 1.00 is a normal day for that period.
| Days | n | Avg ratio | Median ratio | Ratio above 1.5 | Ratio below 0.75 |
|---|---|---|---|---|---|
| Day before FOMC | 232 | 0.93 | 0.79 | 9% | 43% |
| FOMC statement day | 232 | 1.30 | 1.19 | 29% | 13% |
| Day after FOMC | 232 | 1.15 | 1.02 | 20% | 22% |
| All other sessions | 7557 | 1.01 | 0.90 | 13% | 34% |
The pattern is the clearest in this study. The day before a statement tends to be narrow (43% of them under three quarters of normal), the statement day is wide (29% of them over one and a half times normal), and the day after is still a bit wider than usual. That fits a market that waits, then reprices. If you follow the VIX around these dates, our VIX surge study covers what happens after volatility jumps.
Has the Fed-day pattern changed?
Partly. The upside poke has held up; the up-close tendency has not.
| FOMC days | n | Avg range | Avg close-to-close | % up | Above prior high | Below prior low |
|---|---|---|---|---|---|---|
| 1994-1999 | 18 | 1.74% (1.26%) | 0.35% (0.08%) | 56% (53%) | 89% (54%) | 39% (44%) |
| 2000-2010 | 88 | 1.79% (1.58%) | 0.38% (-0.01%) | 64% (53%) | 77% (51%) | 43% (48%) |
| 2011-2018, no press conference | 32 | 1.29% (0.99%) | -0.11% (0.06%) | 41% (55%) | 62% (54%) | 59% (44%) |
| 2011-2018, press conference | 32 | 1.30% (0.99%) | 0.36% (0.06%) | 59% (55%) | 88% (54%) | 34% (44%) |
| 2019-2026 (all with press conference) | 62 | 1.48% (1.19%) | 0.10% (0.08%) | 50% (56%) | 73% (56%) | 50% (43%) |
Figures in brackets are the other sessions of the same years. The 1994-1999 row is small because the Fed’s historical pages list no statement for 30 regular meetings from July 1994 to March 1999, so those meetings are left out. Press conferences start with the April 27, 2011 meeting in the Fed’s records and run at every meeting from 2019. From 2011 to 2018, meetings with a press conference behaved like the old Fed days (88% above the prior high, 59% up), while the ones without behaved worse than an ordinary day. Since 2019 statement days still range about a quarter wider than normal, but they now break the prior low half the time and close up only half the time. Small samples, all of them: 32 or 62 meetings is not much.
The 2010 setup, checked again
The original version of this page, written the morning of the November 3, 2010 meeting, looked at FOMC days that followed a strong pre-meeting session. Here is that idea re-run on SPY, split at the date it was written.
| FOMC days | n | Traded above prior high | Traded below prior low | Avg close-to-close | % up |
|---|---|---|---|---|---|
| 1994 to Nov 3, 2010, after a +0.75% day | 26 | 25 (96%) | 10 (38%) | 0.12% | 54% |
| 1994 to Nov 3, 2010, all | 105 | 83 (79%) | 44 (42%) | 0.38% | 62% |
| Nov 4, 2010 to 2026, after a +0.75% day | 21 | 17 (81%) | 5 (24%) | 0.13% | 48% |
| Nov 4, 2010 to 2026, all | 127 | 94 (74%) | 62 (49%) | 0.11% | 50% |
After 2010 the “strong day before” group still traded above the prior high more often than FOMC days overall (81% vs 74%), but 21 cases is a handful, and the edge over the plain FOMC-day rate is much smaller than it looked in the earlier years. The average FOMC-day gain also shrank from 0.38% to 0.11%. Other old calendar patterns get the same re-check in our presidential cycle study.
What daily bars can and can’t show
Free minute-by-minute history for the S&P 500 going back decades isn’t available, so this study uses daily open, high, low and close. That limits what it can say about “intraday” behaviour:
- It can show how wide the day was, whether it got above the prior high or below the prior low, and where it closed inside its range.
- It cannot show when the high or low printed. The statement now comes out at 2:00 p.m. Eastern (the October 28, 2026 release was marked “For release at 2:00 p.m. EDT”), and a daily bar cannot separate the morning from the post-statement move.
- It cannot see the press conference reaction separately from the statement reaction.
Why SPY and not the index itself: in the Yahoo ^GSPC file the script uses, the open equals the prior close on 90% of sessions from 1994 to 2004, so the index’s early opens and the ranges built from them are not real trading prices. SPY trades, so its bars are real. As a check, the script runs 2015 to 2026, when the index data is clean, on both series:
| Series, days | n | Avg range | Above prior high | Below prior low | Avg close location |
|---|---|---|---|---|---|
| SPY, FOMC days | 94 | 1.36% | 76% | 50% | 48 |
| SPY, other sessions | 2694 | 1.11% | 55% | 43% | 56 |
| S&P 500 index (^GSPC), FOMC days | 94 | 1.34% | 76% | 51% | 47 |
| S&P 500 index (^GSPC), other sessions | 2694 | 1.08% | 55% | 43% | 56 |
The two agree closely. Fed purchase days are a separate question; the POMO study looks at them. All of this describes past sessions. It isn’t a forecast for the next meeting, and none of it is a trade recommendation.
