Mathematical Money | November 13, 2026
The Fed raised rates on 16 Sep. It was the first hike since July 2023. It held on 28 Oct, and after the hot inflation number on 10 Nov the market is reported to be pricing another hike on 9 Dec as close to certain.
So the feed is full of "hikes are bad for stocks". I went and counted. The Fed's own table lists 52 hikes since 1990, and 51 of them are more than a year old. The S&P 500 was higher a year later after 44 of the 51. That is 86%.
That sounds bullish. It isn't. Pick any trading day since 1994 and the S&P was higher a year later 79% of the time. Stocks go up in most years, hike or no hike. So a Fed hike on its own tells you almost nothing.
But one kind of hike does look different from a normal day. September was that kind.
Sort the hike first
Hikes are not all the same event. I sort them into three kinds with two questions. You can answer both from the Fed's table in a minute.
1. What was the Fed's last move? If it was a cut, this is a first hike.
2. If the last move was a hike, how many days ago was it? Under 60 days is a back-to-back hike.
3. 60 days or more is a hike after a pause.
Fed meetings are about six weeks apart. So 60 days is the line between "hiked again at the next meeting" and "skipped at least one".
Since 1994 that gives 7 first hikes, 31 back-to-back hikes and 14 hikes after a pause. For each one I took the S&P 500 close on decision day and measured 21, 63, 126 and 252 sessions later. That is one, three, six and twelve months.
What each kind did
One year later, how many were higher, and the middle result:
Any trading day since 1994: 79% were higher. The middle result was +12.8%.
Back-to-back hikes: 28 of 31 were higher. The middle result was +12.7%.
Hikes after a pause: 11 of 14 were higher. The middle result was +14.7%.
First hikes: 5 of 6 were higher. The middle result was +5.6%.
Three months later:
Any trading day since 1994: 70% were higher. The middle result was +3.4%.
Back-to-back hikes: 20 of 31 were higher. The middle result was +2.9%.
Hikes after a pause: 11 of 14 were higher. The middle result was +2.3%.
First hikes: 1 of 6 was higher. The middle result was -4.1%.
Two of the three kinds are just a normal day. Back-to-back hikes and hikes after a pause sit on the any-day numbers at three months and at twelve.
The first hike is the one that doesn't. Five times out of six the S&P was lower three months later, and 1997 is the only exception. If three-month windows were up 70% of the time like normal, a result this bad should turn up about once in 80 tries.
It doesn't last. A year later five of the six were higher. 2022 is the one that wasn't, down 10.1%.
Where September sits
September's hike came after a run of cuts and then nine months of no change. The last move before it was the cut on 10 Dec 2025. So it is a first hike, the seventh since 1994.
The S&P closed at 7,551.81 on 16 Sep. On 12 Nov it closed at 7,592.40. That is 0.5% higher, 41 sessions in.
At the same point, five of the six earlier first hikes were already below their hike-day close. This one is not. The three-month mark is the close on Tuesday 15 Dec, six days after the next Fed decision.
So the line is 7,551.81. That's six cases, so I'm not calling it. But this is the one window where the record is really different from a normal day, and we are inside it now.
December would be a different kind
If the Fed hikes on Wednesday 9 Dec, that is 84 days after 16 Sep. It is past the 60-day line, so it counts as a hike after a pause. That kind has been a normal day: 11 of 14 higher a year later.
Four past hikes came exactly 84 days after the one before. November 1999, June 2018, December 2018 and July 2023.
A year later the S&P was down 2.6% after the first and up 4.0% after the second. It was up 27.9% after the third and up 19.5% after the fourth. Same gap, four different years. The gap told you nothing.
How much the pause definition matters
I picked 60 days. "Pause" has no official meaning, so that choice needs checking.
If a pause means 90 days or more, only 8 hikes count. Seven of the 8 were higher a year later. If it means six months or more, only 2 count, December 2016 and December 2017. One of the two was higher.
The result does not change. At 60 days and at 90 days this kind looks like a normal day, and two cases at six months is nothing.
What changes is the label. At 60 days a December hike is a hike after a pause. At 90 days it is a back-to-back hike. July 2023 is the same story: they skipped June, hiked in July, 84 days.
So if you read a stat about what happens "when the Fed resumes hiking", ask how long the pause had to be. The December hike sits on both sides of that line.
51 hikes is really six cycles
Hikes come in runs. Two hikes six weeks apart share most of the same twelve-month window, so they are not two separate results.
Seven of the 51 hikes were followed by a lower S&P a year later. Four of the seven are from 1999 and 2000. Two are from 2022. The bad results follow the cycle, not the kind of hike.
There are six finished cycles since 1994. So the honest sample is six. If you see "51 cases" anywhere, including earlier in this post, read it as six.
Does the 10-year explain it?
The 10-year Treasury closed at 5.10% on 12 Nov. On 9 Nov, the day before the inflation number, it was 4.86%.
I checked if a fast rise in the 10-year before a hike made the difference. In the three months before 16 Sep it went from 4.43% to 5.01%. Fourteen past hikes came after a rise of half a point or more. A year later the S&P was higher after 11 of those 14. That is 79%, the any-day number again. So no.
One comparison did show something. S&P 500 companies earn about $3.86 a year for every $100 of index price. The 10-year pays $5.10 on the same $100. Bonds pay more than stocks earn.
Fourteen past hikes happened like that, all between 1994 and 2000. The S&P was higher a year later after 10 of the 14. For the other 37 hikes it was 34.
But those 14 hikes are three cycles. 1994 was fine. 1997 was fine. 1999 and 2000 were not. Two against one, so I'm not calling this either. September was the first hike since May 2000 with the 10-year paying more.
I also compared the equal-weight S&P with the normal one after every hike since 2004. Nothing there.
The midterm claim from 13 Oct
On 13 Oct I wrote that the S&P was higher six months after the end of October in all 19 midterm years since 1950. This year's start line is the 30 Oct close, 7,580.07. On 12 Nov the index was 0.2% above it.
The first-hike record over six months is three higher out of six. So two records now cover the same months. One is 19 from 19 and one is a coin flip. I'll grade both at the end of April.
The hike sort, for any Fed meeting
You can run this yourself the next time the Fed moves.
1. Open the Fed's table of rate changes. It is on federalreserve.gov under Open Market Operations.
2. Sort the hike. Last move a cut means first hike. A hike under 60 days ago means back-to-back. 60 days or more means after a pause.
3. Look up what that kind did, next to what any day did. Only the gap between the two is information.
4. Count cycles, not hikes.
The base rate is the whole test. A stat that says "stocks rose 86% of the time" means nothing until you know a normal day is 79%.
What I'm watching
The 15 Dec close against 7,551.81. And the size of the December move. On 10 Nov the futures were reported at 88% for a quarter point and 11% for a half point.
The weakest part of this is the 60-day line. If you have a cleaner way to define a pause than counting days, tell me. Meetings skipped? Something in the Fed statement? I want a rule where December does not flip sides.
Hike dates and sizes are from the Federal Reserve's table of target rate changes, 1990 to now. From 2015 the table shows the day after each decision, so I moved those back one day. S&P 500 and 10-year closes are from yfinance. "Any day" is every trading day from 1994 with a finished window. The earnings figure is trailing 12-month earnings over price from multpl.com, and the latest months are estimates. October inflation was +0.4% on the month, from the BLS release of 10 Nov; the +0.3% forecast is as reported by the financial press. The futures odds are CME FedWatch as reported by Investing.com on 10 Nov. Everything else is my own arithmetic.
#Fed #rates #macro #SPX #investing
Stop guessing. Start calculating.
Live to fight another day. đ¤
Comments