Gold & Stocks: What eight midterm elections did

I went back through every midterm election since 1994 and asked one question of each: when did the stock market make its low for the year, before the vote or after it?

In seven of the eight cycles, the low came before the election. In six of the eight, it came between mid-June and mid-October, which is to say in the exact window that the "they won't let it fall" argument says is protected. The election month itself was positive in six of eight, and the eleven months that followed were positive in all eight, by an average of about 13 percent. The widely cited version of this is that the S&P 500 has been higher twelve months after every midterm since 1950, nineteen for nineteen.

Read that against the argument. If politicians were holding markets up into midterms, the weeks before the vote would be the calm stretch and the weeks after would be the dangerous one. History shows the reverse. The pre-election stretch is where the lows form, and the vote is what ends them. The reason is not mysterious: uncertainty is highest before an election and gets resolved by it, and the policy support that the argument imagines arriving before the vote has, in practice, arrived after it. The Fed cut rates the day after the 2002 election. It announced QE2 the day after the 2010 election. In 1998 it cut three times, but the market had already bottomed on August 31 before the first cut.

The propping does not show up in the data because the tools available to a White House do not reach the things that move markets in a tightening cycle. Presidents have leaned on oil through the strategic reserve, as the Biden administration did through 2022, and stocks fell into October anyway. Presidents have attacked the Fed chairman in public, as President Trump did repeatedly through the autumn of 2018, and the Fed hiked in December anyway. No president sets the two-year yield, and the two-year yield is what has been setting prices since Jackson Hole.

The one exception, and why it is this year's analog

The exception is 2018, and it is worth a closer look because it is the year that looks most like this one.

In 2018 the Fed was hiking with quantitative tightening running in the background, the S&P 500 set its record on September 20, and the market walked into the election with the next hike still ahead of it. On election day the index sat within 7 percent of that record. It then fell to its low on December 24, nearly 20 percent below the October high, and the trigger was not the election. It was the Fed's December 19 hike, delivered with the message that the balance-sheet runoff was on "autopilot."

2022 had the same Fed and the opposite outcome. Four 75-basis-point hikes had already been delivered, the S&P 500 was 25 percent below its record by October, and the low came on October 12, four weeks before the vote. The variable that separated the two years was not the calendar. It was whether the market entered the vote with the tightening in front of it or behind it. In 2022 the damage was priced. In 2018 it was still coming.

This year looks like 2018. Stocks are at their prior highs, the Fed's hike is eight days away with more priced after it, and the two-year yield is at a 20-month high. So the honest answer to the "no decline before November" argument is narrower than either side wants. There is no evidence that midterms protect stocks. There is evidence that, when the Fed is still tightening and stocks enter the vote at highs, the repricing has landed at the next hawkish Fed meeting, which in 2018 fell after the election. If that is the template, the danger window for stocks is not November 3 itself. It is the FOMC on September 16, and then the one in December.

Two things would break the analog. The S&P 500 making its low before November 3 would put 2026 back in the seven-of-eight column, and a Fed pause on September 16 with dovish guidance would remove the 2018 signature entirely. Neither is on the screen today.

What it means for Gold

The midterm argument gets applied to gold in the same breath as to stocks, on the theory that a supported stock market means a supported everything. The record says two things about gold, and the second one is the one that matters for the next two months. And it’s in tune with my previous gold price forecast for September 2026.

The first is that in the cycles where policy was tightening into the vote, gold fell into the election, not after it. In 2014, gold dropped about 11 percent from Labor Day to its November 5 low, the day after the vote. In 2022, it fell about 6 percent from Labor Day to September 28 and retested that low on November 3. In 1994, it drifted lower. Only 2018 was flat, and there the low had already come in August. The weeks we are in now, between Labor Day and the vote, have been gold's weak stretch in every tightening-cycle midterm, with the September FOMC sitting in the middle of them.

The second is what happened after, and it is the last column of the table. Gold bounced after the vote in 2014, 2018, and 2022, which is where the "buy the midterm" idea comes from. But only two of those three bottoms held. In 2018 and 2022, the Fed was finishing: one hike left in December 2018 and then the January pivot, the last 75-basis-point hike in November 2022 and then the downshift. The dollar had peaked in both. In 2014, the Fed was starting: QE3 had ended six days before the vote and the first hike was thirteen months away, and the dollar was near 87 at the beginning of a run to 100. Gold's post-election bounce in 2014 carried it from $1,140 to about $1,300 by late January, and then it reversed and made a lower low at $1,046 in December 2015. The bounce was real. The bottom was not.

That is the split that matters. Gold's post-midterm rallies held when the Fed was at the end of its cycle and the dollar at the top of its own. They failed when the Fed was at the start and the dollar was too. This year, the first hike is eight days away and the USD Index broke out from a three-month low two weeks ago. On both counts, 2026 is the 2014 configuration, not the 2018 or 2022 one.

This also fits the Labor Day pattern I described yesterday. In the three tightening-cycle midterm years on that list, the weeks after Labor Day were lower or flat for gold, and in the one that most resembles this year's setup, the decline ran straight through the election and the rally that followed was the one that trapped the buyers.

What this means now

Put the two records together and the "they won't let it fall" argument inverts. The weeks before a midterm election have been the weak stretch for stocks in seven of eight cycles and for gold in every tightening cycle, and the one time stocks held up into the vote, they fell harder afterward because the Fed had not finished. Nothing in the data says that a decline waits for the election. The data says the election is, at most, a marker for when the Fed's next decision lands.

For this year, that gives me three conclusions. For stocks, the 2018 analog says the danger sits at the Fed meetings, the first of them eight days away, and the November calendar is a distraction from that. For the metals over the next two months, the record is one-sided: gold fell into the vote in every tightening cycle, and the setup now, a first hike and a fresh dollar breakout, is the one where the decline ran the longest. For the metals after the vote, the question is not the calendar but whether the Fed is finishing and the dollar is topping when November arrives. With oil near $99, a hike on the table rather than behind us, and the dollar two weeks into its breakout rather than two years into its run, that is not the picture, and a post-election bounce in that configuration has been the kind that gets sold.

The signal that would change this remains the one it has been all summer: the USD Index. A dollar that peaks with the Fed done is the 2022 signature and the bottoming signal for the metals. A dollar that is breaking out with the Fed starting is the 2014 signature, and in 2014 the metals’ low was still a year away.

The people who expect the decline to wait for November are, on the evidence, expecting the one thing that has not happened.