Grading half my election scorecard, plus Fed day, to see how much prediction market odds have reduced uncertainty so far.

Last Wednesday, the Fed held rates, just as Kalshi said it would. The day before, the hold was about a 94% favorite. A week before, SPX options had priced Fed day at about 1.37 normal days, meaning 1.37 times what they charge for an ordinary day: a move of roughly 0.9% in either direction. SPY closed up 0.34%.

From just before the 2:00 PM Eastern decision to the end of the press conference, the S&P moved 0.01%. The ancient adage held: the decision giveth nothing, and the press conference taketh it away.

So options charged for about 0.9% and got about a third of it. That’s the pattern I found in my Fed piece: in calm years, Fed day has delivered less than options charged for it, with or without Kalshi. I said I’d report back on how the day actually moved, and there it is.

One quiet day doesn’t prove much, though. Most calm-era Fed days have come in well under what options charged, but options are also paying for the occasional December 2024, when a cut Kalshi had at about 95% came with a 3% drop in the S&P.

It’s also a useful way into the election. In my election scorecard, I wrote down four checks, before any results, for whether prediction markets have made election night less uncertain, using the same measuring stick: what SPX options charge for election night as a multiple of a normal day. Two of those checks finished before the vote, so here’s how they came out.

Check 1 passed

The idea was simple. If options take their cue from the odds, the election-night premium should shrink as the surprise outcome gets less likely. The surprise here is a Republican sweep of Congress, which Kalshi had at 9.25% on October 20. I drew a curve of where the premium should sit at any given odds, and said that if the last three readings before Election Day landed within 0.15 of it, the premium was following the odds.

The odds did move. The sweep fell from about 8% on Thursday to under 6% on Friday, on heavy volume, and sat there Monday. Here’s where the curve put the premium at each day’s odds, against what options charged:

• Thursday, October 29: sweep at 8.15%, curve 1.58, options 1.49
• Friday, October 30: sweep at 5.55%, curve 1.41, options 1.20
• Monday, November 2: sweep at 5.85%, curve 1.43, options 1.42

The average gap is 0.10, inside the line, so Check 1 passes.Friday is the odd one out. Stocks sold off on big tech earnings and AI spending worries, with SPY down 1.77%, and options marked up an ordinary day more than they marked up election night. By Monday that had washed out, and the premium landed right on the curve.

So in the final week, options did what they’d do if they were taking their cue from the odds. That’s a point against my October 21 call, and I’ll take it. The longer record is less tidy, which I’ll get to below.

Check 3 failed

2022 was the last midterm before Kalshi’s congressional markets, so it’s the closest thing to a control group. Over its final five readings, 2022 priced election night at an average of 1.57 normal days. I said 2026 had to average 1.2 or less over the same stretch to count as clearly below.

It averaged 1.55. By Friday it was already out of reach, since Monday would have needed a negative reading to get there. 2026 isn’t below 2022; it’s level with it.

That’s the check that asks the series question most directly, and the answer is that options are charging about what they charged before anyone could trade the House.

Following the odds isn’t the same as listening to them

A premium that tracks the odds for a week fits two stories. Options could be taking their cue from Kalshi, or both could be reading the same things and landing in the same place. The longer record leans toward the second.

Back to the Fed. Across the 15 meetings since December 2024, options charged about 1.3 normal days even when Kalshi had the decision as a near-lock. A market taking its cue from the odds would price a 94% favorite close to an ordinary day. Options charge for the day, not for the chance of an upset.

2018 shows why that’s sensible. The expected result happened, Democrats took the House, and the S&P still rallied 2.1% the next day. Knowing who wins doesn’t tell you how stocks will take it. Prediction markets price what happens. Options price what it means for stocks, and nobody lists odds on that.

The desks that set SPX prices also have their own inputs: volatility models, how past election nights moved, and what their customers want to hedge. Kalshi’s odds are built largely from the same polls and forecasts those desks already read. And the odds are set by a fairly small group. Bloomberg reported this weekend that some of the biggest election traders took field trips to Maine and Los Angeles and pooled early-vote data looking for an edge. That’s real work, but it’s work on who wins, which is the part options don’t seem to need.

Do options disagree with the odds?

You can also run my curve backward and ask what sweep odds the premium implies. Let’s assume options and Kalshi agreed on October 20, where I anchored the curve. The answer depends on how big a sweep’s move would be, which options can’t tell you on their own. I’m assuming it grows and shrinks with ordinary volatility, the same assumption the multiple makes.

In August, Kalshi had the sweep at about 15%, and options priced it at under 6%. In late September and early October it flipped, with options at about 9% to 13% against Kalshi’s 8%. By Monday they’d met at about 6%. (The math can’t say which side is the longshot, only how close the call is, so 6% here means a 6/94 call either way.)

It’s also an “as if” number. It treats the whole premium as being about the sweep, when some of it pays for a 2018-style move on an expected result. Even so, a view that swings from one side of the odds to the other over two months looks less like a disagreement than like two markets keeping their own counsel. They happened to agree on the eve of the vote.

What Wednesday has to show

Two checks from my original scorecard are left. Both apply unless Republicans keep both chambers, which Kalshi has at about 6%.

Check 2 says the expected result should make for an ordinary day: a move of less than 1.0% from Tuesday’s close to Wednesday’s. 2018 started from the same place and moved 2.1%.

Check 4 says the sectors should split, not the index. The S&P moves less than 1.0%, and at least 3 of the 11 sector funds move the other way by more than 0.5%. That’s what an election the odds had fully priced should look like, a quiet index with a reshuffle underneath. Neither 2018 nor 2022 had one.

I’ll grade both next week.

Where I land

Fed day moved about a third of what options charged. Check 1 passed, with the premium following the odds through the final week. Check 3 failed, with 2026 pricing election night about the same as 2022. And over the longer run, options and the odds have mostly gone their own ways.

So my call from two weeks ago, “not visibly,” is holding up on the question that matters most, since election night costs no less than it did before anyone could trade the House. It just isn’t holding up as neatly as I said it would. Same caveats as before: it’s one election and one comparison year, and 2022 had its own noise. What would it take for you to trust the odds over the options market on election night?