Can AI predict CPI? We asked an AI the night before the September 2026 report
Can AI predict CPI? We replayed the night before the September 2026 report: Sikt hit headline inflation at 3.7% but missed core at 2.6%. Here is why.
Read the articleThe Fed held rates at 3.75–4% on October 28, 2026, with two votes for a hike. How Sikt's AI forecast and the markets scored, and the odds of a hike in December.

Short answer: no. On October 28, 2026, the Fed held its target range at 3.75–4%, in a 10–2 vote: Governor Lisa Cook and Cleveland Fed President Beth Hammack wanted a quarter-point hike (Federal Reserve). Sikt Intelligence, our AI forecaster, called the hold: its last forecast before the meeting put a hike at just 20%. The markets called it too, with more confidence, so they scored a little better. The next decision is on December 9, and the markets now expect a hike: as of Nov 22, 2026, Polymarket gives a quarter-point increase 90% and Kalshi 90%. Odds on this page update automatically every few hours.
The Fed kept rates where it put them in September. Its statement was upbeat on growth, saying "economic activity is expanding at a strong pace", and repeated that "inflation remains elevated". It gave no signal on timing: "the accumulation of more information will help the Committee determine the policy stance appropriate" (Federal Reserve).
The two dissents were the news. Cook and Hammack both preferred a quarter-point hike at this meeting, so the case for raising rates again has support inside the committee. At his press conference, Chair Kevin Warsh said politics played "no role" in the decision, six days before the midterms (Yahoo Finance, CNBC).
We tracked this decision live and promised to score every forecast, hit or miss. A hike did not happen, so each forecast is scored against "no hike" with the Brier score: the squared distance from the result, where lower is better.

| Forecast | Sikt: hike | Kalshi at the time | Sikt score | Kalshi score |
|---|---|---|---|---|
| September 29 | 68% | 68% | 0.46 | 0.46 |
| October 14, after the CPI report | 49% | 37% | 0.24 | 0.14 |
| October 16, final | 20% | 11% | 0.04 | 0.01 |
| Morning of October 28 | about 5% |
The verdict: Sikt called the hold; the market was more confident. Sikt's final forecast put a hike at 20%, so it expected no change, and its base case, a "hawkish hold" that keeps a December hike in play, is what the Fed delivered, two hike votes included. On the score, though, the market did better: after matching Sikt on September 29, it was more sure of a hold at every step, and a forecaster is judged on its probabilities, not only on the call.
Two notes for a fair reading. The September 29 and October 14 forecasts came from an earlier version of Sikt. And we had planned updates on October 21, 24 and 27 that were not run, so October 16 stands as Sikt's final forecast, twelve days before the decision; the market kept updating after that. One question proves little either way, which is why Sikt is tested on hundreds of questions, not one. See can AI predict the stock market? for how we test it on markets.
On September 16, 2026, the Federal Reserve raised its target range by a quarter point to 3.75–4%, in a unanimous 12–0 vote, and on October 28 it held it there. Its statement was blunt: "Inflation remains elevated. Today's policy action will support a timelier return to the Committee's 2 percent goal" (Federal Reserve).
The Fed's own projections point to one more step: the median official expects the policy rate to end 2026 at 4.1%, up from 3.8% in the June projections, which implies another quarter-point hike at one of the two meetings that were left, October 27–28 or December 8–9. With October a hold, that leaves December (Federal Reserve projections). The minutes of that meeting, released on October 7, confirmed the lean: most officials judged that another increase would likely be appropriate this year, while keeping every meeting open (Federal Reserve minutes).
| Outcome | Polymarket |
|---|---|
| 25 bps increase | 90% |
| No change | 6.7% |
| 50+ bps increase | 2.1% |
| 25 bps decrease | <1% |
| 50+ bps decrease | <1% |
Source: Polymarket, "Fed Decision in December?", as of Nov 22, 2026. Kalshi gives a quarter-point hike 90% and no change 8.5%. Both markets settle on the Fed's target range after the December 9 statement. For the full picture and price history, see our live Fed rate odds.
Looking further ahead, the markets have all but ruled out cuts this year: Polymarket gives 98% to no rate cuts at all in 2026, and Kalshi puts the chance of at least one cut before 2027 at 1.4%.
The October decision was a moving target. After a hot inflation reading and comments from Fed Governor Michael Barr, markets briefly made a hike in October the favorite (CNBC, September 23). Then the September jobs report came in weak: employers added just 29,000 jobs against expectations of about 84,000, and unemployment rose to 4.2% (CNBC). Traders quickly pulled back their bets on an October hike (CNBC, October 2).
On October 14, the consumer price index showed inflation rising to 3.7% a year in September, up from 3.4% in August and above the 3.6% economists expected. Prices rose 0.6% in the month. Core inflation, which leaves out food and energy, rose to 2.6%, from 2.4%. Energy did most of the damage: gasoline prices jumped as the conflict between the U.S. and Iran kept oil expensive (CNBC).
The markets reacted at once. On Kalshi, the price of an October hike jumped from about 19% to 48% within an hour of the 8:30 a.m. release, then settled near 38% by the afternoon. A week earlier, after the weak jobs report, it had been around 16%. By October 16 it had fallen back to about 12%, as the markets settled on a hike in December rather than October.
That was the Fed's dilemma in one sentence: headline inflation argued for hiking at once, while softer core inflation and a weak job market argued for waiting until December. On October 28 the committee chose to wait, with two members disagreeing.
The best-known gauge of Fed expectations is CME's FedWatch tool, which works the probabilities out from the prices of 30-day fed funds futures. Prediction markets ask the question directly: you buy "hike" or "hold" and get paid if you are right. The two usually tell the same story, but they can differ for good reasons:
When the two disagree by a lot, that gap is worth a closer look.
A Fed decision moves almost everything at once: bond yields, the dollar, mortgage rates and stock valuations. That is why professional investors increasingly treat event probabilities as an input in their own right, alongside futures and options (see how hedge funds and quant firms use prediction markets).
The market's number is one view. Sikt Intelligence is building a second: an AI superforecaster that reads the news and data that exist today, checks every source, and gives an honest probability next to the market's price, so the gap between the two is visible at a glance. Leave your email below for early access. Nothing here is financial or investment advice.
No. On October 28, 2026, the Fed held its target range at 3.75–4%, in a 10–2 vote. Governor Lisa Cook and Cleveland Fed President Beth Hammack dissented in favor of a quarter-point hike.
December 8–9, 2026, the last meeting of the year. The decision is announced at 2:00 p.m. Eastern on December 9, with new economic projections.
The markets think so. As of Nov 22, 2026, Polymarket gives a quarter-point hike 90% and no change 6.7%. The Fed's September projections pointed to one more hike in 2026, and two officials already voted for one in October.
It called the hold. On the score the market did a little better, because it was more confident. Sikt Intelligence's final forecast put a hike at 20%, against 11% on Kalshi at the same time; with no hike, the lower number scores better. The full table is above.
Very unlikely, according to the markets: Polymarket gives 98% to no cuts at all this year. The Fed raised rates in September and its officials expect to raise them again.
A tool from CME Group that converts the prices of 30-day fed funds futures into probabilities for each Fed meeting. Prediction markets such as Kalshi and Polymarket price the same decisions directly.