The Minutes Myth

The finding

Prediction markets are supposed to be fast. The received wisdom is that they reprice within minutes of a headline, which would mean the number on your screen already contains the news. We timed 90,633 Polymarket reactions of two cents or more, dated by when the news broke, across eight weeks from April 29 to June 26, 2026. The median took 80.1 minutes.

No category escapes it. Geopolitics, the subject the speed narrative is built on, took 79.1 minutes. Politics, the fastest major category, still took 75.1. Every major category’s median landed between 75.1 and 81.9 minutes, and once the intervals allow for reactions clustering inside 2,070 markets, five of the six cover the pooled 80.1. Not one is fast.

Prediction markets grind toward a new price. They do not snap to it.

The Minutes Myth · Vera Research

The method

A reaction is one headline paired to one market. We date it by when the news broke, not by when our pipeline wrote the row: on that basis the same records spread across months outside the eight-week window. A move counts when the market’s price change reaches two cents inside a two-hour window, and the time we report is the median minutes from the headline to the peak of that move.

Every figure is a median time, not an accuracy, a return, or a direction, and category names are only population definitions. All of them are float-corrected on one rule: the old test discarded three in ten genuine two-cent moves, and restoring them moved the pooled median from 80.5 to 80.1 and no category by more than half a minute.

Why 80 minutes is the floor

The 80 minutes is a conservative reading, and the reason is built into the measurement. Our window closes hard at 7,200 seconds, 120 minutes, so a move still climbing at the two-hour mark is recorded at whatever it had reached by then. Nothing in this dataset can report a peak later than 120 minutes, because nothing was watched past it. What we call a peak is a two-hour peak, not necessarily the true one, and the distribution is cut off rather than complete.

The censoring is not small. Of the 81,226 moves we could still see at four hours, 35.2% were larger then than at the peak we recorded, and 31.6% peaked in the final 20 minutes of the window, right against the wall. More of the distribution is cut off than a single median admits, which means the real time to a settled price runs longer than 80 minutes, not shorter. Every number in this note should be read as a floor, per category as much as pooled.

What the 80 minutes is not

It is tempting to read a number like this as a gap to exploit, a stretch of time in which the price has not yet caught up. It is not that, and the distinction is the whole point. The 80 minutes describes how long the market took to finish repricing, measured after the fact across tens of thousands of completed reactions.

It is a property of how these markets move, not a signal, not a lag anyone can trade, and not a window a reader could act inside. Knowing that a market took about 80 minutes to reach even its two-hour peak across last spring tells you nothing you can place on a single move as it happens. The open question is a different one, about which moves hold and which give themselves back, and this note does not answer it. What this note settles is narrower and sturdier: the snap does not happen.

The takeaway

1. Across 90,633 measured Polymarket reactions of two cents or more, the median move took 80.1 minutes to reach its peak (95% CI 79.1 to 81.1). Geopolitics, the category the speed story is built on, took 79.1.
2. There is no fast category. The fastest major one, politics, still took 75.1 minutes, and every major category’s median landed between 75.1 and 81.9 minutes. Once the intervals account for reactions clustering inside 2,070 markets, five of the six are consistent with the pooled median and only politics separates from it.
3. Eighty minutes is a floor. The window closes hard at 7,200 seconds, 120 minutes; of the 81,226 moves still visible at four hours, 35.2% were larger then than at the recorded peak and 31.6% peaked in the final 20 minutes. What we call a peak is a two-hour peak.
4. This describes how long the market took to finish repricing. It is a property of the market, not a lag to trade and not a window anyone could act inside.
5. Prediction markets grind toward a new price rather than snapping to it. The dataset that timed these reactions is Vera.

The claim is descriptive: measured after the fact, the typical two-cent-plus move needed more than an hour to reach even its two-hour peak, in every major category. It is a statement about how these markets absorb news, not a timing anyone can act on. Pooled and per-category figures are now on the same corrected rounding rule, and the intervals are cluster-aware rather than assuming reactions are independent.

Method and data

Built on the Vera dataset: 90,633 Polymarket reactions of two cents or more, dated by when the news broke, from April 29 to June 26, 2026. A reaction is one headline paired to one market; the time reported is the median minutes from the headline to the peak of a price move of two cents or more within a two-hour window that is hard-censored at 7,200 seconds. Pooled and per-category figures are float-corrected on the same rule, a market with no recorded move counts as a market that did not move, and the 95% intervals are cluster bootstraps over whole markets and whole news items, the wider of the two reported.

Odds shown are each market’s own price at the time Vera flagged the story, stated as fact, not advice. Nothing here is a recommendation to buy, sell, or trade any market or asset. Vera and Crypto Briefing are not registered investment advisers.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.