A prediction market price is a continuously updated, money-weighted estimate of a probability: it changes within seconds of new information and shows what changed and when. On Nimb, you read that estimate as a chance in %, its move in points and the money behind the move, which makes Polymarket a live instrument for reading the news.
Why do journalists cite Polymarket odds?
Because a price is a live headline. A poll is a photograph taken last week; a market price is a feed that never stops. When something happens, traders with money on the outcome have a reason to act on it at once, and their orders move the price before most articles are written. A chance of 41% at 13:00 and 55% at 13:20 says that something important happened in those twenty minutes, and roughly how important traders think it is.
This is the oldest documented use of these markets. Economists studying US election nights used prediction market prices that moved through the night as results came in, and matched them with moves in financial markets over the same minutes (Snowberg, Wolfers and Zitzewitz, Quarterly Journal of Economics, 2007).
Three properties make the price useful as a news signal:
- Speed. On Nimb, prices stream live from Polymarket's CLOB WebSocket, so a move shows within seconds; volumes refresh every minute from the Gamma API.
- Money behind it. A price only moves if someone trades. Each move comes with the dollars that moved it, which you can see as money flow.
- A time stamp. The price chart records when traders changed their minds, which often dates an event more precisely than the first report.
Are prediction markets better than polls?
They answer a different question and fail in different ways. A poll asks people what they intend; a market asks traders what they expect, with money at stake. The table compares them as news sources.
| Polls | Expert forecasts | Prediction markets | |
|---|---|---|---|
| Update speed | Days to weeks per wave | Irregular, when published | Seconds, whenever someone trades |
| Incentive to be right | None for respondents; reputation for the pollster | Reputation | Money: wrong traders lose it |
| Typical sample | About 1,000 respondents chosen to represent a population | One person or a small panel | Whoever trades that market; can be a few large wallets |
| What it measures | Stated intentions or opinions today | A judgement, often with a model | The price at which traders will risk money on an outcome |
| Known biases | Non-response, house effects, likely-voter models | Overconfidence, herding, slow revision | Long-shot bias, thin books, single large holders, ambiguous rules, traders who are not a cross-section of the public |
| Covers | Elections and opinion | Anything an expert writes about | Anything with a clear rule and a source: elections, central banks, wars, prices, releases, sport |
On accuracy, the evidence is mixed rather than one-sided. A long study of the Iowa Electronic Markets found market prices closer to the final result than polls most of the time (Berg, Nelson and Rietz, International Journal of Forecasting, 2008); a reply argued that polls, once adjusted for how they typically drift before election day, do as well or better (Erikson and Wlezien, Public Opinion Quarterly, 2008). The full picture is on are prediction markets accurate?. For news, the practical point is narrower: markets react first, and the reaction is worth reading even when the level is not.
How fast do markets react to news?
In an active market, within seconds to minutes. The first trades after a report come from whoever saw it first and is willing to act; the price then settles as more traders weigh in. Nimb shows the price live, so you see the move as it happens, not at the next data refresh. The list of markets on the map refreshes every 5 minutes in the app and every 6 hours on the static markets pages.
A move can come before the headline. Sometimes that is because traders are faster than newsrooms; sometimes because someone knows more than the public. Either way, a move without an obvious cause is a reason to look for the news, not a substitute for it.
How do you read a single move?
Market reading is working out what a prediction market's price, flow and holders say about an event, and how much to trust it. For a single move, ask three questions.
- Size: how many points? Moves are measured in points: 41% to 51% is +10 points. Compare the move with what is normal for that market. Nimb's market panel flags a 24-hour move against a typical day, for example "Moved +10.2 points in 24 hours, 3.1× a typical day", followed by the line "Unusual moves often follow news: check what happened before you follow it."
- Money: how much traded? A 10-point move on $500 of trading is one person; the same move on $500k is a consensus shift. Check the volume over the move and whether buying was one-sided.
- Timing: when, and did it hold? Find the minute it started and look for news at that time. A move that holds for hours after the news is a revision; one that snaps back within minutes was often a single large order.
What does a 10-point move in a day tell you?
That traders, with money, revised the chance by a tenth in 24 hours. Nimb treats a move of 10 points or more in 24 hours as notable when at least $20k traded behind it. Below that money, a 10-point move is weak evidence on its own.
What does a large move look like in practice?
On 2026-11-07, in Polymarket's market on who will be Israel's prime minister after the next election, the chance for Gadi Eisenkot rose from 32.5% to 62.6% over 7 days, a move of +30.1 points, while the chance for Benjamin Netanyahu fell from 42.5% to 21.6%. The event's market had traded more than $30M over 30 days, so the move rested on real money. It also illustrates a trap: the market's listed end date had already passed and it was still trading, waiting for its official result. A reading of that move has to include what the rules say will settle it (see how markets resolve).
When should you not trust a market's price?
Use this checklist before you quote or act on a price. Each item weakens the signal; two or more together mean the price says little.
- Thin book. A wide spread or little depth near the price means a small order can move it several points. See spread and order book depth.
- One large holder. If one wallet holds most of one side, the price may reflect that wallet's view, or its attempt to move the price. A large trader kept Intrade's 2012 presidential prices away from other venues for weeks (Rothschild and Sethi, Journal of Prediction Markets, 2016).
- Near-certain prices. Between about 95% and 100% (or 0% and 5%), moves of a point or two mostly reflect fees, the cost of locking up money and the spread, not news. Long shots also tend to be priced too high.
- Ambiguous resolution rules. If the question could be read two ways, the price mixes the chance of the event with the chance of the reading. Read the rules before the price.
- Low volume over the move. A big move on little money is one trader, not the crowd.
- No news at the time of the move. It may be early information. It may also be noise, a mistake or a deliberate push. Wait for it to hold.
How do you cite a prediction market correctly?
Name the platform, the exact question, the chance, the time with time zone, and the source of the data. Say "traders put", not "will happen". A template:
Traders on Polymarket put X at 41% (Polymarket via Nimb, 2026-11-07 13:00 UTC).
Good practice in a sentence or two:
- Give the move with its period and in points: "up 12 points since Monday", not "up 30%".
- Add the money when it matters: "on $2.1M traded this week".
- Do not call the price a forecast by Polymarket or by Nimb. It is a market price set by traders.
- Quote the question as written. "Fed cut in December" and "Fed cut by December" are different markets.
How do you use Nimb as a news dashboard?
The map is built to show where the news is moving money. On the map, circle area follows money traded over the timeframe (from 1H to 30D; the map opens on 7D), so the stories traders are acting on are the biggest circles, and colour gives the category. Open a market and the panel gives the reading in one sentence, for example "Traders see a 41% chance that it happens by December 31, down 4 points this week", then the chart, the flow and the largest holders. A workflow that works:
- Switch to 24H (or 1H, 12H) to see what moved today; 7D or 30D for the slower stories.
- Open the largest circle in a category you follow and read the move: size, money, timing.
- Run the weak-signal checklist above.
- Look for the news at the time of the move. If you find none, treat the price as a question, not an answer.
The full method is on how to read a market.
From reading to execution
If a reading leads you to a view, the order ticket sits in the same panel: one amount in dollars, all-in, with the payout and break-even in plain words. Prediction markets are risky and you can lose your whole stake; in the United States, the United Kingdom, France and about 35 other close-only countries you can close positions but not open new ones (see eligibility).
Last reviewed 2026-11-07 · Data from Polymarket · Not financial advice · Nimb is not affiliated with Polymarket.