On Nimb, the map shows the 110 highest-scoring Polymarket markets per timeframe. The score rewards money traded on undecided outcomes, book depth, recent moves and near deadlines, after four gates. Each category gets room in proportion to the money traded on its undecided outcomes: at least 4 markets, at most 40% of the map.
The rule is the same in the scheduled data build and in your browser, which re-runs it every 5 minutes. It is a ranking of activity, not a recommendation: a market on the map is a good place to start reading the market, not a market worth trading. Nimb is not affiliated with Polymarket.
How does Nimb decide which markets to show?
In five steps, once for each of the 24H, 7D and 30D timeframes; the 1H and 12H views reuse the 24H selection and 3D the 7D one (see how timeframes work):
- Collect candidates. The scheduled build reads the top 2,100 active Polymarket events by 30-day volume and the top 600 by 24-hour volume from the Gamma API. An open map reads the top 300 by 24-hour volume every 5 minutes, and the top 300 by 7-day and 30-day volume every 30 minutes.
- Merge duplicates. One circle per thing: Polymarket's "More Markets" companion events are merged with their main event, and daily or weekly recurring markets (a life of 8 days or less) collapse into one series.
- Apply the gates. Markets that are closed, idle, already decided or untradable are removed.
- Score. Every remaining event gets a selection score, summed over all of its open outcomes.
- Fill the map by category. Each of the seven categories gets its quota of top scorers; any room left goes to the next best scores.
How is the selection score calculated?
The score starts from informative volume: money traded on outcomes whose price is still in doubt. A trade at 50% tells you more about an event than the same trade at 99%.
| Term | Formula | What it means |
|---|---|---|
| Informativeness | I(p) = min(1, p(1−p)/0.16) | 1 for any price from 20% to 80%; 0.30 at 5%; 0.06 at 1% |
| Informative volume | A = Σ volume × I(p) | Timeframe volume of each outcome, weighted by how undecided it is |
| Depth | D | Order-book liquidity reported by Polymarket on outcomes with I above 0.2 (priced roughly 3–97%) |
| Move | M | Volume-weighted absolute price change over the timeframe, counted up to 20 points |
| Deadline | [resolves ≤ 7 days] | 1 if an outcome's end date falls within 7 days and the market is still undecided, else 0 |
Together:
S = log10 A + 0.3·log10(1 + D) + 5·min(M, 0.20) + 0.2·[resolves ≤ 7 days]
The logarithm means each tenfold increase in informative volume adds one point, so a $10M market does not crowd out every $1M market. A 20-point move adds up to 1 point, the same as ten times the money. The deadline bonus is small: it breaks ties in favour of markets about to resolve.
Why does money alone not decide?
Two illustrative markets on the 24H view:
| Market A | Market B | |
|---|---|---|
| Price | 41% | 95% |
| 24-hour volume | $200,000 | $1,000,000 |
| I(p) | 1.00 | 0.30 |
| Informative volume A | $200,000 | $296,875 |
| log10 A | 5.30 | 5.47 |
| Depth term (D = $50,000 / $200,000) | 1.41 | 1.59 |
| Move term (6 pts / 1 pt) | 0.30 | 0.05 |
| Deadline (3 days / 40 days) | 0.20 | 0.00 |
| Score S | 7.21 | 7.11 |
Market B traded five times as much, but most of it was on an outcome the crowd already treats as nearly settled. Market A ranks higher because its price is in doubt, it moved and it resolves soon.
Which gates must a market pass?
| Gate | Rule | 24H | 7D | 30D |
|---|---|---|---|---|
| Tradable | Active, accepting orders and not closed. The end date is ignored, so a market past its deadline that still trades stays in. | Yes | Yes | Yes |
| Active | Informative volume A over the timeframe of at least | $5k | $30k | $100k |
| Not decided | At least 10% of the event's volume on outcomes priced 3–97%. For neg-risk events (one winner among many), the leader must also be below 97%. | ≥10% | ≥10% | ≥10% |
| Tradable book | The lead outcome's spread is at most 4 points, or the event has at least $10k of depth. | 4 pts or $10k | 4 pts or $10k | 4 pts or $10k |
Why are near-certain markets left out?
A market at 98% can trade millions of dollars a day, mostly from holders locking in the last 2 points before resolution. Its price says little that is new, and a large circle for it would crowd out events still in play. Take a single-outcome market at 98% with $2M of 24-hour volume: none of that volume is on an outcome priced 3–97%, so it fails the "not decided" gate.
Why isn't a big market on the map?
- It is decided or nearly so (see above).
- Its spread is wider than 4 points and its book holds less than $10k.
- Its category is already at 40% of the map, and it ranks below that category's last market.
- It is selected on another timeframe only: a long-running election may lead on 30D and miss 24H on a quiet day.
- Between builds, the browser only sees the top 300 events in each trending list.
- Its category or theme is switched off in your filters.
How many markets are shown, and how is the map balanced?
The map shows up to 110 events per timeframe, on every screen size. Fewer appear when fewer pass the gates.
Each category's room is its share of informative volume across all eligible events, multiplied by 110 and rounded, then held between a floor and a cap:
- Floor: 4 markets (or all of them, if fewer than 4 qualify), so a quiet category stays visible.
- Cap: 44 markets, 40% of 110, so one busy category cannot fill the map.
Example: if Sports holds 30% of informative volume, it gets 33 places; if Politics holds 50%, it gets 55, capped at 44, and the 11 spare places go to the best remaining scores in other categories. If rounding pushes the total above 110, the lowest-scoring market of a category above its floor is dropped until 110 remain.
Categories come from Polymarket's tags, with title rules on top: markets on gold, oil, equities, rates or IPOs go to Finance even when tagged as crypto.
How often does the selection change?
The scheduled build re-selects every 6 hours, and an open map re-runs the same rule every 5 minutes on fresh trending lists (see data sources and freshness). A market that qualifies mid-session is loaded with its price history, trades and holders and joins the map; one that falls out disappears at the next refresh. Changes wait while you have a market open or are hovering one.
Nimb does not apply hysteresis, a rule under which a market enters at rank N or better and leaves only after ranking below 1.33N twice in a row. Without it, markets near the cut-off can flicker in and out between refreshes.
What did the tests show?
Before launch, Nimb's own tests compared this rule with the previous one (the largest events by 30-day volume, with fixed per-category limits and recurring markets excluded) on a fresh Polymarket universe pulled on 2026-11-07 at 12:39 UTC: 2,100 events, $46.8M of 24-hour volume.
| Timeframe | Undecided volume | Near-certain leads | Moved ≥ 2 points |
|---|---|---|---|
| 24H | 41% → 56% | 12.5% → 1.2% | 34% → 60% |
| 7D | 46% → 63% | 16% → 0% | 66% → 83% |
| 30D | 45% → 71% | 24% → 1.2% | 63% → 79% |
"Near-certain leads" are events whose main outcome is priced below 5% or above 95%. The new rule also brought back the largest Geopolitics and Finance markets, which the old rule had dropped (one for a passed end date while still trading, one for being a recurring market).
These tests used 80 markets per timeframe and a 30% cap. The live rule uses 110 markets and a 40% cap with volume-proportional room, which has not been re-measured.
What this cannot tell you
- No hysteresis, so markets near rank 110 can swap between refreshes.
- Category labels are rule-based. Unusual titles can land in the wrong category.
- Gamma has no hourly volume, so selection runs only on 24H, 7D and 30D; 1H and 12H reuse the 24H set and 3D the 7D set.
- Depth uses Polymarket's liquidity figure, not depth measured within 2 cents of the midpoint.
- Up to 5 outcomes per event are drawn (priced 1–99%, most 7-day volume); the score and circle size use all of them.
- A high score is not an edge. It says the market is active and in doubt, not that the price is wrong.
The full list, with numbers and status, is in known limitations. The rule is also summarised on How Nimb works, and every market it selects has a page under /markets. Changes to the rule are dated in the changelog.
Last reviewed 2026-11-07 · Data from Polymarket · Not financial advice · Nimb is not affiliated with Polymarket.