Reference · methodology · Polymarket tweet-count markets
How Polymarket tweet-count markets work & resolve
A Polymarket tweet-count market asks how many times someone — usually Elon Musk or Ted Cruz — will post on X inside a fixed window, and pays out on a numbered range called a bucket. This page explains the mechanics in plain English: the three counters you will see, which one actually settles a market, how a bucket wins, and when resolution happens. It is a reference, not a trading pitch — we describe how the plumbing works and leave the betting to you.
The count that resolves a market is the XTracker oracle count.
The three counters, in plain English
FIG. 01 — MEASUREMENTA tweet-count market involves three separate numbers. They usually disagree, and the disagreement is the whole reason to name them apart. Only one of the three — the XTracker counter — decides who gets paid.
| Counter | What it is | Source | Updates |
|---|---|---|---|
polymarket_xtracker_counterXTracker counter | The count Polymarket settles on — the settlement oracle. | XTracker oracle | ~5 min |
real_counterReal counter | Polystrike's independent count, polled fast to show the real posting pace. | Polystrike poll of X | ~60 sec |
polymarket_ui_counterUI counter | The number Polymarket prints on the market page — display only. | Polymarket interface | irregular |
The differences between these three are published as deltas. The settlement delta is the real counter minus the XTracker oracle — how far our fast poll is ahead of the number that settles. The display delta is the real counter minus the UI counter — how stale the page display is. Neither delta is a trading signal: the order book already prices off the settling count, so the UI lag is reference data, not an edge. You can see all three side by side on the live Elon counter, live Cruz counter, or every tracked account on all live counters.
How does Polymarket count Elon Musk's tweets?
FIG. 02 — WHAT COUNTSPolymarket counts qualifying posts on the subject's X timeline through the XTracker settlement oracle. Not every kind of post qualifies. Under the public settlement rules, main-feed posts, quote posts, reposts, and main-feed replies count, while standard in-thread replies and community reposts do not.
| Post type | Counts? | Note |
|---|---|---|
| Main-feed post | Yes | Standard timeline post |
| Quote post | Yes | Quote-tweet with commentary |
| Repost (retweet) | Yes | Plain repost of another account |
| Main-feed reply | Yes | Reply that surfaces on the main feed |
| Regular reply | No | Standard in-thread reply |
| Community repost | No | Repost inside an X Community |
Two details trip people up. First, deleted posts: a qualifying post that existed on the timeline during the window can still count even after it is removed — the full explanation, with our capture-latency data, is on the deleted tweets page. Second, content does not matter to the count: a market tallies how many posts were made, not what they said, so the words inside a post never change whether it counts. If you are curious what the subject actually posts about, the Elon word cloud and Cruz word cloud break that down separately.
How are Polymarket tweet-count markets resolved?
FIG. 03 — RESOLUTIONEach market splits the possible outcomes into numbered ranges called buckets — for example 125–149 or 150–174 posts. When the window closes, Polymarket takes the final count from the XTracker settlement oracle, and the single bucket that contains that number resolves YES while every other bucket resolves NO.
Buckets
The outcomes are a ladder of ranges. You hold a share in one bucket; it pays $1 if the final count lands inside that range and $0 otherwise.
The oracle count
At the window's close the settling number is read from the XTracker oracle — not from our real counter and not from the number printed in the Polymarket interface.
The winning bucket
Whichever bucket contains that oracle count resolves YES. Every other bucket resolves NO. There is exactly one winner per window.
This is why the three counters can diverge without anyone being wrong. Our real counter may be ahead because it polls faster; the UI counter may lag because it refreshes irregularly. But settlement runs on the oracle, so those gaps do not change the payout. For the settled record — final counts and the peak UI lag we logged per window — see the tweet market history.
What is the xtracker counter?
FIG. 04 — THE ORACLEThe xtracker counter is the tweet count Polymarket settles on. It is produced by the XTracker
settlement oracle, which reads the subject's timeline on a roughly five-minute cycle. Because it
is the authoritative number for payouts, it is the one Polystrike mirrors most carefully — our real_counter classifies posts against the same public rules so the pace we show matches
what a market will actually pay on. XTracker is the authority; we are an independent, faster observer,
not a settlement source.
Resolution timing. Every market has a fixed start and end (exposed as end_ts in the API). The number that settles is the oracle's last in-window read, so
a market resolves shortly after its window closes rather than in real time. During busy stretches
the ~5-minute cadence means the settling count can trail the live posting pace by a few minutes —
one more reason the real counter and the oracle rarely match to the second.
Glossary of Polymarket tweet-count terms
FIG. 05 — DEFINITIONS- Bucket
- A numbered outcome range in a tweet-count market — for example 125–149 or 150–174 posts. Each market offers several buckets, and the single bucket that contains the final settled count resolves YES while every other bucket resolves NO.
- XTracker counter
- The tweet count Polymarket settles on. It is produced by the XTracker settlement oracle, which reads the subject's timeline on a roughly five-minute cycle. This is the number that decides who wins a market — not the live counter and not the page's display value.
- Oracle
- The external source of truth a market settles against. For Polymarket tweet-count markets the oracle is the XTracker count; whatever value it reports at the window close is the value the market pays on.
- Settlement delta
- The gap between Polystrike’s real counter and the XTracker oracle — real_counter minus xtracker_counter. A positive value means our fast poll has already tallied posts the ~5-minute oracle has not scraped yet. It is a measurement of scrape lag, not a claim that settlement is wrong.
- Velocity
- The rate at which the subject is posting, usually expressed as posts per hour or per day. Velocity is the input a pace projection uses to estimate where the final count will land inside the market's window.
- Sleep window
- The recurring overnight stretch when the subject rarely posts. A pace projection that ignores it over-counts; a sleep-aware model discounts those hours so the projected final count stays realistic.
- Moneyness
- A descriptive term for how close a bucket sits to the projected final count — i.e. how likely that range is to contain the number that ultimately settles. It describes where a bucket stands relative to the forecast; it is not advice and not a claim of edge.
FAQ
FIG. 06 — FAQPAGE SCHEMAWhy is the Polymarket number different from the live tweet count?
Because they come from different clocks. Polystrike's real counter polls X about every 60 seconds, so it usually runs ahead of the XTracker oracle, which scrapes on a ~5-minute cycle, and ahead of Polymarket's UI counter, which refreshes irregularly and can freeze. The gaps are timing artifacts. Settlement uses the oracle count, so a market pays on the oracle's number regardless of what the live counter or the page display show.
Is the gap between the UI counter and the oracle a trading edge?
No. The gap is a display-staleness measurement, not an exploitable signal. The order book already prices off the settling count rather than the lagging page display, and Polystrike's own research found that trading the divergence comes out to roughly breakeven, not an edge. We publish the delta as reference data, not as a strategy.
Do deleted tweets count toward a Polymarket market?
Often yes. Settlement is about whether a qualifying post existed on the timeline during the window, not whether it still exists at settlement time, so a post that was live when the oracle scraped it can count even after deletion. Whether any specific deleted post was scraped in time is not verifiable after the fact. See the deleted-tweets page for the full explanation.
When does a Polymarket tweet-count market settle?
Shortly after the window closes. Each market has a fixed start and end; the count as of the end of the window is taken from the XTracker oracle, and the bucket containing that final count resolves YES. Because the oracle reads on a ~5-minute cycle, the settling number is its last in-window read.
Keep reading
RELATEDWatch the numbers live on the Elon counter and Cruz counter, or every account at once on all live counters. Dig into settled windows and past deltas in the tweet market history, see how we handle deleted tweets, or pull the same data yourself with the free tweet count API. Developers can go deeper in the core concepts reference.