POLYSTRIKE

Independent counter · metadata only · est. 2026

Deleted tweets and the real Polymarket tweet count

The deleted tweets Polymarket tweet count problem is simple: a post can go up, count toward a tweet-count market, and vanish from the timeline before you ever see it. Polystrike is an independent counter that polls X every 1 second for @elonmusk and every 5 seconds for other tracked accounts — against a ~5-minute oracle cycle — and keeps deleted posts in its record instead of dropping them. Between 2026-02-22 and 2026-06-30, on a slower ~60s poll, we captured 47 short-lived posts, 40 of a type that counts under Polymarket's public settlement rules.

Polystrike home → Live Elon counter
Deleted posts captured 2026-02-22 – 2026-06-30
47
Of a counted type40
Median catch13.5s
Range10.4–21.1s
Poll vs oracle, then ~60s / ~300s

Metadata only — no tweet text is stored or shown

How Polymarket counts: XTracker on a ~5-minute cycle

FIG. 01 — SETTLEMENT ORACLE
One short-lived post, counted two waysThis illustrates how a tweet that is posted and then deleted within a single ~5-minute gap between two XTracker scrapes is exactly the kind of short-lived post that wide-interval sampling can miss. Polystrike polled about every 60 seconds during this sample and captured these posts a median of 13.5 seconds after posting, keeping them flagged as deleted. The poll is now every 1 second for Elon Musk and every 5 seconds for other tracked accounts, with a measured 11.5 second median from post to count.One short-lived post. Two ways to count it.A post can appear and vanish inside the gap between two ~5-min scrapes — the kind of tweetwide-interval sampling can miss. Polystrike polled ~5× more often.tweet postedlives ~a minutedeletedXTracker — settlement oraclesamples ~every 5 minscrapenext scrape · ~5 min laterthe post's whole life fits inside one ~5-min gapsampling can skip a post this short-livedPolystrike — real counterpolled ~every 60scaught in ~13.5scaptured, kept & flagged deletedMedian 13.5s after posting across the 47 deletions we caught (range 10.4–21.1s). We keep the row and flag it deleted; once a deletion is confirmed it stops counting toward settlement, matching the oracle.
~300s

Oracle scrape
interval

1×

Authority
settles payouts

3

Public rules
define counted types

~5m

Gaps short posts can
fall between scrapes

XTracker is Polymarket's settlement oracle: the count it records is the count that determines who wins and loses. It scrapes the timeline roughly every 5 minutes. That interval is the whole story here — if a post goes up and is deleted inside the gap between two scrapes, a ~5-minute sampler can miss it. We are describing a capability limit of any interval-based sampler, not asserting that XTracker missed any particular post. It remains the authority; we do not.

What this is not. We are an independent counter, not a settlement authority. XTracker is Polymarket's oracle — the number that decides who wins and loses. Everything below is metadata: dates, post types, counted and deleted flags, and capture latency. We never publish tweet text, and we do not and cannot prove that XTracker missed any specific post. Nothing here is trading or investment advice.

How Polystrike counts: a 1-second poll that keeps short-lived deleted posts

FIG. 02 — INDEPENDENT COUNTER
Catch latency ~60s poll, then
13.5s
median catch  //  range 10.4–21.1s  //  every one under ~21.1s
10.4s22s

Histogram of capture latency across 47 posts

Classification public rules
40 / 47
of a counted type  //  is_counted exposed on Pro tier
RepostReply

Main-feed posts, quotes and reposts count · regular replies and community reposts do not

Which post types count toward settlementBoth counters settle by the same public rules: main-feed posts, quote posts, reposts and main-feed replies count; regular replies and community reposts do not. Of the 47 deletions Polystrike caught, 40 were a counted type. Polystrike exposes this per-post classification via the is_counted field; a five-minute oracle does not publish it.The same public rules, classified as posts landWe tag every post against the settlement rules. Of the 47 deletions we caught, 40 were a counted type.COUNTSMain-feed postQuote postRepost / retweetMain-feed replyDOES NOT COUNTRegular replyCommunity repostis_countedWe expose this classification per post (Pro tier).A ~5-min oracle does not publish it at all.

Polystrike polls X every 1 second for @elonmusk and every 5 seconds for other tracked accounts — a measured 11.5s median from post to count, against the ~60s poll that produced the sample above — and applies the same public settlement rules to classify each post. When a post is deleted, we flag it is_deleted and keep it rather than dropping it — a deleted record is data, not noise. The is_counted field is exposed on the Pro tier so an agent can see exactly which captured posts qualify.

Why catching deletions matters: the real posting pace

FIG. 03 — PACE & FORECASTING
01

Capability, not a claim

These are 47 short-lived posts a ~5-minute scraper can miss. We are proving we can catch them — not claiming the oracle missed any specific one.

02

Keep, don't drop

A qualifying post that existed in the event window stays visible here after it is deleted, flagged rather than erased — though a confirmed deletion stops counting toward settlement. Keeping it flagged is what makes the pace count complete.

03

Feeds the forecast

An accurate count of deletions is the input a downstream sleep-aware pace model needs. Miss the deletions and the projected pace runs low.

The point of retention is measurement, not settlement. If deleted posts silently disappear from your record, your view of how often someone actually posts drifts below reality — and everything you build on top of that pace inherits the error. We keep the flagged record so the real posting pace reflects what was actually posted, deletions included.

The catches: 47 caught, 40 of a counted type

FIG. 04 — METADATA LOG · NO TWEET TEXT
Two clocks, two speedsPolystrike captured deleted posts a median of 13.5 seconds after posting, ranging 10.4 to 21.1 seconds. XTracker, the settlement oracle, samples on a roughly five-minute (300 second) cycle, so a post created and deleted between samples can be missed. The arc on each dial shows the fraction of one five-minute cycle.Two clocks, two speedsHow soon after a post goes live is it seen? (arc = fraction of one 5-min cycle)13.5smedian capturePOLYSTRIKE poll ~60s (then)range 10.4–21.1s · all under ~21s~5 minsampling cycleXTRACKER scrape ~300sa post between samples can be missedvs~22×tighter
DateSubjectTypeCounted type?Caught after
2026-06-30Ted CruzQuotecounts12.6s
2026-06-17Ted CruzQuotecounts11.9s
2026-06-03Ted CruzRepostcounts11.5s
2026-06-03Elon MuskRepostcounts15.9s
2026-06-03Elon MuskPostcounts15.2s
2026-06-02Elon MuskRepostcounts14.5s
2026-06-01Ted CruzPostcounts11.7s
2026-05-31Elon MuskQuotecounts12.1s
2026-05-26Ted CruzRepostcounts16s
2026-05-26Elon MuskRepostcounts14.7s
2026-05-25Elon MuskRepostcounts10.6s
2026-05-22Elon MuskRepostcounts11.5s
2026-05-18Elon MuskQuotecounts16.2s
2026-05-18Elon MuskRepostcounts12.1s
2026-05-16Elon MuskRepostcounts15.2s
2026-05-15Ted CruzQuotecounts12.2s
2026-05-15Elon MuskReplycounts10.6s
2026-05-13Elon MuskRepostcounts11.1s
2026-05-11Ted CruzPostcounts14.4s
2026-05-09Elon MuskRepostcounts13.5s
2026-05-08Elon MuskPostcounts13.7s
2026-04-30Elon MuskQuotecounts15.7s
2026-04-27Elon MuskReplydoes not count15.9s
2026-04-26Elon MuskReplydoes not count12.1s
2026-04-25Elon MuskRepostcounts12.4s
2026-04-18Elon MuskRepostcounts15.5s
2026-04-13Elon MuskPostcounts12.5s
2026-04-13Elon MuskRepostcounts11.4s
2026-04-09Elon MuskRepostcounts14.3s
2026-04-08Elon MuskPostcounts14s
2026-04-07Ted CruzQuotecounts12s
2026-04-07Elon MuskRepostcounts13s
2026-04-05Elon MuskRepostcounts13.5s
2026-04-02Elon MuskRepostcounts15.6s
2026-04-01Elon MuskQuotecounts14.6s
2026-04-01Elon MuskRepostcounts14.8s
2026-03-24Elon MuskQuotecounts14.9s
2026-03-13Elon MuskRepostcounts21.1s
2026-03-12Elon MuskRepostcounts13.9s
2026-03-09Elon MuskReplydoes not count13s
2026-03-08Elon MuskRepostcounts11s
2026-03-08Elon MuskRepostcounts13.2s
2026-03-06Elon MuskRepostcounts10.4s
2026-03-05Elon MuskRepostdoes not count12.5s
2026-02-25Elon MuskQuotedoes not count10.5s
2026-02-25Elon MuskRepostdoes not count15.1s
2026-02-22Elon MuskReplydoes not count14.2s

The full 47-post log spanning 2026-02-22 to 2026-06-30 (39 from Elon Musk, 8 from Ted Cruz). Every row is metadata only — date, subject, post type, whether that type counts under the public rules, and how many seconds after posting we captured it. No tweet text is stored or shown. "Counted" reflects post-type rules, not a claim about any settlement outcome.

FAQ

FIG. 05 — FAQPAGE SCHEMA
Do deleted tweets count on Polymarket?

Under Polymarket's public settlement rules, settlement is about whether a qualifying post existed on the timeline during the event window — not whether it still exists at settlement time. So a post that was live when the oracle scraped it can count even if it is deleted shortly after — though whether any specific deleted post was scraped in time before deletion is not something we, or you, can verify after the fact. That is why Polystrike keeps deleted posts in its record (flagged is_deleted) rather than dropping them. XTracker is the settlement authority, and we do not claim any specific deleted post did or did not settle.

How fast does Polystrike catch tweets?

Across the 47 deleted posts on this page — collected between 2026-02-22 and 2026-06-30, when we polled X about every 60 seconds — we recorded them a median of 13.5 seconds after posting, with a range of 10.4 to 21.1 seconds. We have since tightened the poll to every second for @elonmusk and every 5 seconds for other tracked accounts; measured on 2026-07-29, a post now reaches our count a median of 11.5 seconds after it goes up (p90 12.6–15.1s, p99 28.2s). The floor is X’s own timeline index, which needs about 10.8 seconds at best to list a new post, so polling harder does not move the median. That is what lets us record short-lived posts before they disappear.

Why is the real count higher than what I see now?

Two reasons. First, deletions: we keep the row for a post that existed during the event window even after it vanishes from the live timeline, so you can still see it — though once we confirm the deletion we stop counting it, matching the settlement oracle, which also drops deleted posts. Second, sampling: a ~5-minute scrape can miss a post that goes up and comes down inside the gap between two scrapes, whereas our poll — 1 second for @elonmusk, 5 seconds for other tracked accounts, a measured 11.5s median from post to count — tends to catch it. The result is a fuller picture of the subject’s real posting pace.

Is Polystrike the official Polymarket tweet count?

No. XTracker is Polymarket’s settlement oracle and the number that determines payouts. Polystrike is an independent counter that polls faster (every second for @elonmusk, every 5 seconds for other tracked accounts, a measured 11.5s median from post to count) and classifies posts against the same public settlement rules to show the real posting pace. We are not a settlement authority and we do not claim to beat Polymarket or XTracker.