Polymarket Flipped on Rocket Lab and Never Flipped Back
Rocket Lab announced an $8 billion deal for Iridium on June 29 and every feed we track leaned the same way at once. Then the stock gave back 24% in eight trading days and the crowds stopped agreeing. Only one of them repriced and stayed there.
The short version
- July 6: Polymarket collapsed from a 96% peak to 28% bullish on the first down day. X printed 57%; Reddit sat in its usual neutral band
- July 6 to 13: six straight down days, about $14 billion of market value gone. Polymarket closed below 50% bullish on every session of the slide
- July 15: X finally gave up: 21% bullish on its biggest mention day of the window
- The honest part: nobody front-ran the deal itself; daily up-or-down markets partly confirm the tape rather than predict it
Deal day: four crowds, one opinion
On Monday, June 29, Rocket Lab announced an agreement to acquire Iridium Communications for $54 a share in cash and stock, roughly $8 billion, pitching itself as an end-to-end space company: launch, satellites and now a global communications constellation. The stock closed up 16% at $98.01.
All four feeds leaned the same way at once, which is rarer than it sounds:
| Feed, June 29 | Activity | Read |
|---|---|---|
| Polymarket | 267 trades, its busiest day of the window | Leaned bullish: +0.50, 56% bullish |
| 287 mentions, more than four times the prior day | +0.11, near the top of its narrow band; the real signal was volume | |
| X | 603 mentions | 68% bullish |
| News | 46 pieces | 91% bullish |
The euphoria compounded for three sessions. By July 1 Polymarket's daily markets printed 96% bullish and the stock crossed $100. The session before the announcement weekend, Polymarket had been outright bearish on Rocket Lab: 10% bullish on June 26. Nobody front-ran this deal, on any feed. What the rest of the window measures is how fast each crowd changed its mind.
July 6: one crowd changes its mind
Monday, July 6, was the first down day: Rocket Lab fell 7.3% to $93.09 as the whole space complex pulled back from its post-SpaceX-IPO run. An $8 billion cash-and-stock deal also brings its own mechanical sellers, merger-arb desks shorting the acquirer against the stock component and plain dilution math on the share issuance. Beck's trust happened to be selling the same week, 3.28 million shares over July 6 to 8 under a 10b5-1 plan adopted in March, but the Form 4 confirming those sales landed only after the drop; at well under 1% of shares outstanding it reads as context rather than catalyst. The sentiment data cannot separate these drivers; it can only show who repriced when.
Here is how differently the crowds absorbed the same tape:
RKLB close (top) against the bullish share per source (bottom), trading days July 1 to 15, 2026. Polymarket: 24-hour prediction-market flow. X and Reddit: share of bullish mentions. Weekends and the July 3 holiday excluded; Polymarket weekend volume drops as low as 5 trades.
Polymarket went from 84% bullish on July 2 to 28% on July 6, on 99 trades. X barely moved that day: 57% bullish, with the mention stream still full of deal commentary. Reddit sat where Reddit had been the whole time, near zero on sentiment with a bullish share inside its usual 22 to 43 band. X did fall below 50% on July 7 and 8, at 45% and 47%, then went back above 60% while the stock made new lows. Polymarket never went back.
An eight-day argument with the tape
The unusual thing about this window is persistence: the crowds disagreed for more than a week while the stock kept falling.
Through July 13 Rocket Lab logged six straight down days, a 23.6% cumulative slide that erased about $14 billion in market value. Polymarket's bullish share across those days: 28, 24, 29, 33, 13, 17. X's across the same days: 57, 45, 47, 62, 61, 51. Bullish share measures what a crowd says, not what it holds. Still, on July 9 and 10, with the stock making new lows for the move, X was back above 60% bullish while Polymarket printed its most bearish reading of the slide at 13%. News ran 69% to 100% bullish on single-digit article counts through July 13; a sample that thin cannot flip.
The last day of the window is July 15. The stock broke to a fresh closing low at $76.20 and X finally cracked: 21% bullish on 1,484 mentions, its single biggest attention day of the entire three-week window. That was nine calendar days, seven sessions, after the real-money feed repriced. Reddit never flipped in either direction: across all 21 days its sentiment stayed inside a narrow band around zero, tracking attention rather than taking a side.
On July 14 the stock bounced 2.7% and Polymarket's daily read swung back to 47% bullish, still below the line, before falling again the next day. Daily up-or-down markets are twitchy by construction. The signal was which side of the argument each feed kept returning to, not any single day's print.
What this shows and what it does not
- Supported: Polymarket flipped bearish on day one of the slide and never flipped back. On a strict below-50% rule it closed bearish on all eight sessions; X did on four, twice early before swinging back above 60%, then on the last two. Reddit never took a side
- Not supported: "Polymarket predicted the crash." Its daily up-or-down markets settle on same-day price action, so part of the July 6 bearish print is the tape itself. The defensible claim is about persistence, staying bearish while social sentiment kept fading the move
- Not supported: anticipation of the deal. Polymarket was 10% bullish one session before a 16% pop. No feed saw the announcement coming
- Caveat: thin sessions are noise. Polymarket's RKLB volume ran 5 to 267 trades a day against up to 1,484 X mentions; weekend trade counts drop as low as 5 and the daily feed counts trades without sizing them. Single-day prediction-market swings on quiet days deserve no weight
- Caveat: these are end-of-day aggregates. The July 6 flip was observable at that day's close at the earliest, after the first -7.3% was gone. From the July 7 open the remaining slide still ran about 18%
- Caveat: one episode, chosen because it is clean. We have not measured how often Polymarket-vs-social divergences resolve this way. The window also ends at the July 15 low, so whether any feed was right about what came next sits outside the data
Where every number comes from
Daily series for RKLB over the 21 days ending July 16, 2026 (UTC days; the chart shows trading days July 1 to 15), pulled July 16 from the four Adanos stock sentiment APIs:
GET /polymarket/stocks/v1/stock/RKLB?from=2026-06-26&to=2026-07-16 # trades, sentiment, bullish % GET /reddit/stocks/v1/stock/RKLB?from=2026-06-26&to=2026-07-16 # mentions, sentiment, bullish % GET /x/stocks/v1/stock/RKLB?from=2026-06-26&to=2026-07-16 # mentions, sentiment, bullish % GET /news/stocks/v1/stock/RKLB?from=2026-06-26&to=2026-07-16 # articles, bullish %
Polymarket sentiment is derived from YES-price evidence per market, weighted by spread, depth, liquidity and trade confidence, then aggregated per ticker; the bullish share is the share of markets pricing bullish in that day's flow. Reddit and X bullish shares are the share of mentions classified bullish by the finance-tuned VADER plus RoBERTa ensemble. Prices are Alpha Vantage daily closes; the +16%, -7.3% and -23.6% moves are close-to-close. Insider-sale facts are from the SEC filings linked below, not from press summaries. All feeds update hourly; a free key covers the calls used here.
FAQ
Can you use Polymarket to track sentiment during breaking news?
Yes, with the right expectations. Polymarket's own market prices move in real time; the aggregated per-stock feed used in this piece updates hourly and the series shown here are daily closes, so the finest claim this data supports is which day each crowd flipped. For the minutes around a headline you need Polymarket's own real-time channels.
Did Polymarket predict the Rocket Lab crash?
No. The data does not support that framing. Polymarket showed no anticipation of the Iridium deal (it was 10% bullish the session before a 16% pop) and its daily up-or-down markets partly confirm same-day price action mechanically. What the data does show: Polymarket repriced to bearish on July 6 and stayed there through the slide, while X's mood cracked only on July 15 and Reddit never flipped at all.
Is prediction-market sentiment better than social sentiment?
It answers a different question. Social feeds measure attention and mood at scale, 1,484 X mentions on July 15 against 64 Polymarket trades; prediction markets measure what a smaller crowd will back with money. In this episode the real-money read was on the right side of an eight-day slide that social sentiment kept fading, but on thin days a handful of trades can swing the Polymarket number.
What data is this based on?
Daily RKLB series from the Adanos Polymarket, Reddit, X and news stock sentiment APIs over a 21-day window ending July 16, 2026, share prices from Alpha Vantage daily closes and the SEC Form 4 filing for the insider sale. Every number in the piece comes from those feeds; the exact API calls are listed in the methodology.
Sources
Sentiment and trade figures are Adanos API output pulled July 16, 2026; the raw calls are listed in the methodology. External sources:
- SpaceNews (June 29, 2026): Rocket Lab to acquire Iridium for $54 a share, roughly $8 billion.
- SEC Form 4 via StockTitan: Beck family trust sells 3,275,779 RKLB shares July 6 to 8 under the March 27 10b5-1 plan.
- SEC Schedule 13D/A via StockTitan: the March 27, 2026 trading plan, up to five million shares, expiring July 8, 2026.
- Trefis (July 14, 2026): six straight down days, 23.6% cumulative loss, about $14 billion of market value erased.