# Who is CKCapitalxx? The record behind the conviction > CK Capital's X banner is a single word: conviction. The account posts small-cap AI and photonics theses at a dozen a day and keeps turning up in our own receipts tables, so we graded the conviction: 2,229 tracked stock posts since March, a fourteen-to-one bullish lean, and 103 graded calls at 54.4% right. Here is who they are and what the record shows. By Maya Koeva · 2026-09-02  The banner on [@CKCapitalxx](https://x.com/CKCapitalxx) is a black rectangle containing one word: "conviction." If you have spent any time this year in the corner of X where AI infrastructure meets small caps, you have scrolled past that banner. And if you have read this blog, you have met the account already: it sits in the receipts tables of six of our data stories, from [Nebius](/blog/nebius-stock-sentiment) to [AMD](/blog/amd-stock-sentiment) to [the Nvidia print](/blog/nvidia-earnings-graded), more than any other account we track. That earns the full profile treatment: who the account is, what it posts, and what happens when you grade the conviction. ## Who they are CK Capital is pseudonymous, and this profile stays within what the account states publicly. The X profile dates to May 2020 and counts 88.3K followers. The bio reads like a sector ETF that does not exist, "investor in defense, space, tech & photonics," and lists co-ownership of @CVresearch_, a research collective on X with [a Substack](https://substack.com/@cvresearch), plus a portfolio that runs live on Autopilot, a copy-trading service where subscribers can see the account's positions, entries, and exits. An account that publishes its own entry and exit prices has already accepted the premise of this page: records are for checking. The feed's signature format is the deep single-name thesis on a company most people have not heard of yet. A $2 billion company [building the first commercial photonic memory product](https://x.com/CKCapitalxx/status/2054243387331449300) with Marvell (Penguin Solutions). The French materials company [it says the entire AI photonics trade runs through](https://x.com/CKCapitalxx/status/2044989114856907187) (Soitec). A satellite play pitched from [the five billion smartphones that have all hit a dead zone](https://x.com/CKCapitalxx/status/2042990020638707886) (AST SpaceMobile). Where [TheValueist](/blog/who-is-thevalueist) argues about Nvidia with the whole timeline, CK Capital hunts two tiers down, in names where a good thread can be most of that day's conversation. ## What they post, in our data Quantral has tracked the account since March 12, 2026, the longest window of any profile in this series: 2,229 stock posts in under six months, roughly a dozen a day. More than nine in ten carry a clear direction, and those run fourteen to one bullish. Nearly every post is a take. The most-posted names, with each one's bullish-to-bearish split among the directional posts: | Ticker | Posts | Bullish : bearish | | --- | --- | --- | | ASTS (AST SpaceMobile) | 237 | 207 : 20 | | NBIS (Nebius) | 166 | 149 : 7 | | AAOI (Applied Optoelectronics) | 76 | 71 : 1 | | PENG (Penguin Solutions) | 70 | 61 : 1 | | HOOD (Robinhood) | 45 | 42 : 1 | | CRDO (Credo) | 45 | 42 : 0 | | RKLB (Rocket Lab) | 39 | 34 : 1 | | KEEL (Keel Infrastructure) | 35 | 33 : 0 | | BE (Bloom Energy) | 35 | 35 : 0 | | HLIT (Harmonic) | 34 | 32 : 0 | The table is the bio made literal: a satellite name on top, then AI infrastructure, optics, and connectivity the rest of the way down, with Robinhood as the lone generalist. Eight of the ten rows hold one bearish post or none, Bloom Energy at 35 to 0, Credo at 42 to 0: when this account covers a name, it is long, and it posts the thesis until the thesis resolves. The exception did not make the table, and it is the most informative name in the book: IREN, the account's twelfth most-posted ticker, splits 9 bullish to 19 bearish, the only net bearish name near the top. A fourteen-to-one bull who still argues the short side somewhere is reading the filings and not just the vibes. ## The graded record Threads are not a track record. On [Quantral's six-month accuracy board](/blog/most-accurate-finance-voices-2026), @CKCapitalxx ranked seventh of every account we track: 103 graded calls, 54.4% right, with a best graded call of Micron at +53% in 30 days. As always, the graded-call count is far below the post count by design: repeated posts on the same name inside a grading window collapse into a single call, so posting the ASTS thesis 237 times earns a handful of grades, not 237 of them. Seventh sounds mid-table, and it is, but the table is compressed: [@aleabitoreddit](/blog/who-is-aleabitoreddit) graded 59.5%, [@jukan05](/blog/who-is-jukan05) 58.6%, [TheValueist](/blog/who-is-thevalueist) 54.9%, and the board's number one, @citrini, 64.2%. From CK Capital to the top of the board is about one extra correct call in ten, and nobody we track escapes the error bar. The honest summary is the same one this series always lands on: a feed this confident is right a little more than half the time, which is both a real edge over a coin flip and a long way from the certainty the threads project. What the mid-table number hides is that this account changes its mind in public, which is rarer than being right. It was one of only two graded bears in [our AMD autopsy](/blog/amd-stock-sentiment), staying skeptical while X ran seventy to twenty-two bullish. It went into [SpaceX's first earnings](/blog/spacex-stock-sentiment) bearish in July and flipped bullish in August when the unlock got absorbed. And it walked into [the Nvidia print](/blog/nvidia-earnings-sentiment) split, three bullish mentions to two bearish, then came out with seven straight bullish posts and a buy alert once the numbers were on the table. The permabull posture is real, but it has hinges. ## The book being tested right now A hit rate is history; the current book is live, and late August has been rough on it. The account's biggest conviction, AST SpaceMobile at 207 to 20, closed September 1 at $55.80, less than half its late-May peak of $133.09 and about 36% below where it traded when we started tracking the account in March. Rocket Lab is modestly underwater over the same window. The AI-infrastructure names sold off hard in the last two weeks of August: Nebius closed at $199.54, about 30% off its June high, and Credo gave back almost 9% on September 1 alone. The wins are real too. Nebius is still up about 85% since March 12, and the account's 26-to-1 bullish receipts [held through the summer drawdown that our Nebius story graded](/blog/nebius-stock-sentiment). Penguin Solutions, the photonic-memory thesis from the table above, closed September 1 at $47.59, up about 168% over the window even after giving back much of its July run, and the near-unanimous credible crowd behind it is [a story we autopsied on its own](/blog/signal-autopsy-penguin-solutions). Some of the open bets will resolve for the account and some against it. The point of grading is that nobody will have to argue about which: every call lands on the record either way. ## Follow the calls, graded The small-cap corner of finance X runs on conviction, and conviction is cheap to type. [Their voice page](/voices/ckcapitalxx) has the profile and the graded highlights, and the live record, updated in the Quantral app as each call resolves, sits next to every other account we track, including the [rest of the accuracy board](/blog/most-accurate-finance-voices-2026). Check the record before you check the thread's ratio. --- *Graded-call figures are Quantral's own methodology (subject calls graded over fixed windows; see [how a track record is graded](/learn/how-a-track-record-is-graded)) as published on the board linked above. Post counts and splits cover the account's tracked stock posts from March 12 to September 1, 2026; price references use closing prices through September 1, 2026. Claims attributed to the account are its own, linked to their sources. Quantral surfaces signals and context from public sources to support your own research. Nothing here is financial advice or a recommendation to buy, sell, or follow anyone. Past performance is not indicative of future results.*