SharkWater Trading  •  Data Center Desk • Insider Filing Nebius Director Keeps Buying Into the Slide: Grimme’s Plan Fills Dropped From $169.99 to $153.96 October 27, 2026 Bottom Line Up Front Nebius (Nasdaq: NBIS) director Arne Grimme bought 208 shares at $153.96 on October 26, per a Form 4 filed that day. That is the lowest of eight consecutive daily plan fills since October 15, which started at $169.99: a drop of 9.4 percent in fill price (desk calc). The plan is also sizing up on weakness. Buys ran 148 to 151 shares through October 21, then 357 shares on October 22, then 205 and 208. He now holds 7,398 direct shares. Total across the eight sessions: 1,513 shares, about $247,600 (desk calc). All of it sits under a Rule 10b5-1 plan adopted June 15, 2026. What the Filing Shows Trade date Shares Price Notional (desk calc) Source Oct 15 148 $169.99 $25,159 Prior SharkWater post (Form 4, 0001513845-26-000130) Oct 16 148 $169.99 $25,159 Same Oct 19 148 $169.99 $25,159 Form 4 0001513845-26-000132 Oct 20 151 $164.62 $24,858 Form 4 0001513845-26-000134 Oct 21 148 $167.94 $24,855 Form 4 0001513845-26-000136 Oct 22 357 $161.97 $57,823 Form 4 0001513845-26-000141 Oct 23 205 $158.98 $32,591 Same Oct 26 208 $153.96 $32,024 Same Source: SEC EDGAR Form 4 filings for Nebius Group N.V. (CIK 1513845), read 2026-10-27. Oct 15 and Oct 16 rows are carried from the October 20 SharkWater post and were not re-read today. Notional figures are shares times filed price, a desk calculation, not a market quote. The Fixed-Price Theory Is Dead Last week I flagged three straight fills at exactly $169.99 and guessed the plan might trigger at or below a fixed limit. The next five fills do not fit a single flat limit. Prices stepped down to $164.62, bounced to $167.94, then fell to $161.97, $158.98 and $153.96. The notional also changed. The first five fills each came to about $25,000. The last three came to $57,823, $32,591 and $32,024 (desk calc). The filing does not disclose the plan terms, so I cannot tell you whether this is a price ladder, a dollar target that scales, or something else. What the tape does show: the buyer sized up when the price dropped, and then stayed larger than the original $25,000 pace. The three most recent fills total 770 shares for $122,438 (desk calc). The eight-session share-weighted average is about $163.67 (desk calc), so the latest fill sits roughly 6 percent below the plan’s average cost. Every filed footnote says the same thing: 10b5-1 plan, adopted June 15, 2026. Nebius is a foreign private issuer, so the filings carry a Section 16(b) and 16(c) exemption remark. A plan that pays out more line the deeper the water gets is a trotline, not a rod. Nobody is feeling the bite. The hooks go where the schedule says. Context From Primary Sources Nebius’s own newsroom lists a press release dated October 22, 2026: a $3.5 billion financing facility with Citi. That was the same day as the 357-share fill at $161.97. Our October 23 post covered the facility. The newsroom also lists an October 21 press release on compute pricing and an October 21 blog post on a Physical AI Workbench with NVIDIA. I found no filing or release that explains the fall in the fill price from $167.94 on October 21 to $153.96 on October 26. I have no verified quote or volume for NBIS today. Source: nebius.com/newsroom, read 2026-10-27. NBIS price and volume: NOT VERIFIED, no primary-source quote available this run. The Bull Case The buyer is price-aware. Size jumped from 148 shares to 357 as the fill price fell from $167.94 to $161.97, and stayed above 200 after that. The plan survives the drawdown. Eight straight sessions of fills while the price slid 9.4 percent says the plan did not pause. The position is growing. Direct holdings went from 5,885 before the October 15 buy to 7,398 after October 26, a gain of 1,513 shares (desk calc). The Bear Case It is a pre-set plan. The June 15 adoption date means the schedule predates this slide. It carries no information about what the board knows today. The size is small. About $247,600 total (desk calc) against the 7,260 shares Director John Boynton IV sold on October 15 alone, per his Form 4 filed October 19. Falling fills are not a floor. Each lower fill means the stock kept dropping. A buyer stepping down with the price is not a support level. The SharkWater Take I am not reading this as a buy signal. A plan adopted in June that scales up as the price falls is a mechanical buyer, and mechanical buyers do not tell you where the bottom is. What I do take from it: the Citi facility news on October 22 did not stop the slide, and the plan sized up into it. The test is simple. If the next fills print well below $153.96 and the size stays near $32,000 a day, the plan is price-insensitive below this zone and tells us nothing. If size jumps again, it is a ladder and the steps are worth mapping. I have no live quote or options chain, so I take no side on NBIS today. Tight lines, SharkWater Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted and accurate as of publication. The author may hold positions in securities discussed. Do your own work.

SharkWater Trading  •  Income Desk • Mortgage REIT Orchid Island Marks Book Value Down 16.6 Percent in One Quarter. The $0.10 Dividend Did Not Move. October 26, 2026 Bottom Line Up Front Orchid Island Capital (NYSE: ORC) estimated its September 30, 2026 book value at $6.02 per share, down $1.20 from June 30, 2026. That is a 16.6 percent drop in one quarter (our arithmetic: $1.20 divided by an implied $7.22). Estimated Q3 GAAP loss was $0.90 per share and total return on equity was (12.5) percent. The company still declared a $0.10 monthly dividend, which annualizes to 19.9 percent of the new book value (our arithmetic). The market price and the premium or discount to book are NOT VERIFIED here. What the filing says The October 14, 2026 8-K exhibit lays out the preliminary quarter. Orchid declared $0.30 per share in dividends during Q3, so the $1.20 loss in book value is the part the payout did not cover. The numbers tie: a $1.20 book decline plus $0.30 of dividends on $7.22 gives the reported (12.5) percent return. Item Figure As of Estimated book value per share $6.02 Sep 30, 2026 Change from prior quarter-end -$1.20 vs Jun 30, 2026 Estimated Q3 GAAP loss per share -$0.90 Q3 2026, preliminary Net realized and unrealized losses on RMBS and derivatives about $1.17 per share Q3 2026 Total return on equity (12.5)% Q3 2026 October monthly dividend $0.10 Ex-date Oct 30, pay Nov 27, 2026 Shares outstanding 194,661,864 Sep 30, 2026 Shares outstanding 192,992,160 Oct 14, 2026 Total borrowings about $10.36 billion Sep 30, 2026 Weighted average repo rate 3.99% Sep 30, 2026 Estimated stockholders' equity about $1.2 billion Sep 30, 2026 Source: Orchid Island Capital Form 8-K exhibit 99, filed October 14, 2026 (SEC EDGAR). Figures are estimated and preliminary. Total mortgage assets are inconsistent inside the exhibit ($10.77 billion in one table, $10.67 billion in others), so we do not use them. A separate filing dated October 23, 2026 carries a September 30, 2026 period and the file name of a 10-Q. We could read the directory listing only, not the contents. The audited-style figures have not been checked against the 10-Q, so treat every number above as the October 14 estimate. Paying out water while the hull takes on more. The deck stays level for a while. The waterline does not. Rate sensitivity The exhibit models a loss in both directions: about $4.2 million for a 50 basis point decline and about $21.0 million for a 50 basis point rise, including hedges. Against roughly $1.2 billion of equity, the larger figure is about 1.75 percent of book (our arithmetic). The portfolio is hedged against a drop and still exposed to a climb. The Bull Case Book is marked, not hidden. A 16.6 percent mark in one quarter is on the table. The 8-K shows the damage instead of burying it. Share count is shrinking. Shares fell from 194.7 million to 193.0 million between September 30 and October 14, 2026. The cause is not stated in what we read. If the stock trades below book, retiring shares adds to book per share. Price is NOT VERIFIED. Rate-fall exposure is small. A 50 basis point decline costs about $4.2 million per the exhibit, a rounding error against $1.2 billion of equity. The Bear Case The dividend outruns the earnings. The quarter lost $0.90 per share and paid $0.30. Book per share cannot fund that for long. The yield is mostly capital returned. $1.20 a year on $6.02 of book is 19.9 percent. Taxable earnings and return-of-capital split for 2026 is NOT VERIFIED, and that split is the real yield story. Rate-rise exposure is five times larger. About $21.0 million for a 50 basis point rise, versus $4.2 million for a fall. A bad rates week hurts more than a good one helps. The SharkWater Take I do not chase a 19.9 percent number on a book that just lost 16.6 percent. The payout is a rate on a shrinking base, and the next mark is a rate decision away. I would want the 10-Q, the November 10 dividend announcement, the live price against book, and the return-of-capital estimate before this is worth a trade. Until then it is an income story I am watching, not one I am taking. Tight lines. SharkWater Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted and accurate as of publication. Options involve substantial risk of loss. The author may hold positions in securities discussed. Do your own work.

SharkWater Trading  •  Weekend Desk • Market Brief Dead Water: VIX 19.45 and a Watchlist With Nothing New to Read October 25, 2026 Bottom Line Up Front The Cboe VIX closed Friday, October 23, 2026 at 19.45, up 2.21 percent from 19.03. Per SEC EDGAR, none of the eight watchlist issuers we read has filed anything dated after October 22, 2026. The only recent one is a Nebius 6-K, already covered. There is no catalyst to trade into Monday from the filing tape. What the filing tape shows We pulled the SEC submissions record for each issuer on October 25, 2026. The table lists the most recent entry in each feed, not a verdict on what is coming. Ticker Latest EDGAR entry Form RKLB 2026-10-02 Form 4 NBIS 2026-10-22 6-K NVDA 2026-10-13 Form type not read MU 2026-10-15 Form type not read IREN 2026-10-13 Proxy materials OKLO 2026-09-11 424B5 / 8-K WULF 2026-09-16 Form type not read ASTS 2026-09-01 Form 4 Source: SEC EDGAR submissions data per issuer, read October 25, 2026. Form types missing where the read did not return them. One flag. The Rocket Lab feed ends at a Form 4 dated October 2, 2026. The company has issued press releases since, and a Neutron timing 8-K we noted earlier never appeared. Either the feed lags or there was no filing. We could not tell which, and a quiet feed is not proof of a quiet company. A flat calm tells you the wind is gone for now. It does not tell you the forecast. VIX at 19.45 A VIX near 19 is below 20 and prices a normal week, not a scare. Index volatility is not where the risk on this list lives. It lives in the single-name stories. We did not pull option chains for this brief. Anyone building a trade needs live quotes from their broker. The Bull Case No new dilution paper. No 424B or S-1 appeared in the feeds we read since Friday. Volatility is contained. VIX under 20 leaves room for a risk-on open if earnings season delivers. Cadence continues. Rocket Lab set its Q3 report for November 9, 2026 after the close, per its release. The Bear Case Absence of filings is not absence of risk. Feeds lag, and the Rocket Lab gap shows it. VIX rose on the day. A 2.21 percent uptick into the weekend is small, but it is the wrong direction. Coverage was partial. Newsrooms were blocked or errored, and half the watchlist was not re-read. The SharkWater Take I am not writing a thesis out of an empty tape. There is no filing worth trading and no setup worth forcing on a Sunday. I will wait for Monday's session to show where volume goes, then read the paper behind it. Tight lines. SharkWater Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted and accurate as of publication. Options involve substantial risk of loss. The author may hold positions in securities discussed. Do your own work.

SharkWater Trading • Space Desk • Rocket Lab Cadence Rocket Lab Flies Electron No. 98 and Sets the Q3 Date October 24, 2026 Bottom Line Up Front Rocket Lab (Nasdaq: RKLB) launched its 98th Electron on October 23, 2026, the 19th of the year, for HawkEye 360 from New Zealand. The company says that moves it closer to a record 100 Electron launches. Third quarter results land after the close on Monday, November 9, 2026, with the call at 5:00 p.m. Eastern. I found no primary-source price or volume data for this run, so I am not telling you how the stock reacted. What the Company Said The mission, named "We're Not Hawking Around," lifted three HAWK satellites to a 520 km low Earth orbit from Launch Complex 1 on October 23, 2026 at 8:11 p.m. New Zealand Daylight Time. It was the third Electron launch dedicated to HawkEye 360. Source: Rocket Lab press release, Mahia, New Zealand, October 23, 2026. Two other items from the company's own newsroom frame the next two weeks. On October 14, 2026 it published an update that adjusts the Neutron target launch timing. On September 30, 2026 it announced a 20-launch Electron contract with Synspective, which it called its largest-ever Electron commercial deal. Item Figure Date Electron launches, all-time 98 Oct 23, 2026 Electron launches, 2026 year to date 19 Oct 23, 2026 Synspective contract, launches 20 Sep 30, 2026 Q3 results release After close, call 5:00 p.m. ET Nov 9, 2026 Source: Rocket Lab investor relations news releases dated Sep 30, Oct 14, Oct 20 and Oct 23, 2026. No SEC filing was read for these items. The Filing Gap The EDGAR submissions feed for Rocket Lab Corp showed its latest filing as a Form 4 dated October 2, 2026. Earlier scans noted an 8-K tied to the October 14 Neutron update that never appeared in that feed. Press releases are not filings. Until the 8-K shows up, the Neutron timing language exists only in the company's own newsroom. A fishing boat that makes a run every day is easy to count. The question on November 9 is what the boat is hauling and how much it costs to run. The Bull Case Cadence is real. Nineteen launches in under ten months is a measurable operating record, not a slide. Contracted demand. A 20-launch Synspective deal, announced September 30, 2026, gives the manifest repeat customers. A dated catalyst. November 9 is on the calendar, so the numbers arrive on a schedule. The Bear Case Electron is not the thesis. A launch count says little about Neutron, and Neutron timing was adjusted on October 14, 2026. No price evidence. I could not source a primary price or volume reading, so I cannot say the market is rewarding the cadence. Unfiled disclosure. The feed gap above means one company statement lacks the legal weight of a filing. The SharkWater Take Electron launch 98 is good news and old news. I am not treating it as a setup. The stock trades on Neutron and the Iridium deal, and neither moved on this launch. I am waiting for November 9 and the 8-K gap to close before I write anything with a thesis in it. Tight lines. SharkWater Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted and accurate as of the dates given. The author may hold positions in securities discussed. Do your own work.

SharkWater Trading  •  Data Center Desk • NBIS Financing Nebius Signs a $3.5 Billion Citi Bridge Loan Secured on Substantially All Assets. The AI Cloud Is Borrowing Short to Build Long. October 23, 2026 Bottom Line Up Front Nebius Group (Nasdaq: NBIS) disclosed on October 22, 2026 a $3.5 billion senior secured term loan from Citi, signed October 17 and maturing October 16, 2027. The rate starts at Term SOFR plus 2.75 percent and steps up 0.25 points every 90 days to a cap of 3.50 percent. Collateral is substantially all assets of each borrower, including the shares of the US and Dutch borrowers. Nebius cites more than $40 billion of investment-grade customer commitments as the base for a later asset-backed deal. This is a one-year bridge, not permanent capital, and the step-up is a clock. I have no primary-source price or volume for NBIS today, so the market reaction is not covered here. What the Filing Says The 6-K is signed by the company secretary and dated October 22, 2026. Exhibit 99.1 is the press release. The terms below are from the 6-K itself. Item Term (per 6-K dated 2026-10-22) Facility $3.5B senior secured term loan, Citibank, N.A. lead arranger and lender Agreement date 2026-10-17 (bridge facility agreement) Maturity 2027-10-16 (364 days) Interest Term SOFR + 2.75%, rising 0.25% every 90 days, capped at 3.50% Prepayment Voluntary, no premium or penalty Collateral Substantially all assets of each borrower, incl. shares in the US and Dutch borrowers Covenant Minimum liquidity requirement, plus customary defaults Stated use GPU infrastructure build-out, data centers, general corporate Backlog cited More than $40B of long-term, investment-grade customer commitments Source: Nebius Group N.V. Form 6-K and Exhibit 99.1, sec.gov, filed 2026-10-22 (accession 0001104659-26-117978). No figure here comes from an aggregator. Why a Bridge Matters A 364-day loan is not a funding plan. It buys time. Nebius says the facility gives it more flexibility over the timing and structure of future capital raises, and points to the contracted backlog as the collateral story for an asset-backed financing. That tells you the permanent solution is still ahead. The step-up margin is built to push the borrower toward refinancing before the cap is reached. Prepayment carries no penalty, so a take-out can happen at any point. The minimum liquidity covenant is the term to watch. The 6-K does not state the dollar threshold in the summary I could read, so the amount is NOT VERIFIED. A bridge loan is a rope ladder thrown across a gap. It gets you over, but nobody builds a house on it. Context on the insider tape: the EDGAR submissions feed for Nebius also lists Form 4 filings dated October 21, 20, 19 and 16, 2026. I could not open the October 21 filing (accession 0001513845-26-000136), so who filed it and in which direction is NOT VERIFIED. The Bull Case Cheap for a secured loan. Starting spread of 2.75 points over SOFR, with prepayment allowed at no cost. Backlog is the anchor. Management cites more than $40 billion of investment-grade commitments behind the build. Flexibility. Near-term cash for GPUs and data centers lets Nebius pick the timing of its next raise. The Bear Case Short fuse. The loan matures October 16, 2027, and the margin steps up every 90 days. Everything is pledged. Substantially all assets of each borrower back the loan, which narrows room for later secured debt. Covenant risk. A minimum liquidity test on a fast-burning build is the tripwire. The threshold is not verified. The SharkWater Take I read this as a company that needs cash to honor the backlog and chose speed over price. That is a reasonable choice, and the disclosed terms are not distressed. It still moves Nebius from asking the equity market for money to owing a bank a full repayment inside a year. I am not taking a side on the stock off this filing. The setup I would want is the take-out: an asset-backed deal that replaces the bridge at a fixed, longer term. Until that shows up, this is a watch item, and the open questions are the liquidity covenant amount and the October 21 Form 4. Execution notes: none. No options chains were pulled and no premiums are quoted. VIX read 19.78 against a prior close of 19.03 (+0.75) on the Cboe page, timestamp displayed 12:15 PM on 2026-10-23, delayed at least 20 minutes, time zone not stated. Tight lines, SharkWater Educational and informational purposes only. Not personalized investment advice. All figures sourced as noted and accurate as of publication. The author may hold positions in securities discussed. Do your own work.