1/ Metaplanet can advertise 14% in the US and pay 7% It isn't an accounting trick It's a Tokyo–Hong Kong–New York structure assembled in 6 months Here's how it works step by step, and why the problem doesn't land on them... It lands on Strategy and Strive 👇 This is a hypothetical scenario, not a description of a transaction Metaplanet has announced The point is to understand how the machinery could work if Metaplanet eventually connected its Japan funding, Hong Kong asset manager and U.S. Bitcoin treasury vehicle

1/3 Don't ask for the 319 million back. Ask for the part that failed the accretion test, and put it where it has to be earned A four-window test identifies 138.939 million potential shares that can move into Metaplanet's existing long-term incentive vehicle without increasing the pool, without creating a single new right, and without touching anything already exercised. The hurdle is a policy choice, and the same four windows fail anywhere from 1.3x to 2x Three terms do a lot of work below, so they need to be precise. Potential shares means the shares underlying the 10th Series rights, which do not exist as stock until a right is exercised. Issued shares are stock in existence. Fully diluted is the company's own reported count, the one in its BTC-per-share KPI, which adds convertibles and options to issued shares but leaves moving strike warrants out until they are exercised and paid for [1][2] The demand circulating is that management return the entire 319,464,000 share option pool. It will not happen, and pressing it mostly guarantees nothing happens. There is a version that could work: the conclusion survives two defensible share-count bases, it touches nothing that has already been exercised, and it fits inside a plan Metaplanet announced three weeks ago and is finalising this month That second point matters more than it sounds. On 28 August 2026 the CEO exercised 92,000 units into 64,032,000 shares, which are now issued stock locked until 2031 [5]. Nothing here asks for those back. The proposal touches neither those 64,032,000 shares nor the board-approval mechanics that already govern any transfer of the remaining rights [2]. What follows operates only on the 367,000 units that remain unexercised The test. The adjustment clause increased the number of potential shares available to the holders as the company's fully diluted count grew. The company's stated purpose was to stop the allottees losing their long-term incentive merely because their ownership ratio fell as capital was raised, and to keep them committed to medium and long-term corporate value: a fixed option grant, in a turnaround where repeated large financings were unavoidable, would have diluted the incentive in proportion to how much rebuilding capital the company raised [2]. The problem is that it increased that entitlement as the share base grew whether or not the issuance driving the growth was sufficiently accretive to justify the increase. So the question is not how many shares it created, it is how many it created during issuance windows that actually increased bitcoin backing per share by enough to justify them The measure is an observed bitcoin-backing accretion multiple: how much additional bitcoin backing per new share appeared during an issuance window, relative to the backing existing shares already carried. It is an attribution test, not a financing-causality test, and the difference matters later. It also measures bitcoin backing per share and nothing else. It is not a valuation of the company The formulas. Everything runs on four quantities: S, the shares outstanding before a raise; B, the bitcoin held before it; n, the new shares issued; and ΔB, the bitcoin added Backing per share is b = B / S Dilution is d = n / (S + n), new shares over the ending count Accretion yield is y = (b after / b before) − 1, the change in backing per share across the window And the accretion multiple compares the two: bitcoin added over the window per new share, against what each existing share already had m = (ΔB / n) ÷ (B / S) Those last three are not independent. Substituting and cancelling gives an exact identity: y = d × (m − 1) which rearranges to the form used in the tables: m = 1 + (y ÷ d) No valuation model. No market data. Once the window and the share-count basis are fixed, the multiple is two quantities and one division The identity is only exact if y and d are built on the same share count. That constraint is what makes the next section necessary, because there are two defensible share counts and they give very different multiples What m = 1 means is arithmetic, not judgement. At 1, each new share arrives carrying exactly the same bitcoin backing an existing share already had, and existing holders receive no accretion. The neutral point sits there by construction Where to set the bar above 1 is a judgement, and I set it at 2. At 1 a new share merely carries its own weight. Requiring 2 means it must arrive with twice the backing an existing share already had. That is deliberately demanding, and the reason is that a pool sized to reward growth should not pay out on issuance that merely avoided going backwards It is a policy choice rather than an identity, and I am not going to pretend otherwise. What makes it safe to argue about is that I do not need to win the argument: the same four windows fail at 1.3x, at 1.5x, at 1.75x and at 2x, and the total that comes out is identical at all four. The sensitivity table below shows the whole range The minting rule, and why it is the company's own arithmetic rather than mine. The adjustment clause set the pool at 20% of a fully diluted total. On 26 August 2025 a second and tighter constraint was added on top of it: every holder gave an undertaking capping exercise so that the post-exercise dilution ratio, measured against actual issued shares, could not exceed 20% [2] That cap is worth writing out. If P is the pool and S the issued shares, post-exercise dilution is P / (S + P). Setting that to 20% gives P = 0.25 × S. The cap therefore permits a pool of up to 25% of issued shares at any moment, which is to say that each additional issued share permits up to a quarter of an additional pool share The company states the result itself. The fixed pool of 319,464,000 is, in its own words, about 24.9% of issued shares at the end of July 2026 and about 20% of the post-exercise total [2]. Those are 25% and 20% before a rounding-down of 160,556 shares that the same notice quantifies. So the relationship is not my model of the clause. It is the clause's operative constraint, stated by the issuer The undertaking is dated inside the period examined here, which is worth saying plainly rather than leaving for someone to find. Three of the four windows that carry the conclusion fall entirely after 26 August 2025 and the fourth straddles it, so the cap covers all of them. Whether it bound earlier is a separate question, and nothing below depends on the answer What none of this means is that every quarter's gap between issued and diluted shares is 25% of issuance. In the fourth quarter of 2024 issued shares rose 169 million while the reported fully diluted count rose only 35 million, because rights already inside that count were being exercised, which moves shares from one bucket to another without adding to the total. So the relationship is checked rather than assumed, in the two quarters the failing windows sit in, and that check is at the end What comes from the company, exactly. The dated bitcoin holdings are irreducibly company-sourced: there is no statutory register of wallets. The fully diluted count is a company construct by definition, and I use it as the company reports it. Issued shares are different, and that is where the work went The standing objection to any analysis of this pool is that the company's metrics are defined and calculated by the issuer, and the adjustment clause moved the diluted count itself. So I rebuilt the issued-share series from statutory filings rather than taking it from the KPI dashboard: period-end counts in the quarterly and annual earnings reports, and the day-by-day exercise tables in the monthly notices filed under the Companies Act [6][7]. Those are company filings too, but they are filed counts rather than issuer-computed metrics, and they are auditable line by line. The company's own published counts then serve as a check on the rebuild rather than as its input Everything after that, the backing per share, the yield, the dilution and the multiple, I compute The rebuild reconciles. Anchoring on 654,714,340 shares at 30 June 2025, adding every daily delivery on its date and the international offering of 385,000,000 shares on its payment date of 16 September 2025 [4], the series lands on 1,140,974,340 at 30 September 2025, the figure in the quarterly report. The difference is zero. It also reproduces the company's own issued-share count at each of the eight intra-quarter dates, to the share, once ordinary settlement is applied: a share delivered on day X counts from day X plus one. Counting from the day of delivery instead would put both the 10 July and the 14 July deliveries inside the 7 to 14 July window, showing 18,000,000 shares where 9,000,000 is what matches That leaves two share counts, both defensible. The honest test is not to choose the denominator that produces the answer you want. It is to run the calculation on both and see whether the conclusion survives A word on which diluted count is which, because the company draws the distinction itself and it is easy to miss. The notice states that the fully diluted share count in its BTC-per-share KPI and the "fully diluted total number of issued shares" defined in the 10th Series terms are similar in name but different in scope [2]. The KPI count is issued shares plus paid-in convertibles plus options, and it includes the 10th Series pool as adjusted. The contractual count is issued shares plus other acquisition rights, and it excludes the 10th Series itself The first table below uses the KPI count, because that is the series the company publishes alongside its bitcoin holdings and the one its own per-share figures are built on. The second uses issued shares rebuilt from filings. Neither is the contractual definition, and neither needs to be: the accretion test measures bitcoin backing per share, and the pool mechanics are verified separately, against the exercise cap above The two counts are not interchangeable. Fully diluted rises when new rights or other potential shares enter the calculation; issued shares rise when those rights are exercised and stock is delivered. A quarter with heavy conversion can therefore add far more to the issued count than to the diluted one Here is the same test on both. Each table uses its own share count throughout, so the identity holds within each and the columns can be checked against each other Company-reported fully diluted basis Period new fully diluted shares dilution change in bitcoin per share multiple passes/BELOW Oct to Dec 2024 - 35,402,320 - 7.2% - +309.8% - 43.85x - passes Jan to Mar 2025 - 85,175,005 - 14.8% - +95.6% - 7.45x - passes Apr to Jun 2025 - 251,788,750 - 30.5% - +129.4% - 5.25x - passes 30 Jun to 7 Jul 2025 - 10,125,000 - 1.2% - +15.1% - 13.48x - passes 7 to 14 Jul 2025 - 11,250,000 - 1.3% - +3.7% - 3.81x - passes 14 to 28 Jul 2025 - 18,000,000 - 2.1% - +2.6% - 2.25x - passes 28 Jul to 4 Aug 2025 - 8,375,000 - 1.0% - +1.7% - 2.79x - passes 4 to 12 Aug 2025 - 11,625,000 - 1.3% - +1.6% - 2.21x - passes 12 to 18 Aug 2025 - 18,625,000 - 2.1% - +2.1% - 2.04x - passes 18 Aug to 1 Sep 2025 - 42,625,000 - 4.5% - +1.1% - 1.25x - BELOW 1 to 8 Sep 2025 - 5,950,000 - 0.6% - +0.1% - 1.08x - BELOW 8 to 30 Sep 2025, the offering - 481,250,000 - 33.6% - +1.7% - 1.05x - BELOW Oct to Dec 2025 - 25,235,000 - 1.7% - +11.9% - 7.89x - passes Jan to Mar 2026 - 164,871,250 - 10.1% - +2.8% - 1.28x - BELOW Apr to Jun 2026 - 7,044,205 - 0.4% - +6.6% - 16.21x - passes Independently reconstructed issued-share basis. Same windows, same formula, different denominator Oct to Dec 2024 - 168,754,110 - 46.5% - +136.2% - 3.93x - passes Jan to Mar 2025 - 97,223,000 - 21.1% - +81.1% - 4.84x - passes Apr to Jun 2025 - 194,808,000 - 29.8% - +131.8% - 5.43x - passes 30 Jun to 7 Jul 2025 - 8,100,000 - 1.2% - +15.1% - 13.35x - passes 7 to 14 Jul 2025 - 9,000,000 - 1.3% - +3.7% - 3.77x - passes 14 to 28 Jul 2025 - 14,400,000 - 2.1% - +2.6% - 2.23x - passes 28 Jul to 4 Aug 2025 - 6,700,000 - 1.0% - +1.7% - 2.77x - passes 4 to 12 Aug 2025 - 9,300,000 - 1.3% - +1.6% - 2.19x - passes 12 to 18 Aug 2025 - 14,900,000 - 2.1% - +2.1% - 2.02x - passes 18 Aug to 1 Sep 2025 - 34,100,000 - 4.5% - +1.1% - 1.24x - BELOW 1 to 8 Sep 2025 - 4,760,000 - 0.6% - +0.05% - 1.07x - BELOW 8 to 30 Sep 2025, the offering - 385,000,000 - 33.7% - +1.4% - 1.04x - BELOW Oct to Dec 2025 - 1,300,000 - 0.1% - +13.8% - 121.84x - passes Jan to Mar 2026 - 131,897,000 - 10.4% - +2.6% - 1.25x - BELOW Apr to Jun 2026 - 7,112,284 - 0.6% - +6.4% - 12.59x - passes The multiples differ, sometimes by a great deal. October to December 2024 prints 43.85x on fully diluted shares and 3.93x on issued shares, because that quarter converted a large number of warrants into stock: the issued count rose 169 million while the fully diluted count rose only 35 million. Neither number is wrong, and neither is the more correct basis in the abstract. They answer different questions But the four windows below the hurdle are the same four on both bases, in the same order, with the same three September 2025 windows dominating. That is the point of running it twice. Across these two defensible share-count bases, the denominator changes the multiple. It does not change the conclusion Dilution and yield are shown to one decimal place, and the multiples are computed on the unrounded figures, so recomputing them from the displayed columns will differ slightly where dilution is under one percent One limit of the measure itself, stated before anyone else states it. The multiple tracks the change in bitcoin backing per share across a window. It does not know, and cannot know, whether that bitcoin was bought with the money those particular shares raised. Bitcoin acquired with debt, with option income, or with cash raised in an earlier window all count toward the window in which it lands. That is what makes it an attribution test rather than a causality test Note which way that bias runs. It is favourable to the issuance: the window is credited with bitcoin that may have been funded somewhere else entirely. Four windows still fail The resolution matters Fifteen issuance windows, reconstructed from sixteen dated anchors, instead of eight quarter-end observations, because at quarterly resolution the September 2025 raise is blended with unrelated warrant exercises and its real character disappears Isn't this just mNAV. Not quite, and the difference is worth being exact about. For a clean raise, where all the new bitcoin comes from the raise itself and is deployed at spot, the bitcoin bought is ΔB = (n × issue price) ÷ BTC price, so: ΔB / n = issue price ÷ BTC price m = (issue price ÷ BTC price) ÷ (B / S) = issue price ÷ NAV per share So for a pure equity raise the accretion multiple collapses to the issue-price-to-BTC-NAV multiple: the price at which shares were sold, over the bitcoin backing each existing share already carried. That reading is cleanest on the issued-share basis and does not carry over mechanically to the fully diluted one, because the 481,250,000 fully diluted shares that window added include 96,250,000 pool shares that raised no cash. Only 385,000,000 of them carried proceeds And it is the issue price that matters, not the screen price: the pricing notice sets the reference at ¥614 on 9 September 2025 and the discount at 9.93%, giving ¥553 [4]. Read back through the identity, a 1.04x accretion multiple implies shares sold at roughly 1.04 times the bitcoin backing each existing share carried, against a screen price about 1.16 times it. Those are implied figures rather than observed ones, because the window was not a pure raise: part of the proceeds became bitcoin only after quarter end. Trading above backing and raising above backing are different facts, and it is the second one this test measures Where it stops resembling a price multiple. Two things pull the accretion multiple away from the issue-price-to-NAV multiple, and they pull in opposite directions It can run very high whenever bitcoin arrives from something other than the shares issued in that window. On the issued basis, October to December 2025 is the extreme case: 1,300,000 new shares, a dilution of one tenth of one percent, against bitcoin rising from 30,823 to 35,102, which was September's proceeds finally landing. The multiple prints 121.84x. Nobody sold stock at 122 times NAV. That is a limitation of the attribution method, and it is deliberately conservative in management's favour It runs low when proceeds take weeks to become bitcoin, which is the same effect seen from the other side. Cash raised on 16 September 2025 was not bitcoin on 30 September, which is why the offering scores about 1.05x at quarter end and much higher once the money had been deployed. That bias runs against management, so it cannot simply be left there. It is corrected below Why the passes pass On both bases, the three large 2024 and early 2025 windows cleared the bar with room. Note April to June 2025 in particular, roughly 30% dilution on either basis, and bitcoin per share still rose more than 100%. Heavy dilution is not the problem. Unproductive dilution is. The six summer windows are warrant exercises, small and continuous, and every one brought in more bitcoin per share than the company already had. They are the strongest evidence in favour of the strategy in this dataset, and they should not be touched Before the failures, the thing that has to be said first Every tested window was accretive. Not one of them destroyed bitcoin per share That is not the test. The test is whether a window was sufficiently accretive to justify the additional compensation it created Why the failures fail In each case the mechanism increased the potential-share count by an amount equal to 25% of the issued-share increase, while bitcoin per share barely moved. 18 August to 1 September 2025: 4.5% dilution bought 1.1%. 1 to 8 September: 0.6% bought almost nothing. The offering: 33.6% dilution bought 1.7%. January to March 2026: 10.1% bought 2.8%. The money did buy bitcoin, and the raises stayed accretive, but only marginally so against the backing each share already carried, so the new shares diluted almost as much as they contributed and the pool grew regardless The offering deserves its own paragraph Between 8 and 30 September 2025 issued shares rose by exactly 385,000,000 and not a single warrant was exercised in that window, which the daily reconstruction confirms rather than assumes, so it is cleanly isolated with no averaging. At quarterly resolution the whole quarter reads 1.78x on the diluted basis and 1.76x on the issued one, because the offering gets blended with the July and August exercises scoring above 2x. Separated, the exercises pass and the offering falls below the hurdle. That is the single reason the finer data mattered Two of the passes are artifacts, and I will say so. October to December 2025 scores high on only 1,300,000 new shares because bitcoin rose 30,823 to 35,102, which is September's money being deployed, not a new raise. April to June 2026 does the same against the first quarter. So merge each failing raise with the window that spent its proceeds, the fairest possible treatment. On the diluted basis the September 2025 offering goes from 1.05x to 1.40x and January to March 2026 from 1.28x to 1.91x; on the issued basis, 1.04x to 1.45x and 1.25x to 1.85x. Both still fall short on both bases The same treatment has to be given to the two smaller failing windows, and it is. Take all three 2025 failures together, from 18 August 2025 through to the quarter that spent the proceeds, and the answer does not move: 1.40x diluted, 1.45x issued. Stop each of them at the next boundary instead, with no credit for the offering's proceeds at all, and they score 1.23x and 1.05x diluted, 1.22x and 1.04x issued. There is no version of the merge that rescues them Extend the offering all the way to June 2026 and it reaches 1.60x diluted, 1.63x issued. Do the same from 18 August 2025 and it reaches 1.59x and 1.62x. Even after giving the proceeds every chance to deploy, nothing clears 2x under either share-count basis Robustness Which is the part that matters more than the hurdle itself. Hurdle, windows falling below, pool shares involved: 1.3x / 4 / 138,939,250 1.5x / 4 / 138,939,250 1.75x / 4 / 138,939,250 2.0x / 4 / 138,939,250 2.5x / 7 / 148,589,250 3.0x / 8 / 150,264,250 Nothing in either table sits between 1.28x and 2.02x, so the answer is identical anywhere in that range. Pick 1.3x, 1.5x, 1.75x or 2x and the same four windows fail on both bases and the same total comes out. So the argument about where the bar belongs, which is a real argument, does not have to be settled to reach the number. Above 2x it does start to move, and I would rather say so than claim otherwise: at 3x four small warrant windows join and the total rises eight percent. The four windows this proposal rests on remain below every hurdle from 1.3x through 3x One last sensitivity, on the only input that is irreducibly company-sourced. For the offering window to reach 2x, reported holdings at 30 September 2025 would have to have been understated by 32%; for January to March 2026, by 7.5%. Those two windows carry 93% of the 138,939,250. The two small September windows are tighter, and they carry the other 7% The number, computed directly Four windows fall below the hurdle on both bases: 18 August to 1 September 2025, 1 to 8 September 2025, the offering window of 8 to 30 September 2025, and January to March 2026 The minting is measured against the increase in issued shares, so it takes the issued column. Those four windows issued 34,100,000, 4,760,000, 385,000,000 and 131,897,000 shares, which at 25% increased the pool by 8,525,000, 1,190,000, 96,250,000 and 32,974,250 potential shares. That is 138,939,250, and the three September 2025 windows are 105,965,000 of it on their own And here is the check, read straight off the company's own diluted series. Over the third quarter of 2025 issued shares rose 486,260,000 while fully diluted shares rose 607,825,000, a gap of 121,565,000, which is exactly 25.000% of the issued-share increase [1]. The six passing summer windows account for 15,600,000 of that gap and the three failing windows for 105,965,000, and the two sum to 121,565,000 exactly. The first quarter of 2026 checks the same way: a gap of 32,974,250 on 131,897,000 issued, again exactly 25.000%. That is a demonstration of how the 10th Series behaved in those two quarters, not a general claim that every historical gap between issued and diluted shares is 25% of issuance So the figure does not rest on adding up eleven passing windows and subtracting, and it does not rest on the algebra alone either. It rests on two quarters where the minting can be read straight off the company's own diluted series, and on a cap the company has since restated in its own words Target identified by the test: 138,939,250 potential shares. Transferable in whole units: 199,625 units at 696 shares each, or 138,939,000. Rounding difference: 250 shares The 199,625 is not a second number and not a new grant. It is the same 138,939,250 shares expressed in the whole 696-share units the rights actually come in One thing this is not It is not a claim that whatever remains is the mathematically correct amount of compensation, nor that the rights were improperly granted. They were granted under terms shareholders approved, the allottees paid third-party-appraised fair value for them out of their own funds [2], and the holders have contractual rights under those terms. What the test produces is a candidate amount for reconditioning, not a legally owed clawback: 138,939,250 is the portion attributable to issuance windows that failed a stated accretion hurdle, and that portion is a better candidate for conditioning than for keeping unconditional

Metaplanet's two monthly filings today (Aug 3) both read zero and both are deliberate - 27th Series warrants: 0 exercised in July. The mNAV gate (≥1.01x) still isn't met, so the moving-strike engine stays idle, no dilution, even after the recent price bump - Buyback: ¥0 in July, still 0 cumulative since October. At ~0.90x mNAV, buying BTC beats buying back stock, so the ¥75B authorization stays untouched, dry powder intact The stock is sitting in the dead zone between the two triggers (issue above 1.01x, buy back below ~0.6x), so both tools correctly read zero Discipline, not inaction contents.xj-storage.jp/xcont… contents.xj-storage.jp/xcont…

1/ Metaplanet just announced a ¥20M Bitcoin shareholder benefit program with Coincheck Most analysts will dismiss it as marketing Read it again The partner choice and the launch date tell you what's actually being set up and it's bigger than a giveaway 👇

1/ Everyone hates Metaplanet right now. "Brutal dilution," "worth less than its own BTC," "death cross. Sentiment is in the gutter I'm going to take each bearish point and show you why the surface read is wrong 👇

#metaplanet #strive #strategy

🚨🚨🚨Metaplanet acaba de revelar algo importante. Tres SPVs de Caimán recibieron acciones en colocaciones primarias de feb/mar y las transfirieron inmediatamente a las mesas de swaps de Morgan Stanley y Goldman Sachs en Londres. 👇 Lo explicamos...