BMNR / MARA short puts: close, roll to Jan-28, or change nothing
Built 17 Aug 2026 from the eight IBKR Risk Navigator exports (per-leg Beta Delta $ + Margin Sensitivity), Monday TWS balances, IBKR order previews for the rolls, and Trade Query income. Advisory only; nothing here places orders; not licensed advice. Marks are Sunday/Monday mids.
What the problem is, in plain English
The BMNR and MARA fortresses were built in 2025 with the short puts near the money. The stocks fell through them. Those puts are now deep in the money with almost no time value left: BMNR 65P (Main 75, Joint 82) at ~33c, MARA 25P (Main 200) at ~55c, MARA 15P (Neville 250) at ~$1.1. Three things follow:
- They are the accidental second long. The put was the financing leg; at delta 0.7 to 0.85 it now behaves like ~48,500 extra shares stacked on top of the long calls you planned to own. That is most of the book's beta-delta.
- They will be exercised early. A deep put with less time value than the interest on its strike gets exercised. That is $2.0M of forced stock purchases (Joint alone $533k against $17k of cash) on someone else's timing.
- They set the equity slope in a drawdown. On the corrected liquidation model (NLV vs a TIMS margin requirement that shrinks as prices fall) IBKR does not sell you out at -10%; but an S&P -20% (miners -55%) still takes total NLV from $3.29M to $1.28M and Retired/Joint through IBKR's ~$100k portfolio-margin floor. Those puts are the reason.
The aim you set: allow time for recovery without margin liquidation, keep the CC income on the rest of the book (~$84k/mo), keep the shares free, and exit these fortresses at roughly BMNR 30 / MARA 20 next year to rebuild risk-defined. Everything else in the book (MU, GOOG, RKLB, IREN, COIN, COPX, SPY, SNDK, ETHA, NOW, PATH, every long call and hedge put, every share block) stays untouched in all three options below. Retired's small BMNR 10C/18P sleeve and the dead UAMY/IBIT/ECHO legs are closed in A and B (~$16k of realized, no income).
The three options, side by side
| A. Close the puts | B. Roll the puts to Jan-28 | C. Change nothing |
|---|
| What you do | Buy back the deep puts at intrinsic (BMNR 65P x157, MARA 25P x200, MARA 15P x250) and leave the sleeves as long call + hedge put + free shares. | Same buybacks, and in the same ticket sell a new Jan-28 put at/near the money (BMNR 18P, MARA 8P), same count. IBKR previews: 65/18 at ~41.00, 45/18 at ~22.00, 25/8 and 15/8 combos. | Nothing. Let the puts sit; accept assignment when it comes. |
| Cash out now (all accounts) | $2,370,285 → $1,066,568 (-$1,303,717) | $2,370,285 → $1,285,429 (-$1,084,856; ~$240k of credit comes back) | no cash now; ~$2.0M of strike money leaves on assignment (Joint -$625k, Neville -$53k) |
| Net Liq change | $0 (a marked liability is paid off) | $0 | $0 (until the stocks move) |
| Realized P&L booked | -$108,322 | -$108,322 (same buyback leg) | none now |
| Delta / leverage | the puts' delta gone entirely; book 4.1x → ~3.5x | ~60 to 65% of it gone; book → ~3.8x | unchanged 4.1x; rises to ~4.3x as puts become shares |
| Assignment risk | gone | gone (new puts have $2 to $6 of time value) | near-certain; ~$2.0M of stock arrives on their timing |
| Time value | none sold | ~$240k sold; ~$8k/month of decay in your favour while BMNR > 18, MARA > 8 | none |
| Maintenance margin | ~-$272k | ~-$160k (IBKR previews) | +$330k as puts convert to stock at 51% / 29% intervals |
| Total NLV at S&P -20% | $1,692,193 (-49%) | $1,465,870 (-55%) | $1,281,664 (-61%) or worse once assigned |
| Total NLV at S&P -30% | $1,184,545 (-64%) | $834,325 (-75%) | $601,832 (-82%) |
| IBKR liquidation crossover | none within -50% in any account | none within -50% in any account | Neville ~-26% (flat vol), Joint ~-40%; after assignment Joint ~-30%, Main ~-40 to -55% |
| ~$100k PM floor | Retired ~-39% | Retired ~-39%, Joint ~-43% | Retired ~-25%, Joint ~-32%; after assignment Neville ~-36% too |
| Structure afterwards | long call + hedge put + shares: a paid-for leveraged long, no financing leg | a fortress again (LC + new SP + HP), safe strike reset near today's price, same expiry as the calls (no naked-put window) | as today, then shares at 65 / 25 / 15 basis |
| Recovery upside | full: every call and every share kept | full, plus the new puts decay to zero on the way up | full, plus ~$700k more stock if assigned |
| Downside on a further 30 to 50% drop | shallowest | gives back about a third of A's protection (new puts go ITM again below 18 / 8, at a quarter of the old dollar size) | today's slope, then steeper (more shares) |
| Joint specifically | loan ~$443k (~$21k/yr at IBKR Pro tiers: 5.13% on the first $100k, 4.63% above) | loan ~$395k (~$19k/yr) | loan ~$625k on assignment (~$29k/yr), PM floor ~-30% |
| Neville specifically | buyback $174k from $319k cash | roll credit $59k; Jun-27 sleeve still expires Jun-27 | MARA 15P assigns for $375k vs $319k cash (~$3k/yr on the ~$53k shortfall) |
| CC income | unchanged, ~$84k/mo | unchanged | unchanged |
| Exit at BMNR 30 / MARA 20 next year | no put to close | the 18P / 8P are worth ~$1 / ~$0.15 then; ~$10k to close | you'd be selling ~32k BMNR + ~55k MARA shares |
How to choose. A and B are the same trade with one difference: whether you re-establish a financing put at today's price. A is maximum protection of the margin curve, paid for in cash. B hands ~$240k of that cash back and rebuilds the fortress structure, paid for with about a third of A's downside protection (worst in Joint). C is not a risk-management choice at all: it is a decision to own ~$700k more BMNR/MARA on margin, on the put-holder's timing, and it brings back every crossover and PM-floor exposure the other two remove. Between A and B, the account that matters is Joint: it carries the loan and the least cushion, so if you roll, re-sell only 60 to 75% of the 82 BMNR puts there (see the sp_roll_fix rows below); Main and Neville can carry a full roll.
Liquidation risk by account (corrected model: NLV vs shrinking TIMS margin)
| Account | NLV now | NLV at S&P -20%: today / A / B | at -30%: today / A / B | IBKR crossover: today / A / B | PM floor: today / A / B | Cash: today / A / B |
|---|
| Retired | $386,900 | $139,784 | $183,236 | $183,236 | $66,222 | $123,831 | $123,831 | none within -50% | none within -50% | none within -50% | ~-25% | ~-39% | ~-39% | $493,581 | $506,189 | $506,702 |
| Neville | $770,500 | $264,323 | $378,518 | $306,195 | $129,717 | $298,765 | $186,107 | -26% | none within -50% | none within -50% | not within -60% | not within -60% | not within -60% | $319,094 | $144,964 | $206,944 |
| Joint | $634,700 | $264,951 | $359,529 | $309,316 | $115,978 | $247,937 | $170,932 | -39% | none within -50% | none within -50% | ~-32% | not within -60% | ~-43% | $16,763 | -$443,128 | -$394,303 |
| Main | $1,500,000 | $612,606 | $770,909 | $667,124 | $289,916 | $514,012 | $353,455 | none within -50% | none within -50% | none within -50% | not within -60% | not within -60% | not within -60% | $1,540,847 | $858,544 | $966,085 |
| TOTAL | $3,292,100 | $1,281,664 | $1,692,193 | $1,465,870 | $601,832 | $1,184,545 | $834,325 | | | $2,370,285 | $1,066,568 | $1,285,429 |
Model: full Black-Scholes revaluation of every leg for beta-weighted index moves with the v28 crash-vol ramp, against a TIMS scan margin per underlying at IBKR's own margin intervals (BMNR 51%, MARA 29%, ...), calibrated x1.21 to 1.25 to today's actual maintenance. Crossover shown on flat vol (the conservative one). PM floor: IBKR removes portfolio-margin treatment below roughly $100k NLV, after which Reg-T requirements apply; verify the exact threshold. Retired is identical in A and B (no BMNR/MARA puts there beyond the small sleeve closed in both).
The trades
A. Close the puts (and the dead legs)
| Account | Leg | Kind | Beta-delta removed | Cash | Maint margin (IBKR per-leg) | Realized |
|---|
| Retired | UAMY NOV 20 '26 10 Call | DEAD | -$7,197 | +$727 | -$700 | -$12,242 |
| Retired | UAMY NOV 20 '26 12.5 Put | DEAD | -$29,796 | -$17,742 | -$2,475 | -$7,316 |
| Retired | UAMY NOV 20 '26 7.5 Put | DEAD | +$20,695 | +$6,381 | +$5,450 | +$2,878 |
| Retired | IBIT AUG 21 '26 50 Call | DEAD | -$46 | +$2 | +$0 | -$2,983 |
| Retired | BMNR JAN 21 '28 10 Call | LINEAR | -$232,146 | +$53,233 | -$41,750 | -$13,342 |
| Retired | BMNR JAN 21 '28 18 Put | LINEAR | -$79,462 | -$31,369 | -$15,500 | +$5,405 |
| Retired | BMNR NOV 20 '26 10 Put | LINEAR | +$16,439 | +$1,375 | +$6,600 | -$1,183 |
| Neville | MARA JUN 17 '27 15 Put | BUYBACK | -$411,703 | -$174,130 | -$39,500 | -$43,118 |
| Joint | BMNR JAN 21 '28 65 Put | BUYBACK | -$364,357 | -$387,710 | -$74,374 | +$6,011 |
| Joint | BTBT JAN 15 '27 4 Put | BUYBACK | -$37,084 | -$25,124 | +$5,200 | -$8,994 |
| Joint | QUBT JAN 21 '28 22 Put | BUYBACK | -$39,276 | -$35,560 | -$4,100 | -$7,241 |
| Joint | EOSE JAN 21 '28 15 Put | BUYBACK | -$6,598 | -$11,497 | -$950 | +$384 |
| Main | ECHO SEP 18 '26 150 Call | DEAD | -$291 | +$14 | -$15 | -$3,485 |
| Main | BMNR JAN 21 '28 65 Put | BUYBACK | -$333,253 | -$354,613 | -$68,175 | +$7,885 |
| Main | MARA JAN 21 '28 25 Put | BUYBACK | -$401,777 | -$327,704 | -$44,200 | -$30,981 |
| Total | -$1,905,852 | -$1,303,717 | -$274,489 | -$108,322 |
Order: Joint's BMNR 65P first (the account that can carry neither the loan on assignment nor the drawdown), then Main's BMNR 65P and MARA 25P (from cash), then Neville's MARA 15P, then Retired's three legs and Joint's small BTBT/QUBT/EOSE puts. Work the deep-ITM puts at or near intrinsic; there is almost no time value in them.
B. Roll the puts to Jan-28: same buybacks, plus the new short puts (sp_roll_fix, IBKR previews)
| Old put | Buyback (mid) | New put | IBKR combo mid | Credit on new put | Net cash out | New delta (IBKR) | Delta removed vs A | Time value / carry | Give-back if the stock drops 20% / 40% | Maint margin (IBKR preview) |
|---|
| BMNR 65P Joint x82 | $388k | Jan-28 18P x82 | 41.00 | $48k | -$340k | -0.30 | 64% | 8.6x | -$9.6k / -$22k | -$46k |
| BMNR 65P Main x75 | $355k | Jan-28 18P x75 | 41.00 | $44k | -$311k | -0.30 | 64% | 8.6x | -$8.8k / -$20k | -$44k |
| BMNR 45P Main x50 | $142k | Jan-28 18P x50 | 22.00 | $29k | -$113k | -0.30 | 56% | 8.6x | -$5.9k / -$13k | -$23k |
| MARA 25P Main x200 | $327k | Jan-28 8P x200 | ~14.0 | $47k | -$280k | -0.24 | 67% | 7.8x | -$10k / -$23k | ~-$45k |
| MARA 17P Joint x50 | $46k | Jan-28 8P x50 | ~6.8 | $12k | -$34k | -0.24 | 57% | 7.8x | -$2.5k / -$6k | ~-$7k |
| MARA 15P Neville x250 | $173k | Jan-28 8P x250 | ~4.5 | $59k | -$114k | -0.24 | 58% | 7.8x | -$12.6k / -$29k | ~flat |
Why Jan-28: same expiry as the calls and hedges, so the whole sleeve is always closed or rolled together and there is never a naked-put window; the longer Jun-28 / Dec-28 puts pay ~$6 to $15k more per block but cost about that much more to close at your exit. Why 18 / 8: sp_roll_fix prices every real strike, expiry and quantity, gates on the 1B-roll goals (delta cut ≥ 55%, no early-exercise incentive, strike ≤ spot) and ranks; 18 is the ATM strike on BMNR's sparse chain (13/15/18), 8 is the first OTM strike on MARA's (7/8/10). Partial re-sell for Joint: 62 of 82 BMNR puts removes 72% of the delta for $36k of credit; 25 of 50 MARA puts removes 81% for $6k. Give-back is per stock move (an S&P -20% is BMNR/MARA -55%). Re-run in RTH: cd /Users/abhi/fortress/scripts && python3 sp_roll_fix.py
C. Change nothing: what the book needs to reach $5M NLV, and what assignment does
| Scenario | Horizon | Multiplier | MARA | IBIT | BMNR |
|---|
| Only MARA/IBIT/BMNR move | now | x2.10 | $19.2 | $74.7 | $37.9 |
| in 6 months | x2.28 | $20.9 | $81.3 | $41.2 |
| in 12 months | x2.47 | $22.7 | $88.0 | $44.6 |
| Whole crypto cluster moves together | now | x1.65 | $15.1 | $58.7 | $29.8 |
| in 6 months | x1.76 | $16.1 | $62.6 | $31.7 |
| in 12 months | x1.88 | $17.2 | $66.9 | $33.9 |
Black-Scholes revaluation of the whole book with the crypto names uncapped, IV constant. Roughly a doubling of the three names (or +65% across the whole crypto complex) reaches $5M today; more the longer it takes, because the book bleeds ~$37k/month of net time value if nothing moves (CC income of ~$80k/mo, none of it from these names, would offset that). The path is the problem: the same book is -61% at an S&P -20% on the way, with Retired and Joint through the PM floor at -25% / -32%, and the deep puts convert to ~$2.0M of stock on the holder's timing (Joint's loan ~$625k, ~$29k/yr at IBKR Pro's tiered USD rates: 5.13% on the first $100k, 4.63% from $100k to $1M). Accepting assignment is worse than A or B on cash, margin, drawdown and crossovers; the only thing it is better at is owning ~$700k more BMNR/MARA if they rally.
Recovery grid: the whole BMNR and MARA book (after A or B; identical to within ~$10k)
Every remaining BMNR/MARA leg in every account plus every share, short puts closed (B: rolled to Jan-28 18P / 8P, worth $0 to $2 at these prices). Prices reached on the date shown; Black-Scholes at constant IV (BMNR 95%, MARA 85%). Neville's Jun-27 MARA sleeve is valued to its own expiry, which is why MARA decays hard between Dec-26 and Jul-27 (at $15 to $25 in June 2027 that sleeve's 25C and 13P expire worthless; it needs a roll or an exit before then).
Position today
| BMNR book (Joint + Main) | MARA book (Main + Joint + Neville) |
|---|
| Shares | 16,300 sh, avg basis ~$61, mark $18.24: value $297k, unrealized -$701k | 10,000 sh, basis $14.75, mark $9.32: value $93k, unrealized -$54k |
| Long calls | 275x 23C/25C Jan-28 + 25x 35C Jan-27: $164k | 450x (20C Jan-28 x250, 25C Jun-27 x250) + 200x 40C Jan-27: $64k |
| Hedge puts | 275x 25P/15P Jan-28 + 25x 45P Jan-27: $242k | 450x (15P/10P Jan-28, 13P Jun-27): $304k |
| Short puts (the problem legs) | -157x 65P, -50x 45P Jan-28: liability -$884k | -200x 25P, -50x 17P Jan-28, -250x 15P Jun-27: liability -$546k |
| Book value today, short puts still on | -$179,962 | -$221,565 |
| Book value after A or B (puts paid off from cash) | $704,038 | $324,435 |
| Distance to original basis = unrealized loss still on the book | -$746,615 | -$215,453 |
Cash cost of A ≈ $884k + $546k = $1.43M; of B ≈ $1.19M after the new-put credits. Neither changes Net Liq (a marked liability is paid off), so the grid below is not netted against them; the only accounting change is ~$83k of MARA loss moving from Unrealized to Realized.
BMNR book (today -$179,962 with the puts on; distance to basis -$746,615)
| Date | $25 vs today | $30 vs today | $50 vs today | $70 vs today | $100 vs today | $150 vs today | | $25 vs basis | $30 vs basis | $50 vs basis | $70 vs basis | $100 vs basis | $150 vs basis | Breakeven vs today | Breakeven vs original basis |
|---|
| Dec-26 | +$156,957 | +$318,242 | +$1,078,839 | +$1,937,743 | +$3,281,143 | +$5,565,127 | | -$589,658 | -$428,373 | +$332,224 | +$1,191,128 | +$2,534,528 | +$4,818,512 | $19.6 | $41.7 |
| Jul-27 | +$54,111 | +$202,835 | +$961,459 | +$1,847,836 | +$3,217,823 | +$5,525,806 | | -$692,504 | -$543,780 | +$214,844 | +$1,101,221 | +$2,471,208 | +$4,779,191 | $23.0 | $45.0 |
| Dec-27 | -$87,369 | +$55,642 | +$884,333 | +$1,809,370 | +$3,198,352 | +$5,513,351 | | -$833,984 | -$690,973 | +$137,718 | +$1,062,755 | +$2,451,737 | +$4,766,736 | $28.2 | $47.0 |
MARA book (today -$221,565 with the puts on; distance to basis -$215,453)
| Date | $15 vs today | $20 vs today | $25 vs today | $40 vs today | $100 vs today | | $15 vs basis | $20 vs basis | $25 vs basis | $40 vs basis | $100 vs basis | Breakeven vs today | Breakeven vs original basis |
|---|
| Dec-26 | +$60,103 | +$205,428 | +$397,993 | +$1,200,744 | +$5,802,739 | | -$155,350 | -$10,025 | +$182,540 | +$985,291 | +$5,587,286 | $11.8 | $20.3 |
| Jul-27 | -$69,393 | +$31,489 | +$157,215 | +$990,743 | +$5,761,541 | | -$284,846 | -$183,964 | -$58,238 | +$775,290 | +$5,546,088 | $18.6 | $26.1 |
| Dec-27 | -$136,306 | -$66,136 | +$68,462 | +$952,629 | +$5,752,425 | | -$351,759 | -$281,589 | -$146,991 | +$737,176 | +$5,536,972 | $22.7 | $27.6 |
Reading the breakevens. "Vs today" is the price the stock must reach on that date just for the book to be worth what it is worth now; it climbs (BMNR $19.6 → $23 → $28; MARA $11.8 → $18.6 → $22.7) because the Jan-28 calls burn time value while you wait, so standing still is not free. "Vs original basis" is the price at which everything you paid, including the assigned shares at 65 / 25, is recovered: BMNR ~$42 by Dec-26 drifting to ~$47 by Dec-27; MARA ~$20 by Dec-26 drifting to ~$28. At the BMNR 30 / MARA 20 exit levels: about +$0.5M from today if reached by Dec-26, roughly flat if it takes until Dec-27, and $0.4M to $1.0M short of basis either way. The book is genuinely convex above ~$40 BMNR / ~$25 MARA where the calls take over from the shares (BMNR $50 in Dec-26 is +$1.08M from here; $70, +$1.94M). The shares carry the recovery with no clock; the calls only until Jan-28.
Method and assumptions
- Data: IBKR Risk Navigator per-leg exports (Beta Delta $, marks, unrealized) and Margin Sensitivity exports (per-unit maintenance change and per-underlying margin interval), Sunday 16 Aug 2026; TWS balances Monday 17 Aug; IBKR order previews for the roll combos (delta, margin impact, mids); Trade Query realized-income legs for the CC run-rate ($84,509/mo on 90 days, $82,322/mo on 180).
- Liquidation model: full Black-Scholes revaluation vs a TIMS scan margin using IBKR's per-underlying intervals, calibrated x1.21 to 1.25 to today's maintenance, beta-weighted index moves, v28 crash-vol ramp (flat-vol shown as the conservative crossover). It replaced an earlier flat-margin proxy that wrongly said "-10%"; margin shrinks as prices fall in these structures, so the real risk is equity destruction and the PM floor, not a sale at -10%.
- Recovery grid and $5M path: Black-Scholes at constant IV; a recovery would compress IV, trimming the near-ATM call values a little and mattering not at all where intrinsic dominates.
- Roll pricing: sp_roll_fix.py (scripts/, tests in tests/test_sp_roll_fix.py) prices every strike x expiry x quantity through module_ibkr_fetch; off-RTH FROZEN quotes have no greeks so delta falls back to Black-Scholes; IBKR's own preview deltas were used where you supplied them. Give-back is per stock move.
- Not modelled: house margin hikes in a crash (they pull crossovers closer), IV compression on the way up, exact TIMS recomputation after the trades. Re-export the eight files after each tranche and the model re-runs.
- Engine: delever.py, plan3b.py, liq.py, opt3.py, sp_roll_fix.py, render4.py in ~/fortress/_md_files/2026-08-17/ and scripts/.