Fortress · Coverage Insights · weekly · 2026-W45 · as of 2026-11-04 · source coverage_insights.py

Coverage Insights

The Desk's Coverage > Insights tab as a reference page: your registered writing rules and how they are grading, the record of how you write and close calls, whether your rolls worked, and each fortress against its own design. Refreshed every night (03:30 SGT), every Saturday 08:30 SGT and on every Regenerate latest.

1. How to read the numbers

$ per $1M per day = net premium kept / (value of the shares covered x days the call was on) x 1,000,000. It puts a $20 stock and an $800 stock, and a 3-day and a 30-day call, on one scale. Example: 10 contracts on a $100 stock ($100,000 of stock), open 7 days, $700 kept: $700 / ($100,000 x 7) x 1,000,000 = $1,000 per $1M per day.

Three yearly returns on the same income (x 365, no compounding): ROI on Notional = on the value of the stock the call covered; ROI on IC = on the Invested Capital (Net Debit x Shares) of the fortress shares the call covered; ROI on ML = on the Max Loss of those shares (today's IC and ML from the sheet). IC and ML exist only for calls on a live fortress, so those two cover fewer calls.

Per $1M per day	Per month on $1M covered	ROI on Notional
$1,000	$30,000	36.5%
$684	$20,500	25.0%
$318	$9,500	11.6%

2. Your writing rules

Written down on 2026-10-15, before any new data, in ledgers/write_rules_registered.json. Graded each week ONLY on calls sold after that date: a rule needs 8 calls on each side before it is graded; confirmed = the 90% range excludes zero in the registered direction, holding = same direction, not yet sure, slipping = the other way, not yet sure, refuted = the 90% range excludes zero against it.

Rule	Found Dec to Oct	Since registration	90% range	Status
On stocks with 20-day realized volatility of 40% or more, write 10%+ out of the money rather than 5-10% out.	+$742 per $1M/day (27.1%/yr)	-$828 (-30.2%/yr), 20 vs 14 calls	-$1,803 to +$194	slipping
Sell fresh calls 14 days or less out (days left to expiry) rather than 15 to 60 days out.	+$366 per $1M/day (13.4%/yr)	42 vs 0 calls	–	collecting
Do not let a call run across an earnings report (its first reacting session after the sale and on or before expiry).	+$459 per $1M/day (16.7%/yr)	+$1,936 (70.7%/yr), 38 vs 1 calls	–	collecting
Closing a call early once 80%+ of the premium is kept, then writing again, beats holding it to expiry.	+$83,198 over 119 closes	+$245 over 1 closes	–	collecting
Buying a call back early at a loss does worse than holding it to expiry.	-$30,445 over 57 closes	0 closes	–	collecting
In FIGHT territory (stock 20%+ below its fortress's safe strike at the sale), write 10-15% out of the money rather than 5-10% out, with 14 days or less to expiry.	+$1,510 per $1M/day (55.1%/yr)	0 vs 0 calls	–	collecting

3. How you write calls

Every write since 2025-12-15 (607 writes, incl. 25 defensive buybacks), each judged on its whole chain: its own premium plus every roll it led to, open chains marked to close today. The sheet books a roll's buyback on the new [roll] row, so a rolled-out call otherwise shows its full premium.

ROI on IC and ROI on ML cover the 244 writes that map to a live fortress; across them the stock covered was 2.96x the Invested Capital and 2.53x the Max Loss.

Tenor

Calls sold 14 days or less out (days left to expiry) earned $803 per $1M per day, 29%/yr ROI on Notional (472 calls, 9% lost money) vs $315 per $1M per day, 11%/yr ROI on Notional for 15 to 60 days (130 calls, 22% lost money): 2.55x.

Days to expiry at the sale	Writes	Per $1M/day	ROI on Notional	ROI on IC	ROI on ML	Kept	Lost money
0-7d	215	$727	27%	72%	61%	76%	11%
8-14d	257	$850	31%	96%	82%	86%	8%
15-30d	101	$347	13%	55%	45%	58%	25%
31-60d	29	$236	9%	40%	33%	45%	14%
61d+	5 (few)	-$275	-10%	-15%	-15%	–	60%

Takeaway Sell calls 8-14 days out (that many days left to expiry): the best ROI on ML (82%/yr) and the fewest losers (8% lost money). Weakest: 31-60 days out (33%/yr on ML).

Distance, by the stock's volatility

On volatile stocks (40%+ volatility), calls 10%+ out earned $1,160 per $1M per day, 42%/yr ROI on Notional (276 calls) vs $506 per $1M per day, 18%/yr ROI on Notional for 5-10% out (136 calls).

Stock's volatility	in the money	0-2% out	2-5% out	5-10% out	10%+ out
calm (<40%)	$191 7% on Notional · – on IC · – on ML 100% kept · 0% lost · 2 calls	$382 14% on Notional · 46% on IC · 46% on ML 76% kept · 16% lost · 25 calls	$417 15% on Notional · 73% on IC · 67% on ML 92% kept · 4% lost · 57 calls	$300 11% on Notional · 21% on IC · 18% on ML 47% kept · 10% lost · 48 calls	$546 20% on Notional · 61% on IC · 47% on ML 73% kept · 27% lost · 11 calls
active (40-80%)	$558 20% on Notional · 65% on IC · 48% on ML 83% kept · 20% lost · 5 calls	$451 16% on Notional · 62% on IC · 49% on ML 62% kept · 25% lost · 8 calls	$650 24% on Notional · 26% on IC · 22% on ML 68% kept · 4% lost · 24 calls	$588 21% on Notional · 70% on IC · 52% on ML 77% kept · 13% lost · 97 calls	$1,005 37% on Notional · 89% on IC · 77% on ML 81% kept · 16% lost · 111 calls
wild (80%+)	$11 0% on Notional · – on IC · – on ML 10% kept · 50% lost · 2 calls	$583 21% on Notional · 84% on IC · 72% on ML 73% kept · 25% lost · 4 calls	$520 19% on Notional · 46% on IC · 38% on ML 49% kept · 14% lost · 7 calls	$363 13% on Notional · 22% on IC · 17% on ML 58% kept · 18% lost · 39 calls	$1,286 47% on Notional · 136% on IC · 119% on ML 77% kept · 10% lost · 165 calls

Takeaway On every kind of stock, write 10%+ out (calm $546, active $1,005, wild $1,286 per $1M per day) (on calm stocks it also lost money most often: 11 calls, 27% lost).

Calls across earnings

Calls that ran across a report earned $245 per $1M per day, 9%/yr ROI on Notional and 38% lost money (55 calls) vs $795 per $1M per day, 29%/yr ROI on Notional and 10% for calls clear of earnings (458 calls).

Days to expiry at the sale	Across a report	Clear of earnings
all writes	$245 9% on Notional · 53% on IC · 46% on ML 31% kept · 38% lost · 55 calls	$795 29% on Notional · 82% on IC · 66% on ML 80% kept · 10% lost · 458 calls
0-7d	$1,544 56% on Notional · 1808% on IC · 1808% on ML 70% kept · 0% lost · 6 calls	$944 34% on Notional · 79% on IC · 64% on ML 76% kept · 11% lost · 180 calls
8-14d	$550 20% on Notional · 19% on IC · 16% on ML 41% kept · 14% lost · 7 calls	$896 33% on Notional · 109% on IC · 91% on ML 86% kept · 9% lost · 216 calls
15-30d	$137 5% on Notional · 48% on IC · 43% on ML 18% kept · 58% lost · 26 calls	$632 23% on Notional · 68% on IC · 49% on ML 72% kept · 14% lost · 49 calls
31-60d	$282 10% on Notional · 95% on IC · 76% on ML 31% kept · 31% lost · 13 calls	$232 8% on Notional · 18% on IC · 14% on ML 56% kept · 0% lost · 12 calls

Takeaway Keep calls clear of earnings: clear of a report they earned $795 per $1M per day and 10% lost money; across one, $245 and 38% lost.

When you close

Closing early at 80%+ kept earned $659 per $1M per day, 24%/yr ROI on Notional over the whole cycle (incl. 4.5d uncovered on average before the next write, 119 calls) vs $360 per $1M per day, 13%/yr ROI on Notional for holding to expiry (9.6d gap, 263 calls).

Buying back early at a loss did worse than holding to expiry in 39 of 54 cases (-$31,730 after the calls written next).

How it was closed	Writes	Per $1M/day while on	ROI on Notional	ROI on IC	ROI on ML	Kept	Lost money	Uncovered after	Whole cycle /day	Whole cycle ROI	vs holding to expiry
early, 80%+ kept	119	$1,058	39%	129%	112%	87%	0%	4.5d	$659	24% N · 81% IC · 69% ML	+$80,047 (63 of 119 better held)
early, 50-80% kept	74	$784	29%	190%	165%	69%	0%	6.6d	$400	15% N · 81% IC · 70% ML	+$42,915 (42 of 74 better held)
early, under 50% kept	36	$548	20%	48%	41%	27%	0%	7.0d	$221	8% N · 20% IC · 17% ML	+$66,660 (24 of 36 better held)
early, at a loss	55	-$1,444	-53%	-183%	-178%	-160%	100%	10.6d	-$695	-25% N · -99% IC · -97% ML	-$31,730 (39 of 54 better held)
held to expiry	263	$717	26%	78%	66%	92%	4%	9.6d	$360	13% N · 38% IC · 32% ML	–
rolled into a new call	60	$430	16%	47%	38%	170%	17%	5.3d	$362	13% N · 41% IC · 32% ML	–

Takeaway Take profits early once 80%+ of the premium is kept, then write again ($659 per $1M per day over the whole cycle); never buy a call back early at a loss (-$695, -$31,730 vs holding).

Whole cycle adds the days the ticker sat uncovered before your next write (capped at 30d). vs holding to expiry = the early close plus the call written next, minus what the call would have made held to its expiry.

FIGHT calls

Distance out of the money	FIGHT territory, 14 days or less	FIGHT territory, 15 days+	Outside FIGHT, 14 days or less
under 5% out	–	–	–
5-10% out	–	–	–
10-15% out	–	–	–
15%+ out	–	–	–

FIGHT's own targeted picks since 2026-07-02 (1189): median 8.8% out; under 5% out 18%, 5-10% out 44%, 10-15% out 29%, 15%+ out 10%.

FIGHT territory = the stock 20%+ below its fortress's safe strike (CSV) on the sale day; only calls sold on the fortress's current structure can be placed (0 of 0 such writes were in FIGHT territory).

4. Did your rolls work?

Every roll of a covered call since July (216 rolls). A roll-up works when the upside it bought beats what it cost; a roll-down works when its premium beats the upside the lower cap gave up.

Rolling up: 36 of 65 worked, paid +$134,385, bought +$65,756 of upside. Rolling down: 138 of 149 worked, collected +$560,329, gave up -$168,730 of upside.

Your roll rules	Rolls	Worked	Net	Verdict
Roll-up for a credit or about even (up and out)	32	31 of 32	+$214,339	works
Roll-up for a debit	33	5 of 33	-$14,198	avoid
Roll-down to 5%+ out of the money	104	98 of 104	+$148,525	works
Roll-down to under 5% out of the money	45	40 of 45	+$243,074	works
Rolling back down within 14 days of a roll-up	40	39 of 40	+$59,847	works
Rolling back up within 14 days of a roll-down	31	16 of 31	+$98,562	mixed

Takeaway Roll down at any distance (5%+ out: 98 of 104 worked; closer: 40 of 45); roll up only for a credit or about even (31 of 32 worked); rolling back down after a roll-up is fine (39 of 40 worked). Don't roll up for a debit (5 of 33 worked).

5. Fortresses vs their own design

Each fortress is judged on what its calls actually earned over the last 90 days (closed, plus open calls marked to close today) against max(its hedge carry, half of what it is designed to earn). Design = 40% of its Max Loss a year, the same in FIGHT or normal mode; once a fortress built from a v28 run has held 90 days, v28's own income for that structure x the book's keep rate (recorded at the build). Today's offer = the expected net of the pick FIGHT or the scanner recommends now, a diagnosis only. 0 not paying their hedge, 0 below design, 0 earning their keep, 13 not judged. The judged book earned $0/mo against $0/mo of hedge carry.

Fortress	Verdict	Earned /mo	Design /mo	Hedge /mo	vs design	% of ML / yr	Today offers	Calls on
AXP Main:1299	TOO NEW	$0	$1,215	$322	0%	0%	$1,044	–
PEP Main:1299	TOO NEW	$0	$1,231	$503	0%	0%	$793	–
CI Main:1299	TOO NEW	$0	$2,340	$526	0%	0%	$312	–
ABT Main:1299	TOO NEW	$0	$1,636	$314	0%	0%	$941	–
LLY Main:1299	TOO NEW	$0	$2,062	$433	0%	0%	$1,301	–
MCD Main:1299	TOO NEW	$0	$3,040	$109	0%	0%	$51	–
JPM Main:1299	TOO NEW	$0	$1,533	$269	0%	0%	$1,309	–
XOM Main:1299	TOO NEW	$0	$2,603	$739	0%	0%	$2,947	–
NVDA Main:1299	TOO NEW	$0	$2,204	$412	0%	0%	$2,334	–
SPCX Main:1299	TOO NEW	$0	$2,366	$331	0%	0%	$4,218	–
MU Main:1299	TOO NEW	$0	$1,783	$245	0%	0%	$2,938	–
DELL Main:1299	TOO NEW	$0	$3,075	$0	0%	0%	$4,237	–
COIN Main:1299	TOO NEW	$0	$3,456	$249	0%	0%	$4,518	–

6. The causality study behind the rules (2026-10-15)

Run on every closed call since Dec 2025 to find which writing choices CAUSE better income, not just which buckets earned more. Bundle: _md_files/cc_check/2026-10-15_write_causality/ (analysis.py, results.json, README).

Whole chains: each write charged with every roll it led to (open chains marked). This cut the short-call yield from $997 to $684 per $1M per day; the edge over longer calls fell from 3.0x to 2.2x but survived.
Controls: weighted regression with a per-ticker baseline, the stock's drawdown and trend at the sale; 90% ranges from resampling ticker-weeks (calls on one ticker in one week are not independent bets).
Out of sample: found on Dec to Jun, tested on Jul to Oct; also re-run without the biggest ticker.
Skill vs luck: implied vol backed out of each premium vs the volatility the stock then realized. The typical call was sold at a fair-to-rich price (median edge +$115); the dollar total was negative because of a few big rallies (a third from the 10 worst calls). Income came from managing calls, not from pricing edge.
Baselines (held to expiry): your strikes -$464 per $1M/day; a mechanical 30-delta weekly at your own implied vol -$41 to -$214; since July the scanner's picks +$418 to +$554 vs yours +$58 (it writes ~20% out over ~15 days, you ~9% out over 7).
Failed, not registered: calm stocks 2-5% out beats 5-10% (flipped on Jul to Oct); selling when implied vol is rich vs trailing realized (flipped); 0-7 day vs 8-14 day calls (no difference); weeklies across a report (12 calls, too few).
Reconciliation: IBKR fills only exist from 2026-10-05; where clean, the sheet matches IBKR to the dollar (COPX, CRCL) or within $24 (SPY). A full check needs an IBKR Flex trades export from Dec 2025.

Outside evidence

Out-of-the-money writing beating at-the-money: Cboe's 30-delta BuyWrite index (BXMD) vs BXM, 9.59% vs 8.14% a year 1990 to 2018 (Neuberger Berman).
Short tenor: practitioner evidence is mixed (tastytrade via Luckbox favours monthlies on risk/reward); your own Jul to Oct test is the stronger evidence.
Earnings: straddles bought before reports earn positive returns, i.e. sellers underprice the event (Chung & Louis 2017); you sold reports at 47% implied vs 72% realized.

Generated by coverage_insights.py (module_insights_reference.py) from the same state the Desk shows. Yahoo daily closes and earnings dates (ETFs have none); the stock price used for a sale is that day's close. Portfolio level: per-ticker samples are too small. Light theme only.