Held at 1.33× for six years, the option construction costs +1.37% a year more than the margin loan on the index fund and +1.43% on the volatile name. That is a real price for removing a real risk. But at that ratio the margin book's worst equity ratio over the whole window was 75% against a 25% maintenance requirement — it was never called, and it was never close. Insurance, correctly priced, against something the construction had already excluded.
Granted, not measured: that long-dated deep-in-the-money calls really do remove the margin call. They do. There is no loan behind them, so no broker can force the position closed. That half of the pitch is true and we are not testing it.
Tested: what it costs. We ran the same 1.33× exposure two ways over the same six years and charged both sides honestly — bought at the ask, marked at the bid every day, commission on both sides, dividends credited against the loan. 13 entries, 11 of them rolls.
Committed 6beaa55 on before any code ran, before the probe was run — the wager you cannot move after seeing the cards.
Does the non-callable construction cost less than the margin loan it replaces, at a ratio a careful person would actually hold?
We were wrong, and wrong in the direction that made the product look better than it is. Both names came in above the band we guessed.
| Parameter | Locked value |
|---|---|
| Leverage | 1.33× (1.33 shares per $1 of equity) |
| Maintenance requirement | 25% |
| Margin rate | Fed funds + 1.5% — a pre-registered approximation, never a broker quote |
| Cash sweep | max(0, Fed funds − 0.5%) |
| Option selection | largest listed strike at or below 70% of spot |
| Roll trigger | time to expiry < 180 days |
| Commission | $0.66 per contract, both sides |
| Fill convention | bought at the ask, marked at the bid daily |
| Window | 2020-07-01 → 2026-06-22 |
| Tickers | QQQ, MU |
| # | Pre-registered goalpost | Measured | Pass |
|---|---|---|---|
| 1 | Pre-registered goalpost — drag ≤ 1.0 pt/yr on both names | +1.37% / +1.43% | NO |
| 2 | Pre-registered goalpost — implied financing inside FFR → FFR+4% on every entry | QQQ 4 of 7 inside; MU 0 of 6 | NO |
| 3 | Pre-registered goalpost — the margin arm is called at least once in the window | never called on either name | NO |
| 4 | Pre-registered goalpost — the verdict survives a ±1% move in the margin rate | holds on 2 of 2 | YES |
1 of 4. The one that passed is the sensitivity — and it is the one that had never been run until 2026-08-28. The three that failed all failed in the same direction: the option construction is more expensive than registered, and the risk it insures against never arrived.
| Figure | Value | What it means | ||||
|---|---|---|---|---|---|---|
| QQQ · the index fund | 23.69% | 22.32% | +1.37% | 19.76% | 75.2% | never called |
| MU · the volatile name | 78.37% | 76.94% | +1.43% | 70.38% | 71.0% | never called |
The margin arm's worst equity ratio over the whole six years was 75% on the index fund and 71% on the volatile name, against a 25% maintenance requirement. On the index fund that is where the book started — the cushion never narrowed at all.
This was not a kind window. At 1.33× and a 25% maintenance requirement the hinge is a 67% fall, and the deepest fall in the record was 37.8% intraday. The construction excluded the event. The weather never got a vote. The converse is just as true and it is ours to say: never called at this ratio is a statement about arithmetic, not about the future.
Buy-and-hold, unlevered, over the same window: QQQ · the index fund 19.76% · MU · the volatile name 70.38%. Realized beta lands where 1.33× says it should — margin 1.21 vs LEAPS 1.18 (QQQ) · margin 1.15 vs LEAPS 1.10 (MU), so the two arms really are carrying the same exposure and the comparison is like-for-like.
“Your answer rests on a borrowing rate you made up.” It is a fair objection and it is why we registered a sensitivity before writing any code — and then did not run it, which we said on camera. It has now been run.
| drag @ FFR+0.5% | drag @ FFR+1.5% | drag @ FFR+2.5% | breakeven spread | verdict | |
|---|---|---|---|---|---|
| QQQ · the index fund | +1.51% | +1.37% | +1.23% | FFR + 9.02% | holds |
| MU · the volatile name | +1.46% | +1.43% | +1.41% | FFR + 26.67% | holds |
Move the borrowing rate a full point in either direction and the gap never closes on either name. The rate that would actually close it on the index fund is Fed funds + 9.02% — 7.5 points above what we charged ourselves. The verdict does not rest on the rate guess.
Two things run against us and we are stating them rather than glossing them.
Our strike rule is not their strike rule. We chose the largest listed strike at or below 70% of spot; the two educators whose lesson this prices choose by delta and aim for 70–80. Those are not the same contract. On the index fund, moving to an 80% strike adds a median of about 1.4 points a year in entry premium — so our figure is a floor on their version, not a price for it.
We rolled at six months; at least one creator teaches three. Different positions on the time-decay curve. Which is more expensive is something we did not measure and will not guess.
leaps_rolls in the summary file equals the entry count; the roll count here is derived, and the report and the episode both say entries and rolls separately for that reason. The implied financing column is the interest rate the option is quietly charging: the pre-registered band was Fed funds to Fed funds + 4%, checked on every entry, and rows outside it are marked. Check our arithmetic: the CSV below is the same file the figures above are computed from.| # | Sym | Entry | Expiry | Strike | Spot | Ask | Moneyness | Implied financing | Premium over intrinsic |
|---|---|---|---|---|---|---|---|---|---|
| 1 | MU | 2020-07-01 | 2022-01-21 | 32.5 | 49.70 | 22.95 | 65.4% | 12.50% | 11.57% |
| 2 | MU | 2021-07-26 | 2023-01-20 | 52.5 | 76.30 | 28.20 | 68.8% | 5.89% | 5.77% |
| 3 | MU | 2022-07-25 | 2024-01-19 | 40.0 | 60.15 | 25.70 | 66.5% | 10.05% | 9.23% |
| 4 | MU | 2023-07-24 | 2025-01-17 | 45.0 | 65.48 | 26.50 | 68.7% | 9.66% | 9.19% |
| 5 | MU | 2024-07-22 | 2026-01-16 | 80.0 | 115.28 | 47.55 | 69.4% | 11.20% | 10.64% |
| 6 | MU | 2025-07-21 | 2027-01-15 | 75.0 | 113.23 | 48.85 | 66.2% | 10.27% | 9.38% |
| 7 | QQQ | 2020-07-01 | 2021-12-17 | 175.0 | 250.49 | 86.00 | 69.9% | 4.24% | 4.20% |
| 8 | QQQ | 2021-06-21 | 2022-12-16 | 240.0 | 344.36 | 113.49 | 69.7% | 2.61% | 2.65% |
| 9 | QQQ | 2022-06-21 | 2023-12-15 | 195.0 | 281.08 | 104.46 | 69.4% | 6.67% | 6.54% |
| 10 | QQQ | 2023-06-20 | 2024-12-20 | 255.0 | 366.90 | 135.06 | 69.5% | 6.33% | 6.31% |
| 11 | QQQ | 2024-06-24 | 2025-12-19 | 329.8 | 473.96 | 173.87 | 69.6% | 6.35% | 6.26% |
| 12 | QQQ | 2025-06-23 | 2026-12-18 | 370.0 | 531.65 | 192.85 | 69.6% | 5.93% | 5.87% |
| 13 | QQQ | 2026-06-22 | 2027-12-17 | 515.0 | 735.73 | 275.94 | 70.0% | 7.63% | 7.50% |
A margin quote has two prices. The rate is on the screen. The second one — the fall that ends you — is in the maintenance schedule, your broker recomputes it nightly, and it moves with what you post. A threshold without its maintenance requirement is not a number.
And the thing worth carrying: correctly priced protection, bought against a risk your construction has already excluded, is still a cost. At 1.33× you are paying about a point and a half a year for a margin call that the arithmetic says cannot reach you.
Everything we make is free — the reports, the ledgers, all of it. The only thing that costs us real money is the market data.