The Time I Almost Got Rich

My first options trade was ten Affirm $80 calls, September expiry, bought on August 26, 2021 for $2.05 a contract. The next day, Amazon announced Affirm would power its buy-now-pay-later checkout, and the stock opened the following Monday up 47%. I held those calls twenty-two days and sold them at $36.65 — a 1,667% return, $34,555 on a $2,050 ticket, on the very first pull of the lever.

That trade is the whole story, really. Everything after it was an attempt to have it again.

PART I — THE THESESAffirm, Ford, and why those tickers

Affirm wasn't a random meme. In the fall of 2021 the buy-now-pay-later thesis was as clean as retail theses get: BNPL checkout was spreading through e-commerce, Affirm had the premium brand, and the knock on the company — that it was dangerously concentrated in Peloton's exercise bikes — was dissolving in real time. The Amazon deal landed August 27. Partnerships with Target and Apple financing followed in October. Then on November 10 the company beat earnings and announced it would be Amazon's exclusive BNPL partner in the U.S., and the stock jumped ~19% overnight. The diversification-away-from-Peloton story wasn't hope — it was happening on a weekly cadence, in press releases. The stock went from the low $60s in August to an all-time high of $176.65 on November 4. I was long calls for almost every day of that run.

Ford was the value version of the same energy. The F-150 Lightning had over 150,000 reservations — three-quarters of them from people who had never owned a Ford. The Mach-E was selling. And the relative-valuation argument wrote itself: Tesla had just crossed a trillion dollars of market cap while Ford — a company shipping actual electric trucks to actual customers — traded around $60 billion. If the market re-rated legacy-EV even a fraction of the way toward Tesla's multiple, Ford didn't need to become Tesla; it needed to become 1.5× Ford. On October 27 the company beat Q3 estimates nearly 2-to-1, raised full-year guidance, and reinstated its dividend; the stock jumped 8.5% after hours and opened the next morning at a seven-year high.

I was positioned for that one before it happened: a thousand $16 calls bought October 18 at $0.33, sold into the earnings pop at $0.84 — +$48,723 — plus five hundred $15.5 calls the week before earnings, +$37,971. Ford ended up my best underlying: +$119,640 across eleven round trips. The irony is exact: Affirm, the thesis I loved most and traded most (88 round trips), netted −$84,321 — while containing my single best contract, +$122,850 on the $155 calls. Same conviction, same stock. The difference was never the thesis.

PART II — READING THE CHAINWhy that strike and not another

My chain-reading habit was consistent enough that the ledger reads like a fingerprint. 94.4% of every premium dollar went into calls. Median time to expiry at entry: three days; a quarter of all entries had one day or less. Some of that was style — weeklies near the money are where a dollar of premium buys the most gamma, the most convexity per contract — and some of it was budget. Far-dated options on a running stock cost multiples more per contract; when I was flush I occasionally bought a month out, and when I was chasing, I couldn't afford to. The expiry date of what I bought is practically a mood ring for the account.

The strike selection followed the stock up like rungs on a ladder. Watch it in the Affirm sequence: stock in the $130s in mid-October, I'm buying $150s for ~$4.80. Stock crosses $150, I'm in the $155s. Stock prints $176 on November 4 — and I'm buying $170s. Each rung kept the option in the same psychological place: a few percent out of the money, a few dollars a contract, a few days to work. Near the money, that's a gamma engine. Ten percent out with two days left, it's a lottery ticket wearing the same clothes. The chain doesn't warn you when you cross that line; the premium just quietly gets cheaper, and cheaper felt like more contracts.

PART III — THE CLOCKWhen I bought, when I sold

Every one of 769 fills carries a timestamp, recovered by matching its price against that day's quote tape. Put them on a clock and the account has a signature: buying concentrated late in the week, selling concentrated in the first half of the New York session — and a sharp, expensive asymmetry in which day and hour the money was made and lost.

FIG. 1 — The clock. Late-week buying, morning selling — and the hour of entry mattered more than either.

The buying leaned late-week — Friday (94 fills) and Thursday (91) were the heaviest days — and the selling was a morning habit, 55% of it done before noon Eastern. The asymmetry shows up in the exits: Thursday exits netted +$140,058 at a 49% win rate, the best of any day. Tuesday exits were the worst — −$70,768 at 36% — and Tuesday barely featured in the selling at all (46 fills, the second-lightest day). Whatever was being sold on Tuesdays, it was being sold badly.

The P&L-by-entry-hour chart is the one I'd frame. The single worst hour of my trading life was 10 a.m. Eastern: −$322,579 net across every trade entered in that hour — the hour after the open, the hour you buy yesterday's story at today's markup, the hour both big Affirm $170 entries printed. Two of those entries are outliers by any statistical test; Part IV strips them out properly — and 10 a.m. still loses six figures without them. The full accounting, dollars and percentages both, is below.

PART IV — THE CALENDARWeekends, Mondays, and the shrinking clock

FIG. 2 — The calendar. 56 weekend holds beat the other 320 trades by $443k.

The weekend numbers are the ones that rearranged how I think about that year. I held positions across a weekend 56 times. Only 24 of them — 43% — came back up. But those 56 trades netted +$175,536, while the 320 trades opened and closed within a single week netted −$267,664. Every trade this account is remembered for was a weekend hold: the first Affirm ticket (three weekends), the Ford earnings run (one weekend, entered ten days early), the Tesla deliveries gap (bought Friday morning, New Year's weekend, sold into a +8% Monday gap). And when I did carry across a weekend, I wasn't quick to fold Monday morning either — the average weekend hold ran about two and a half trading days into the new week, and a third of them rode all the way to Friday. The robustness check reframes rather than rescues it: five of the eight outliers were weekend holds. Strip them and crossers still net +$53,067 against −$189,968 for same-week trades — the direction holds — but the median weekend hold lost 27%, worse than the median same-week trade (−13%). Holding over weekends didn't improve the typical trade. It bought exposure to the tails — and the tails were where all the money was.

Put the two calendars together — carefully, because the ledger is tail-driven and raw dollar totals will lie to you. Eight trades out of 376 sit beyond three standard deviations, and they carry most of every total:

FIG. 3 — The percent view. Median trade return by when it was entered: every bucket is negative; Thursday (−2%) and the open (−7%) were least bad, Tuesday (−26%) worst.

With the outliers removed, the story sharpens rather than dissolves. Monday's heroic +$179,542 collapses to −$7,715 — its greatness was three big winners, not a good habit. Tuesday's crater shrinks from −$248,848 to −$33,927, but Tuesday keeps the worst median return (−26%) and the worst win rate (31%) of any day: the two Affirm disasters didn't make Tuesday bad — they happened to a day that was already the worst, which is precisely the point. And Thursday emerges as the only honestly good day: zero outliers, +$39,575 that survives trimming untouched, the best win rate (48%) and the best median (−2%). The same test on entry hours: 10 a.m. keeps its crown even after losing its two famous disasters — still −$107,658 across the remaining 71 trades, with the worst morning median (−20%) — while the midday hours' positive totals turn out to be three outlier winners wearing a halo; no entry hour had a positive median return. The difference between hours was never edge. It was how expensive the habit was.

The holding-period gradient survives the same scrutiny, with one honest asterisk. Same-day scalps: −$102,634 over 204 trades, zero outliers, a tight −9% median — a consistent, grinding bleed. Overnight-only holds were robustly the worst thing I did: −$144,898 on 65 trades, zero outliers, a −59% median and a 26% win rate — the buy-the-close, panic-the-open cycle, fully earned. Past two days everything turns net positive even after trimming (+$44k for 2–4 days, +$23k for 5–10). But the >10-day bucket deserves the asterisk: +$109,612 raw, +$42,962 trimmed — and a median of −63%. Long holds weren't "patience pays." They were lottery exposure with a 436-point standard deviation: most died, and the survivors paid for everything. My edge lived at swing-trade speed; my habit, by December, was scalping — a hundred-plus trades a month, each one paying the spread and the commission for the privilege of anxiety.

PART V — THE PROFIT CLOCKWas there ever a moment to get out ahead?

One more way to grade every exit, in the only currency that matters: for each lot, measure the share of minutes the option's bid — the price actually on offer — sat above what I paid, from purchase to sale. Then extend the window through expiry, as if I'd never sold, and measure again. The first number says how often I was looking at a profitable exit while holding; the comparison says whether the profitable stretch happened on my watch or after I left.

FIG. 4 — Above the diagonal: the option kept trading above my cost after I sold (the good period outlived the exit). Below: the profitable share collapsed once the full life is counted — I owned the good period and stayed through the bad one. Bottom-left corner: 24 lots that never traded above cost for a single minute.

The diagonal splits the book into two behaviors. Above it — lots that kept trading above my cost after I sold — sit 112 lots worth +$402,854: the winners, sold into strength, with more good minutes still to come. "Too early," and profitable, every time. Below it sit 173 lots worth −$185,756: I owned the profitable stretch, didn't take it, and stayed through the decay — by the time these were sold, counting the full life just dilutes what little good period there was. Selling early made money; outstaying the window lost it.

But the number that changes the diagnosis: only 7% of sold lots never traded above cost at any point through expiry — and of the 58 lots that rode all the way to expiration, 46 passed through profit on the way to zero. The median losing lot had the bid above entry for 10% of its held minutes; 109 losing lots, −$354,670 between them, each spent more than 5% of their full life with a profitable exit sitting on the screen. The problem was almost never that a winning exit didn't exist. It's that the window opened, stayed open for hours, and closed — with the position still on.

PART VI — THE CHEAT SHEETBest and worst, by the numbers

Read the table cynically and it says one thing five ways: the account made money when a position was allowed to express a thesis, and lost money when it expressed a reaction. Thursday entries held across days were thesis. Tuesday-10 a.m. entries closed by Wednesday were reaction — a habit that loses money with the disasters removed, and lost catastrophically with them in. The tickers were fine — Affirm did sign Amazon, Ford did re-rate, Tesla did beat deliveries. The theses paid retail investors who simply held the stock. What they couldn't survive was being expressed through three-day options bought at the top of the morning, one strike higher each week.

The account touched $619,502 U.S. — about $771,000 Canadian on my broker's home screen — at 9:42 a.m. on November 4, 2021, the exact minute Affirm printed its all-time high. It finished at $183.12. Between those two numbers sits everything in this article: a right thesis, a real edge with a documented day and hour, and a clock that sped up exactly when it should have slowed down.

None of this was visible from inside the account in 2021 — at that speed, nothing is. It took parsing 769 confirmations, matching every fill to the quote tape, and rebuilding the account minute by minute to see that the wins and the losses weren't two runs of luck. They were two different behaviors, each with its own day, its own hour, and its own holding period — running side by side in the same account, under the same thesis, about half a million dollars apart.