Posted in My Daily Journal · 22 October 2026
A month-and-day swap in the 2015–19 option files is found and every seller result re-run, one engine prices the buyer and the mirror seller on identical fills, the buyer wins per rupee but with drawdowns ten to twenty-five times deeper, an independent reviewer re-derives the chain at scale, and the fourth quarter is confirmed as the weakest for buying.
What was the date bug?
In the older option files, 2015–19, month and day were swapped on ambiguous dates, so some trades had been priced on the wrong day. Corrected, the expiry-day seller comes to ₹7.07 lakh over the period, 9 of 12 years positive, worst drawdown ₹1.36 lakh — still a seller, just a smaller one. The daily 9:20 straddle on all days is +₹4 per day: +₹566 on expiry days and −₹145 on non-expiry days, positive in only 1 of 8 non-expiry years.
You built one engine for both roles. What does it say about buying versus selling the same option?
With both sides priced on the same fills, the ATM buyer makes ₹84 per trade at a 30% win rate, 5 of 11 years positive, worst drawdown ₹172.7k, five straight red years and 1,877 days without a new equity high; the ITM-150 buyer improves to ₹144 per trade. The seller earns ₹178 per trade with 46% winners. The parity check between the two sides is −₹8 per trade, so the engine is consistent, and a fill delay is worth about ₹900 per trade to both roles, which is why fill speed matters more than strike choice.
At fixed lots, with a fixed capital base and periodic top-ups, how do the books compare year to year?
Over 2025 and then 2026 the Williams buyer made ₹1.06 lakh then ₹6.15 lakh, the door ₹4.33 lakh then ₹0.04 lakh, BankNifty −₹0.20 lakh then +₹0.47 lakh, and the mirror buyer ₹8.55 lakh then ₹2.59 lakh — with worst drawdowns of ₹9.38 lakh and ₹20.8 lakh. Per rupee at risk the buyer earns two to four times the seller but with drawdowns ten to twenty-five times deeper, and no option-price exit rule survives two consecutive years. The chain itself needed only one top-up in the whole period.
Does the time of year matter?
The fourth quarter is the weakest for buying — the door earns about ₹4.4k per quarter in Q4 against ₹8–10k in the others — and the first quarter leads only because it contains the crash quarters. "Following the year", scaling up after good quarters and down after bad ones, was worse than fixed lots. Quarters are a description of the past, not a dial.
An independent reviewer re-derived the chain. What came back?
The chain alone at ten lots over twelve years comes to ₹63.98 lakh, positive in 10 of 12 years, 44% win rate, worst drawdown ₹4.64 lakh. The mirror buyer looks better until you notice that 141% of its profit comes from its ten best trades, against 31% for the chain. Three risks were listed and accepted as true: the edge has been thinner since 2023, the most recent months are modelled rather than traded, and execution at size is unproven.