A podcast trader's setups under test, the standing answer, a charter, and why the stop is the lot count

His demo trade reproduced exactly on our data (+746.75 points, ₹46,252 on one lot). Over December 2019 to October 2026 the rule fired 44 times and filled 33: ₹6,830 a trade, about ₹2.25 lakh in total, win rate 39%, profit factor 1.78, drawdown −₹1.08 lakh, confidence interval including zero — and the top five trades made ₹3.96 lakh, so the other 28 averaged −₹6,076. The 30-minute version he says "auto-adjusts to any timeframe" lost ₹168 a trade over 537 trades. The delta-one claim is half right: on identical nights the deep-ITM monthly saved ₹360–590 a night of friction, but it is a future plus a protective put, and the put cost about ₹257 a night, leaving +₹106 ± 405 since late 2024 — unprovable — while clearly winning on risk (worst night −₹15k instead of −₹26k, ₹41k of premium instead of ₹1.9 lakh of margin). Verdict: a real but five-signals-a-year lottery, not a bot; the instrument, though, is the one our pivot buyer had just adopted.

Oversold-breakout on 159 liquid stocks since 2018: as stated, 433 trades at +0.09R with profit factor 1.21 and −0.27R since November 2024; drop the "structurally bullish" filter and it improves to +0.31R; buy on a random day with the same tight-stop, far-target exits and you get +0.25R. The trigger adds about 0.06R — the exit geometry made the money, and every dip-buying cell has been at or below zero since late 2024. Candles: eighteen pre-registered cells on 3-, 5- and 15-minute bars (Supertrend, EMA crosses, first-candle and inside-bar breaks, VWAP reclaims, impulses, gap fills, lunch breakouts, option-volume momentum), 23,713 trades on real option prices. Only the control — the gap-and-go at ₹713 a trade — works; the best non-control cell was floor-pivot breaks at ₹194 with a modern figure of ₹70 ± 443. A round trip on a ₹10k premium costs ₹300–400, and the typical three-to-five-minute move is smaller than that.

The shape is real: under hold-to-expiry every entry we tried shows 71–87% losers and 4–10% of trades above +200% of premium, and even the trailed gap-and-go has its top 5% (13 trades of 259) carrying 104% of the profit. The shape is not the edge: the identical skew appears for Supertrend, prior-day-high and every-day entries whose means sit at or below zero since late 2024. Letting gap trades run to expiry beat the trail in 2020–23 with five times the drawdown and has lost ₹1,044 a trade since November 2024; the famous carriers were Budget day 2021, March 2020, June 2022 and April 2025. An out-of-the-money variant looked positive in every window until it was re-run with the live bot's exact opening-range definition, where it fell to ₹210 a trade and −₹440 in the modern window — which earned a standing process rule: nothing is called a lead until it survives the live entry definition on both data sources.

The pivot engine's signals bought as a deep in-the-money current-month NIFTY option, index-trail exits, overnight holds, roll at three days to expiry: 1,526 trades on real prices 2020 to September 2026, ₹671 a trade, seven of seven years positive, interval excluding zero, profit-to-drawdown 4.4, charges ₹126 a trade. Three risks stated with it: it is in its deepest drawdown as we speak (−₹2.30 lakh, April to September 2026); it needs a deep-ITM half-spread of 0.35% or less (₹359 a trade and two red years at 0.5%); its worst trade is an overnight gap worth 94% of the premium. The change criteria were written down too — a ten-session median half-spread above 0.35%, thirty paper fills more than 0.5% worse than fair value, or the modern window turning negative with 500 or more trades — and nothing else moves the answer. His pyramid idea was tested the same night on all 1,526 signals: both variants beat one lot per signal, neither beat a plain two-lot book on profit-to-drawdown (4.1 and 3.4 against 4.4), so the second lot is better deployed at entry than after the move.

Written down as a charter: option buying only, with the loss defined by the premium and the profit uncapped — "unlimited profit, limited loss" is the soul of the system. It changed how research is judged (convexity first: profit factor, win-to-loss size, tail share, worst trade against premium — not win rate), what gets researched (long-premium, defined-risk structures; the two sellers on the fleet stay secondary and are flagged as the one payoff shape the charter rejects), and how size is set (capital per lot equals the premium plus the measured worst drawdown, stepped up as the book grows, fixed lots per trade at each step). The truth rules — pre-registration, the two-window test, both data sources, the live entry definition, thirty-trade paper gates — were left exactly as they were: ambition sets the target, not the evidence bar.

Yes, and there are two. On the same pivot signals and the same trail exit, an instrument ladder showed the pattern cleanly: deeper pays more per signal — ₹678 for ITM-500 monthly, ₹634 for ITM-300, ₹514 for ATM monthly, ₹475 for ITM-150 weekly, ₹139 for OTM-100 weekly and ₹1 for OTM-200 weekly; the lottery ticket earns nothing. On BankNifty the same engine bought ITM-1000 monthlies for ₹724 a signal, seven of seven years, profit-to-drawdown 5.4, and its strong years (2022, 2024) are NIFTY's weak ones. Two filters passed in both halves as size-up candidates, not rules: signals on a ≥0.5% gap day (₹1,098 versus ₹515) and boxes of 0.6% or wider. The BankNifty twin was approved as a paper book; its open risk is the spread, since at 0.5% per side it loses 2026.

The logic held — the replay reproduced the live bot's October signals eight for eight within a minute — but the double-check found what the headline hid. Both buyers are in the same drawdown at the same time: NIFTY from the 9 April 2026 peak to the 28 September trough −₹2.30 lakh, BankNifty −₹2.31 lakh over the same dates, so one lot of each was −₹4.6 lakh together and yearly diversification does not help inside a drawdown. All the profit is overnight: 63% of signals hold overnight and earn ₹2,927 a signal, while same-day exits lose ₹3,099 a signal — the engine pays whipsaws to be in the runners. At the measured 0.25% spread the NIFTY book is ₹559 a signal with 2026 at −₹19k; 2026 win rates were 33% and 29% with losing runs of fifteen; the worst days were election-count day in June 2024 (−₹35k) and gap mornings in April 2020, June 2022 and April 2026. The sizing changed, not the answer: ₹2.5–3 lakh per lot per index, with the two books sized as one bet.

There is no option-price stop: the stop is on the index — the last ten-minute fractal pivot low for a call, high for a put — ratcheting one way only, with no target; the median planned risk is 95 points, about ₹6.4k per lot. Realised losses run at a median 0.58 times the plan, 81% stay within it, 5% exceed 1.5 times, and every one of the ten worst but one is an overnight gap through the stop. Eight pre-registered exit variants — premium stops at −10/−15/−20%, a half-box initial stop, a break-even lock, flatten at 15:15 if losing, and no re-entry until a fresh pivot — and zero passed: premium stops cut the total 4–20% with no drawdown benefit (a premium stop cannot stop a gap), the tighter stop traded 22% more for the same money, the break-even lock was a wash. The one variant that changed the risk shape — never hold a loser overnight — removed the gap losses entirely (−₹6.0 lakh → +₹0.9 lakh) and cut the worst trade 49% and the drawdown 20%, at a cost of about 3% of total and −₹40k in the modern window, inside noise; not adopted, re-examined once at the December review with a shadow ledger logging it on live quotes meanwhile. The honest answer to "minimise losses" is the lot count: the gap that produces the tail is the same mechanism that produces all of the profit.