Verdict
Passes the re-costed rule; forward paper run scheduled from 2026-10-26
Same finalist, cheaper standard costs: Nifty +₹1,476 per night on the sealed period (t 2.6), +₹1,681 overall; Sensex +₹693 (t 1.8) but only +₹344 in its second half.
- Rules fixed before the re-run (STRATEGY_OVN.md section 12). Five conditions, all passed: Nifty positive in development, validation and sealed; Sensex positive in both halves; Nifty positive at 1.0% slippage per fill; Nifty positive with model-priced exits 10% worse; one-lot tail rule.
- The re-computation reproduces v1.0 exactly (sealed mean at 2x costs +₹870.70), so only the cost inputs changed.
- Why it now passes: v1.0 required the result to survive 2x and 3x costs (3% to 4.5% slippage per fill). The standard model is 0.5% per fill. Even v1.0's own 1x base (1.5% per fill) gave +₹1,469 a night on Nifty; v1.0's closure came from the stricter stress and the Sensex replication.
- Every period has been seen: Nifty's sealed period was opened once in v1.0 and Sensex ran on its whole sample. This version is therefore a sensitivity of a known result. The only clean test is a forward paper run.
- Fragile points: about 60% of exits use a modelled price (Black-76 fill); with those 10% worse for the seller Sensex is about zero (₹-64) and at 20% worse Nifty falls to +₹892 (t 2.0); Sensex's second half is weak (t 0.6).
- The only slippage figure from real fills in our records is the 1.5% per fill the Tradetron report measured; 0.5% per fill is the owner's assumption and is unmeasured for selling the ATM strangle at the 15:15 open. The result at 1.5% is shown in the table.
Promotion gates
No gates published for this strategy.
Explain the strategy
Same trade as before: sell a pair of Nifty options near the close and buy them back next day. With cheaper trading costs than we tested before, the average night makes money, in every period we have data for. But all that data was already used, so only a live pretend-trading run can prove it.
Report
The finalist re-costed, by period and cost level (₹ per night per lot pair)
| Underlying | Period | Nights | Mean per night (new base: 0.5% per fill) | t-stat | Win rate | Worst night | At 1.0% per fill | At 1.5% per fill (v1.0 1x) | New base, model exits 10% worse | New base, model exits 20% worse |
|---|---|---|---|---|---|---|---|---|---|---|
| Nifty | development | 99 | ₹1,700 | 4.11 | 70% | ₹-9,726 | ₹1,599 | ₹1,498 | ₹1,428 | ₹1,156 |
| Nifty | validation | 47 | ₹1,860 | 1.7 | 81% | ₹-39,606 | ₹1,735 | ₹1,610 | ₹1,225 | ₹589 |
| Nifty | sealed | 50 | ₹1,476 | 2.65 | 76% | ₹-10,995 | ₹1,378 | ₹1,281 | ₹1,064 | ₹653 |
| Nifty | all | 196 | ₹1,681 | 4.65 | 74% | ₹-39,606 | ₹1,575 | ₹1,469 | ₹1,287 | ₹892 |
| Sensex | first half | 106 | ₹1,048 | 1.99 | 70% | ₹-23,939 | ₹935 | ₹822 | ₹320 | ₹-409 |
| Sensex | second half | 108 | ₹344 | 0.64 | 65% | ₹-25,904 | ₹233 | ₹121 | ₹-440 | ₹-1,224 |
| Sensex | all | 214 | ₹693 | 1.84 | 67% | ₹-25,904 | ₹581 | ₹468 | ₹-64 | ₹-820 |
Rules and parameters
Rules
- Underlying and contract: Nifty 50 weekly options, the contract that expires soonest. Sessions to expiry is counted in trading days; entries on expiry day are not allowed (the contract would expire before the exit).
- Entry (finalist): on a session 1 or 2 sessions before expiry, sell one ATM call (strike nearest the spot at the close of the 14:xx bar) and one put 100 points below it, at the open of the 15:15 bar.
- Exit: buy both back at the open of the 12:00 bar of the next session. No stop and no target in the finalist (stops, targets, wings and other exits were searched and did not help).
- Prices: observed 1-minute opens from Dhan's rolling series where the strike is inside the -2..+2 window; Black-76 (forward from put-call parity at the ATM pair, ATM implied volatility, skew fitted on observed strikes) otherwise. Model error on observed strikes: median about 0% to 2%, mean absolute error 10% to 12%.
- Costs: slippage 1.5% of the premium on every fill (3.5% in a session's first 15 minutes, the Tradetron report's measurement on 21,011 live fills); brokerage ₹20 an order; exchange 0.03503%; SEBI 0.0001%; GST 18% on those; stamp 0.003% on buys; STT on the premium sold 0.10% to 2026-03-31 and 0.15% from 2026-04-01; all times 1x, 2x or 3x.
- Periods fixed before any result: development 2024-12-03 to 2025-11-07, validation 2025-11-10 to 2026-04-29, sealed 2026-04-30 to 2026-10-16. A diagnostic on a baseline configuration printed sealed figures before the finalist was chosen; the finalist's sealed result was not looked at.
- Sizing for the tail rule: one lot pair (65 units from 2025-12-30), margin about ₹1.5 lakh per strangle (assumption), capital ₹15 lakh.
Parameters
- finalist
- sessions to expiry 1-2; sell ATM call + put 100 points lower at 15:15; buy back next 12:00 (unchanged)
- charges
- standard Dhan: Rs 20 an order, exchange 0.03503%, SEBI, GST, stamp 0.003% buy, STT 0.10% of premium sold to 31 Mar 2026 then 0.15%
- slippage
- 0.5% per fill = 1.0% per full cycle (owner); stress 1.0% and 1.5% per fill
- pre-registration
- research/ovn/STRATEGY_OVN.md section 12
- next
- forward paper run: nightly 18:40 IST on weekdays from 2026-10-26, spec frozen 2026-10-24 (paper_spec.json v1.1); no orders, Dhan /charts data only
Instrument
Nifty weekly options, one lot each side (lot 75, then 65); Sensex used as a replication · capital ₹1,500,000
Known limitations
- Re-costing known data, not independent evidence; no untouched period remains.
- About 60% of exits are priced by a model because the strike left Dhan's -2..+2 window (about 10% to 12% average error on observed strikes).
- The slippage assumption (0.5% per fill) is below the 1.5% measured in the Tradetron report.
- Sensex replication is weak in its second half.
- Tail: worst night about -₹41,000 per lot pair (2.7% of ₹15 lakh); two lots would break the 5% rule.
- The paper run reads real 1-minute candles of the listed contracts through Dhan's charts endpoints (no orders, no broker calls). Its first live plumbing test on 2026-10-22/23 candles worked end to end; that forced test trade lost Rs 468 and is not part of the record.
Logs and raw data
Downloads appear once a data window is loaded.
Files are plain CSV so results can be re-computed independently. Trade dates are IST sessions; returns are net of modelled costs and slippage.