Six months of the equity book
159 closed trades, 27 March – 29 September 2026 · realised only ·
show_performance --tenant t-default --months 6
| month | trades | wins | win % | realised |
|---|---|---|---|---|
| April 2026 | 25 | 7 | 28.0 | ₹2,726 |
| May 2026 | 61 | 13 | 21.3 | −₹1,08,593 |
| June 2026 | 20 | 7 | 35.0 | ₹83,683 |
| July 2026 | 19 | 6 | 31.6 | ₹1,22,456 |
| August 2026 | 11 | 5 | 45.5 | ₹9,353 |
| September 2026 | 23 | 13 | 56.5 | ₹3,06,757 |
| total | 159 | 51 | 32.1 | ₹4,16,382 |
Average win ₹20,212. Average loss −₹5,689. Gain‑to‑loss ratio 1.68. On a book of roughly ₹60 lakh that is about +6.9% realised over six months, before any open position is counted.
The shape matters more than the total
Two out of three trades lose money. That is not a flaw being admitted, it is the design: a win is 3.6× the size of a loss, so a 32% hit rate still compounds. A strategy with this profile that suddenly shows a 60% win rate is usually one that has started cutting winners early.
May was the worst month and the most useful one. Sixty‑one trades, a 21% hit rate, −₹1.09L. The loss analysis that followed produced the two hard rules the system still runs under — a cap of ₹2L per position, and no entry unless market breadth is above its own five‑session mean and the Nifty is above its 20‑day average. Trade count fell from 61 in May to 11 in August. That is the rules working: the gates refuse in conditions where the edge is not there, and refusing is most of what they do.
September's ₹3.07L is one month and should be read as one month. It is the largest single contribution in the window by a wide margin, on 23 trades. Six months of 159 trades is a small sample; one month of 23 is not a track record.
What these numbers are, exactly
- Realised only. Closed positions. Open profit is excluded entirely, which makes this conservative in a rising market and flattering in a falling one.
- Real money, not paper. The paper book is tracked separately and is never added in; the two are different fields for exactly this reason.
- Zero estimated exits across all 159 trades. Where a position vanishes without execution evidence the system books it at its recorded stop and flags it; on 2026-09-15 nine such exits produced nine wrong prices. None of the trades here are of that kind — every exit has an observed fill behind it.
- Our bookkeeping, not the broker's ledger. Reconstructed by
equity-servicefrom fills. That reconstruction has been wrong before and was corrected four separate ways in September: split‑adjusted bars, misdated fills, a hand‑entered phantom price, and SIP contamination. The verified Zerodha statement is the only figure that does not depend on it.
What this is not
It is not a backtest. The ₹21.94L and ₹26.96L figures that appear in the research notes are conditional historical replays of 127 supplied opportunities under a fixed ₹60L capital ledger. They are not realised trading, the production scanner has never been shown to generate that sequence, and they are not quoted here as performance. The table above is what the account actually did.
It is also six months. A strategy that survives one correction has survived one correction.