Posted in My Daily Journal · 19 June 2026
The first real results of the NIFTY pivot-breakout system in paper mode, including the day a gap jumped the trailing stop and why that is the strategy, not a bug.
What did the first three paper trades of the pivot system look like?
Two losses and one winner that paid for both: a long and a short whipsawed on the first day for −₹8,423 and −₹5,170, then a long carried over the weekend caught a multi-day trend and closed for +₹24,461 — net +₹10,868 after three trades, one win in three, payoff about 3.6 to 1. That is the exact shape the backtest promised: a win rate near a third and a payoff above two, with the money arriving in a few trends rather than many trades.
A few days later the position lost ₹4,155 by gapping through its stop at the open. Was something wrong?
No — that is the system working as designed. The carried long had a trailing stop at 24,075; NIFTY opened at 23,950 and the exit filled there, 57 points below the stop. Holding overnight means gap risk is symmetric: a gap up extends the winner, a gap down jumps the stop. The lesson we wrote down for every future report: a trailing stop is where the system tries to exit, not a guarantee, so never quote an open profit or a trail level as "locked".
Three straight small losses followed. Does a breakout system have a failure regime?
Yes — a range. Within a week the record went from a +₹10,868 peak to +₹3,159 (six trades, one winner) as NIFTY sat in a 300-point band. Breakout systems bleed in ranges and earn in trends; the drawdown is the price of being in position when the range breaks. By the end of the month the carried short survived a three-day gap, the trail ratcheted below the entry, and the record stood at a new high of +₹28,204 after ten trades with four winners.
You replayed the bot's own logic on historical data to check it. Did the replay match live paper results?
The replay overstated them. The idealised replay of the first big trend trade showed +₹31,936; the live paper book booked +₹24,461. About ₹7,500 of the gap was real downtime (a late manual start one morning), the rest perfect-fill optimism. We treat any replay P&L as a ceiling, not an expectation.
What operational lessons came out of the first month?
Three. Automate the start and the holiday calendar — a missed 09:05 start cost a morning, and the expiry rollover (carrying an open position from the expiring future to the next) needed its own code and its first live test. Second, a bot that restarts itself can survive being "disabled"; retiring a system means masking it, not switching it off. Third, measure every bot against its own backtest from day one, because drift between the two is invisible until you record it.