Memory is not a database; it is a spotlight, and the spotlight sits on whatever happened last. When the brain estimates “how likely is this trade to work,” it does not consult your 300-trade backtest — it consults the vivid, emotionally-charged sample of the last three. That sample is statistically meaningless and psychologically overwhelming.
- Streak sizing. Three winners and size creeps up “because you’re in form”; three losers and you skip the next valid setup — which is, statistically, just as likely to work as any other.
- System-hopping. A tested strategy has a normal losing week and gets abandoned for whatever worked most recently on YouTube — the graveyard of Indian retail accounts is paved with strategies abandoned mid-drawdown.
- Regime assumption. A trending Nifty week convinces you every dip is a buy; the range week that follows collects the bill.
- Expiry memory. One remembered expiry-day windfall outweighs the forgotten dozen that decayed to zero — recency plus selective memory is how lottery-ticket option buying sustains itself.
Fix size by rule, not by form. Position size changes only on a written schedule (e.g. after a month of clean execution) — never after a streak in either direction.
Judge systems on samples, not weeks. Pre-commit the evaluation window (e.g. 30 trades) before trading a system; inside the window, results are data, not verdicts.
Keep the long-run numbers visible. A journal dashboard showing your 100-trade win rate and expectancy gives the spotlight something bigger to look at than yesterday.
Name the regime daily. Writing “trend / range / event day” in the pre-market plan forces the question recency skips: is today actually like yesterday?