trade-behavior-reviewlisted

# Trade Behavior Review Read-only. `reference/READ-ONLY-POLICY.md` (hard rule) and `reference/RESEARCH-STANDARDS.md` (data efficiency, delegated parsing, disclosure) apply. Kite is the only broker with working order/trade history (Groww's is dead; Upstox's is doc-verified but unconfirmed live) — check `reference/BROKER-CAPABILITIES.md`. With no such broker active, say so rather than substituting current holdings: a holding's average price is not a trade log. ## Steps 1. **Pull the history** via the order/trade-history capability, scoped to a lookback (default 6-12 months). Use the summary/history call, not a per-order loop. Large dumps go to a subagent per **Delegate bulk parsing** in `RESEARCH-STANDARDS.md`, with an output contract (one row per closed trade: symbol, buy date/price, sell date/price, qty, realized P&L) — judgement stays here. 2. **Reconstruct round trips** — match fills per symbol FIFO into closed trades with holding period and realized P&L. Still-open positions are excluded from win-rate math and noted separately. 3. **Metrics**: win rate; average win vs. average loss and the ratio; median *and* mean holding period (median matters more — a few long holds skew the mean); trade count and traded value as churn; best/worst names by realized P&L. 4. **Compare against stated intent.** Check median holding period against the horizon or style in `PORTFOLIO-PLAN.md` and name the gap in days/weeks. If the plan is silent, say s